Introducing literate modeling

Organizational modeling has much in common with programming. If you create or ask for organizational models in any one of a number of text-based systems, you likely realize that. If you use any one of a number of GUI-based systems, you may recognize what you're doing as visual programming.

In organizational modeling, the challenges are many:

• How to create a competent model
  
• How to make the process transparent so others can collaborate and still others can benefit
  
• How to organize the model in ways that are understandable both to people and the computer
  

It's both easier and harder than it sounds: it's easier to produce useful, insightful models than many might think, and it's harder to create solid models than some would have you believe.

Despite the claims of some, I don't think there's one right way: I think we need a diversity of approaches. To that end, I've been exploring text-based modeling and simulation for the last few years. Recently I've begun to merge the ideas of literate programming and organizational modeling into what you might call literate modeling. Strictly speaking, that's only doable with text-based modeling languages. It brings with it a few features that seem important:

• Thinking about the model and creating the model go hand-in-hand. I've already found that text-based modeling has helped me think in new and deeper ways about the problems I've addressed, and literate modeling strengthens that thinking by linking them even more tightly.
  
  Why is that so? Part of the reason has little to do with literate modeling and textual programming per se: I found that splitting up the work of modeling and simulation into different types of tasks (conceptualizing and modeling a problem, designing experiments to test theories of action, analyzing and thinking about experimental results, and communicating those results to others) helped me do a better job at each stage.
  
  Literate modeling adds a writing component, which may help us think more carefully about our modeling decisions, to the modeling process. Forcing the explanation of the model to go hand-in-hand with its creation seems to keep me from making assumptions I can't support. In fact, I tend to let the writing drive the modeling, which I think drives models even more from the aspect of hypothesized causality than from the aspect of what works technically. You can see what others have said about literate programming; I think those ideas apply to literate modeling, too.
  
  Of course, you can document normal models extensively, too. As others have said, literate approaches link the thinking required to explain a model (or a program) to other people tightly to the model (or program) itself; the model flows from the thinking and writing instead of the writing becoming partially a reverse engineering of the model code.
  



• Model consumers, collaborators, and even developers think most naturally about models in a sequence that doesn't necessarily match the sequence needed by the simulator. Literate modeling allows you to decouple those two sequences completely.
  



• To explain a model, you sometimes need graphs, diagrams, pictures, tables, or other non-textual components. A good literate modeling system can integrate all those components easily without being tied down to the features offered by a particular simulator. In a way, this is the old Unix philosophy: use a collection of smaller tools, each well-suited to the task at hand, and put them together flexibly as you need them.
  



• Literate modeling is one approach to applying some of the best of the lessons of software engineering to a related field in order to do better work.
  

Why do you care? Why should you care?

If you work with or use organizational simulation models to make better business or organizational decisions, perhaps literate modeling offers you a way to have more transparent, more collaborative, more understandable, and more well-thought-out models. Better models, applied appropriately to generate better insights, might just help you make better decisions. Better decisions, carried through with good implementation, might just lead to better results over the long haul.

Is literate modeling or, for that matter, text-based modeling the only way to go? By no means! By the rule of diversity, if nothing else, that would be wrong. Each of the graphical modeling tools I use has brought its unique strengths to the problem-solving process, and I continue to use them. "Horses for courses," as they say. Yet don't count out text-based tools for effectively engaging those with whom you're working, don't think that text-based modeling and literate modeling has to be ponderous, and don't count out the power of text and writing to help you think more effectively about the challenges you face.

What do you think?

Dealing with risk and uncertainty

Times are uncertain. Risks seem high. We may feel that the price of missteps is high; we know it's hard to decide what steps to take.

In situations such as this, how do you make decisions in and for your organization? How do you plan effective actions? How do you solve the inevitable problems that arise?

The German psychologist Dietrich Dörner, author of The Logic of Failure, has made a career of studying why people make mistakes and what we can do to improve. One of his key pieces of advice is to use computer simulation to get insight about the situations we face so that we can make better decisions in real life.

Perhaps today's uncertainties are your signal that the time is right to apply more systemic approaches in your work and to ground your planning, problem solving, and decision making with simulation that takes into account factors important to your business. Perhaps it's time to test and rehearse your plans before you implement them.

That's what we've been discussing here, and that's how I help others. If you're concerned that your standard approach to business may need augmentation in today's world, perhaps I can help you, too. Drop me an email or give me a call. There's no obligation—only opportunity.

System dynamics in Seattle

I just ordered texts for IMT 586, Information Dynamics I, in the Information School of the University of Washington, which reminds me to tell any of you at UW or within commuting distance who have been interested in system dynamics that we do plan on teaching system dynamics again this winter quarter. Registration starts November 7.

If there's enough interest, we hope to teach the follow-on IMT 587 in spring 2009. If you take IMT 586 this winter, consider leaving time in your spring schedule. If you took IMT 586 last year and would like to go further, think of IMT 587. If you've already got a system dynamics background (equivalent roughly to the first fourteen chapters of John Sterman's Business Dynamics) and would like to go straight to IMT 587, let's talk sometime before spring quarter enrollment.

I look forward to seeing you there!

Prediction, system dynamics, and Future-Fusion

Recently, I made the claim that we're better off focusing on adapting to the present than predicting the future. I've made similar claims in the past, too. I've even given one example in which predictions serve a useful purpose.

That's all a bit simplistic, of course. Even system dynamicists could be said to predict the future in a way: we show behavior over time we feel is more likely to occur (although we may warn people away from point predictions based on a behavior over time graph). In other words, I might suggest that your current policies could produce a boom and bust effect in your business, but I wouldn't want you to draw the conclusion that your business will grow another 172.3% by June 15, 2009 before taking a tumble that afternoon.

Because we all would like to know the future, I've experimented with blending system dynamics and Bayesian analysis to quantify the probability of a particular behavior pattern, for example. Of course, that probability is conditioned on both the historical data and the model being correct, which is a loophole big enough for a good-sized locomotive to run through: models are always incorrect. Still, I think this approach may give more useful insight in certain cases.

Now Kshanti Greene of Stottler Henke Assocates, Inc. has shown me a Bayesian tool they've developed called Future-Fusion, and I've been exploring it a bit. They are using Bayesian networks and the power of groups to get a better handle on what the future holds. Much as Data360 looks at the past, Future-Fusion attempts to look at the future. As of this writing, they've created four test areas which you can explore: the 2008 US presidential election, the Iraq war, corporate strategy, and energy. Try it out: learn how to use the system, see current predictions, and add your own (I think you only have to create a free account if you want to add your own predictions). Perhaps you'll learn something, and perhaps they will, too.

Kshanti has pointed out a recent addition to Future-Fusion that may intrigue some of you: time. They've enhanced their technology to allow limited dynamic execution of a network model, which begins to narrow the gap between Bayesian networks and system dynamics from the Bayesian network side, much as what I've tried has narrowed it from the system dynamics side. To try that out, go to the energy model, select a prediction (e.g., "Reduced SUV sales"), click "view graph," note the numbers, and then click "Next Time Step."

I think this is all still experimental in many ways, but it's a good opportunity to learn a bit about this technology by trying it out on real-life issues. I'll be curious what you discover.

Bittersweet asphalt

No, it's not that I find asphalt bittersweet (I haven't tried tasting it). It's the news that I find bittersweet.

Five and a half years ago, I published Out of Gas: A Systems Perspective on Potential Petroleum-Fuel Depletion. In that column and in the accompanying simulation model, I suggested that delays due to debates over how to allocate shrinking petroleum stocks might hurt our ability to replace energy resources in a timely fashion.

Today, I read Asphalt shortage disrupts road projects. I'm sure you can find other examples, perhaps closer to your home, in which the imbalance in supply and demand of petroleum is leading companies to prioritize one usage over another, which can cause pain for the unfavored group.

Models such as this aren't designed to predict the future, at least in the sense that they tell you that a certain event will happen in a certain year. They're intended to give insights into the likely and potential ramifications of current and proposed policies, both formal and informal, that we've created. They're intended to help us test policies quickly, inexpensively, and at low risk, so that we can be more confident when we implement a policy in our organizations. They do not provide guarantees, but they can provide very useful and sometimes unexpected insights.

In which areas would you like to think more effectively about the effect your current policies could have on your organization's future?

Alternative energy: not as easy as it sounds?

Five and a half years ago, Pegasus Communications published my Out of Gas: A Systems Perspective on Potential Petroleum-Fuel Depletion. If you try out the downloadable model (see the column for instructions), you will discover my concern that we wouldn't start soon enough or be able to move fast enough to replace petroleum. Such delays could have a significant, perhaps massive, impact on society and on our economies.

Yesterday Forbes published America's Best Places For Alternative Energy and noted SRI's estimates that we need "[r]oughly 4.2 billion solar rooftops, 3 million wind turbines, 2,500 nuclear power plants or 200 Three Gorges Dams" to replace the amount of oil we use annually and that "no single category of renewable energy is growing anywhere near the speed it needs to bear the full brunt of displacing carbon-emitting fossil fuels anytime soon."

So my simple concept model identified a problem that's substantiated by more research at SRI (and by our daily experience, for if alternative energy sources were fully replacing petroleum, would we see overall energy price increases?).

That's part of the message of Is predicting the future really worthwhile?. My simple model didn't predict the future.

• It did identify past patterns of action that could credibly lead to a problem.
• It did use information that's known reasonably well (quantities of petroleum, even admitting that we don't know reserves as well as we'd like, as well as something about the dynamics of petroleum discovery and use) and apply simulation to explore what ramifications those factors might have over time.
  
• It did allow one to try different scenarios to see whether one's conclusions were sensitive to assumptions. For example, does it make a fundamental difference if petroleum reserves are 25% higher than assumed? (No, it just changes the timing of the problem.)
  
• It did provide a test bed for exploring strategies to see which might be more effective.
  

More research (in this case, in the form of the SRI study) substantiates the nature of the problem and helps us understand its magnitude and timing.

What systems are at work in your organization, your business, or your part of the world that might lead to consequences you don't want? How might you test your ideas? What can you change that might lead to better results? How might you test those ideas?

Is predicting the future really worthwhile?

Predicting the future, also called forecasting, is a popular business activity. Some managers want to know what the future holds so they can plan to accommodate it.

Yet I'm reminded by The Oil Drum: Europe's The Fantasy World of the UK Government that our record in forecasting isn't so great.

In that report, the U.K. government published a prediction on May 7, 2008 that gave four oil price scenarios. In the highest of the four, the "high high scenario," the price of oil hits $107 per barrel in 2010 and stabilizes at $150 per barrel by 2015.

As of my writing this, upstreamonline.com shows crude oil spot prices ranging from $138.96 to $152.58.

In 58 days, we've hit the prices they forecast for 7 years in the future.

Instead of predicting the future and then designing a business system to work well if the prediction is true, wouldn't it be better to design a business system that responds appropriately to whatever the future brings?

Isn't that hard? Yes, but so, apparently, is prediction.

That's one of the goals of system dynamics: to give us models which we can test against multiple futures to see if our modeled business systems work as well as we'd like independently of the outside environment. Once we are satisfied with the insights we've gained from modeling, we can implement our real business system with higher confidence it will work as we expect no matter what the future throws at it.

If you'd like to talk about such adaptive business systems, give me a call.

Sneak peak: Information Dynamics I / II

If you are a current University of Washington graduate student or you live within commuting distance of the University of Washington and if you are interested in learning about system dynamics in an academic setting, put IMT 586 and IMT 587 on your calendar for the coming winter and spring quarters.

If you took IMT 586 last year or if you have a solid background in the material of the first half of John Sterman's Business Dynamics, put IMT 587 on your calendar for the coming spring quarter (yes, that's nine months away). We plan to offer it, assuming we have sufficient enrollment.

I'll make a fuller announcement as we get closer. Ask if you have questions, and let me know if you think you're interested: I'm curious and interested.

President Bush and greenhouse gases

I haven't been a political blogger, and I'm not about to start now. Yet the news of the past few days does offer ways to illustrate systems concepts I've mentioned before, and so I thought I'd point out what I hope is obvious to all here.

For but one example, take US President Bush's goal of having greenhouse gas (GHG) emissions stop growing by 2025, which is stirring up comment world-wide.

In system dynamics terms, GHG emissions (largely CO₂) are a flow, and the amount of CO₂ in the atmosphere is a stock. If you recall what I've written before on stocks and flows, you'll see that stopping the increase of a flow does not mean that the stock will decrease; it simply means that it will increase less rapidly.

In other words, even if we do meet this goal, things may well continue to get worse well after 2025, but they will at least get worse less rapidly after then.

I want to show you a little model that demonstrates that behavior, but, to publish it here, I'd like to get the numbers at least close to right, and that would take a bit of research time I don't have tonight. Let me try an analogy, instead; those of you who studied and remember the calculus can probably make a more elegant argument, and those who do system dynamics models can create one on your own in a few minutes (if you have the needed parameters, let me know, or post a pointer to your model).

In the real world, we are emitting CO₂ into the atmosphere by breathing, burning fossil fuels, and the like. That stock of CO₂ in the atmosphere is growing and threatening climate havoc.

Some of that CO₂ is taken out of the atmosphere each year through the action of photosynthesis and perhaps other mechanisms.

According to the science I read, we have too much CO₂ in the atmosphere at present, and our global CO₂ emissions per year, already above what the environment can naturally purge, are increasing. If that weren't the case, there would be little reason for President Bush's call to action.

Let's look at an analogous situation. For example, let's say you have a bathtub that's three-fourths full of water. The drain is open, but it's partially clogged, and so it's draining slowly.

In addition, the faucet is turned on, putting more water in the tub. It so happens that the water is currently coming into the tub faster than the partially-stopped drain can take it out, so the water level is rising, causing fears for the well-being of the bathroom floor.

The person controlling the faucet is opening the faucet as we speak, letting water come into the tub at an ever faster rate. That person, realizing the risk to the floor, promises to stop opening the faucet anymore in about 15 minutes.

What do you think will happen to the floor?

Even with the rough data I supplied, I hope you can see that the water will rise increasingly rapidly for the next 15 minutes. If the person takes their hand off the faucet in 15 minutes, the water will continue to rise until it overflows the tub (assuming it doesn't overflow sooner). The only way to save the floor is to reduce the flow of water from the faucet to below the flow of water out of the drain before the tub overflows. Even if they started reducing the flow of water out of the faucet now, the water in the tub would still rise until the inflow was less than the outflow.

Of course, this is a silly little example; the real world of GHG emissions is much more complex. Yet the general principle of stocks and flows holds: as long as the inflow exceeds the outflow, the stock will rise.

I'm not about to use this short, informal essay to argue for or against specific GHG or climate proposals or to try to balance climate stability against economic stability. I am suggesting that we all remember the lesson of stocks and flows when we are thinking about or evaluating policies such as these.

PS: Thanks to colleague Wayne Wakeland for, in a totally different situation, reminding me of the effectiveness of simple bathtub models (and I hope it worked here!).

Cassandra's curse and LTG

Almost two years ago, I posted about Limits to Growth: The 30-Year Update. Whether you saw that posting or not, I suspect you know Limits to Growth, often referred to by its initials as LTG.

Now Ugo Bardi has written Cassandra's curse: how "The Limits to Growth" was demonized in The Oil Drum: Europe. It's his view how LTG started to stimulate true dialog about a major challenge for the planet and how it then became "everyone's laughing stock" (well, perhaps not everyone's).

That's changing. As Bargi notes,

Climate studies have also brought back the limits of resources to attention; in this case intended as the limited capability of the atmosphere to absorb the products of human activities. In this field, the LTG study can be seen as having taken the right approach from the beginning; modeling for the first time the interaction of the environment with the human industrial and agricultural system.

If you've not read LTG, I encourage you to read it now. If you'd like, you can buy Limits to Growth: The 30-Year Update online, or you can find it in your favorite library (you can change the country or specify the location more precisely).

IMT 586: Information Dynamics I

If you're in the Puget Sound area and have been thinking about enrolling in IMT 586 (Information Dynamics I, called system dynamics by most of the rest of the world) at the University of Washington, now's the time; the quarter starts next week.

For more on the course, see my two prior announcements.

System dynamics course

Have you heard of system dynamics here or in your reading elsewhere? Would you like to learn more, including how to create computer simulation models to make sense of some of the challenges and puzzles you face, be they at work or in the news?

The University of Washington Information School is offering IMT 586, a first course in system dynamics, in the winter quarter. Yes, I'll be teaching it. You can learn more about it, including tips on how to register, in my earlier posting called Information Dynamics: IMT 586. My instructor class description lists the three goals I have for this course. For anyone concerned about the level of mathematics required in this course, that page also points to a brief description by the author of our text describing the level of mathematics needed to do this work.

I look forward to meeting some of you in that class!

Information Dynamics: IMT 586

Have you ever wondered ...

• what causes some ideas, products, and companies to become fads that peak and die, while others have staying power?
  
  
• why there are business cycles?
  
  
• what causes some diseases to become epidemics and others to subside with little effect?
  
  
• why real change often takes so long?
  
  
• the role information plays in the answer to each of these questions?
  

Would you like to learn to answer these and other such questions
yourself? Are you a student at the University of Washington, or do you live within commuting distance?

Then sign up for the Information School's IMT 586, Information Dynamics I, in the Winter Quarter 2008. I look forward to seeing some of you there.

Focus on the patterns, not the events

If you've ever seen the video that accompanies The Beer Game, there's something eerily familiar in the news about real estate in the USA. (Disclaimer: The video I saw was a VHS tape with a PBS segment on a previous boom and bust cycle in real estate. I can't promise the current DVD contains the same material.) Despite 50 years of knowing that the principles of feedback control theory apply to human and organizational systems, we still create systems with poor information feedback that get us into the ecstasy of boom times followed by the despair of busts.

What does this mean to us, assuming we're not directly impacted by current real estate woes? Where do you see the potential for boom and bust in your world? How do you know? How do you test your hunches?

I can't tell you when you'll experience a bust, but I can help you discover how you can design a business or organizational system that is less likely to experience such boom and bust cycles.

Incidentally, if you'd like to play The Beer Game but don't think you have an opportunity, check out the online version.

Rehearsing

I often help people with presentations, and I've noticed that those who rehearse seem to be those who do better. Now Garr Reynolds of Presentation Zen has done an excellent job of explaining the creative process of presenting ideas to others in his Steve Jobs and the art of the swordsman.

Note the two keys to presentation success:

• Intense rehearsal in a team setting
  
• Absolutely no attention to technique or form in the actual presentation
  

Reread Garr's comments, if you need to, and note comments such as, "...once we allow our mind to drift to thoughts of success and failure or of outcomes and technique while performing our art we have at that moment begun our sure decent." [sic]

How can we possibly get through a presentation while following the second key? By following the first key until we have internalized what we want to say, how we want to say it, how others will hear it and respond, and what we can do if something goes differently than we expect. Then we have to rehearse it some more.

As someone once noted, we often rehearse something until we get it right. That means we may have done it wrong 20 times and right once; which do you think will stick with us better?

I think the same thing applies in other areas of our professional lives, and I think Dietrich Dörner and Harald Schaub might agree. That's why I wrote A somewhat unified view of decision making: to suggest the importance of spending time wrestling with what we do at a time that's apart from the actual doing. Whether we use computer simulation, scenario planning, role playing, or something else, the opportunity to rehearse what we do professionally before we do it and to learn from what we actually do afterwards to improve for next time is exceedingly valuable. And it's the cyclic action learning that helps us improve and helps keep us from getting fixated on a bad idea.

If you're still thinking of presentations, check out Garr's presentation tips.

Is business getting worse?

Bloomberg says "CEOs See `No Clear Signs' of Crisis as Woes Intensify." Are things really getting worse, even as people put smiles on their faces, as that article seems to indicate?

We obviously won't know for a while. Even if things get worse for some companies, others will likely do okay, and some will thrive (or, if things go well, others will likely do okay, and some will suffer).

To a large degree, the key is being good at responding to what happens, not simply what happens. We get good by being lucky, by thinking clearly, or by having been in this situation before and having learned (or by some combination of those). While I have no help for you in the luck category, there are myriad approaches to thinking clearly, and I've tried to touch on a few in Making Sense With Facilitated Systems.

You might say that there's no way to experience the future before you get there (the third alternative). As Dietrich Dörner and Harald Schaub point out, that's not necessarily the case. Simulation (system dynamics, usually) is a way to explore challenges we might face in the future and to learn which strategies are likely to be more successful.

How are you preparing for the challenges you might face? If you'd like to talk about some of the possibilities, drop me a line.

Systems thinking and art

Not long ago, The Diagram published one of my system dynamics diagrams because they were attracted by its design ("But is it art?"). Recently I discovered a quotation I wanted to share with you, for I think it conveys something important about about organizational or societal simulation and modeling as well as about art.

Suggestion—the part standing for the whole—is a principal means by which art communicates; this is why art often tells us so much with such economy.

Jane Jacobs, The Death and Life of Great American Cities, p. 377.

That's what system dynamics in particular and systems thinking in general is all about: economy in expressing the essence of a problem to foster economy in solving the problem and economy in creating deeper insights to be able to solve the next, similar problem.

I like that quotation.

I like the book, too; I'll probably write more about it soon.

Why are we headed there?

Yesterday I posted a rather dark story about what we could be facing. As with any system dynamics model, it's showing the likely effect of the modeled policies; it's not predicting the future. (There is a difference; call if you'd like to chat about it.)

If that message is so dark, why don't we see much action to change our course? (Admittedly, we're beginning to see more action now.) The Oil Drum published Living for the Moment while Devaluing the Future, an essay exploring just that question. For a more academic approach, see Larry Karp's Global Warming and Hyperbolic Discounting, an article referenced at the bottom of The Oil Drum article (now on my reading list—I've only had time to skim it so far).

I think these ideas are important to understand and explore as we try to craft a "soft landing" from our ecological overshoot.

I think they may also be important to us in business. We get used to constant discount rates, because that's what we use. Do our customers and our bosses (they do have similar roles in our lives) really think that way, or do they do hyperbolic discounting, too? Would it it make a difference which they use? Would that difference imply we should act differently than we do?

Where are our policies leading us?

A policy is a set of guidelines or rules by which we make decisions. We have certain policies by which we work and live, even if we don't always make those explicit. If we see a pattern of decisions, decisions that seem cut out of the same mold, that's likely evidence of a policy.

François Cellier of the Institute of Computational Science of ETH Zürich, Switzerland has published an article, Ecological Footprint, Energy Consumption, and the Looming Collapse, at The Oil Drum that examines the potential effects of our policies towards growth. It's a high-level view, to be sure, but sometimes those offer great insights. Be sure to read both the article and the accompanying slide set (the article isn't that long; it's the 333 comments that take up most of the length).

I think this is a very important discussion. That's why I think it's important for each of us to be skeptical about such claims. It's not because I think he's wrong; his analysis, at least so far, seems good. It's not a call to wait for "proof," for, as John Sterman points out, we're not really waiting; we are doing things to the environment every day. It's not a call to ignore the claims, for that's not being skeptical; it's a call to test them and then to act based on what we determine. It's not a call for depression; Cellier does show a way forward (especially in the slides).

By suggesting we be skeptical, I may give the impression I think we can ignore this for a bit. I want to re-emphasize that the IPCC and others have given some pretty clear signals that the time to act is now (actually, the time to act was some years in the past; the next available time to act is now).

What does this mean for our businesses and for business in general? I think it means figuring out what to do to ensure the sustainability of our businesses and our economic system in the face of the challenges the best science says await us. The key lesson from "Out of Gas: A Systems Perspective on Potential Petroleum-Fuel Depletion" was that we not wait too long to attend to signals we get, for our systems have inertia, and we can't, as much as we might wish, always change direction instantaneously. Pay attention to Cellier's description of easy and difficult problems starting on slide 38; the signals may not be as we'd normally expect. Sometimes we can't wait to feel the wind from an impending storm; we have to rely on forecasts from meteorologists to know when to board up windows in the face of an approaching hurricane.

We can also apply that message to more typical business decisions. Do we discover we will need to add (or remove) capacity well in advance, so we can react smoothly, or do we make such discoveries only when the market begins to complain loudly? How do we figure out whether our latest initiative is about to make real progress or it's about to fail and we should abandon it and change course?

Reacting too soon can lead to the Chicken Little trap: if we respond too quickly, we diffuse our energies by responding to simple noise; if we respond too slowly, we're trapped. One of the lessons I've learned is that feedback models (of the sort I've sometimes discussed here, also called system dynamics models) can help us find what things to monitor so that we have a clearer picture to guide our decisions.

What do you think?

IMT586: a course in system dynamics

While I have taught system dynamics and systems thinking courses and workshops for various organizations, some have asked if I was planning any public courses. I can now tell you that I'm the lecturer in spe for IMT586, a graduate course in information (system) dynamics for the Information School at the University of Washington. IMT586 will be offered starting in the winter quarter of 2008. It will be offered in the MSIM program to both day and executive students. It currently appears it will also be open to non-matriculated students, so, if you live within commuting distance of the University of Washington's main Seattle campus and are interested in learning system dynamics, check it out.

As we get closer and more information is available on the UW iSchool Web site, I'll post updates on Making Sense With Facilitated Systems.

Of course, if you want custom training tailored for your organization's needs, feel free to contact me.

An accidental experiment

I've published a number of At Any Rate™ columns through Pegasus Communications. They consist of text designed to capture people's interest up front and to remind them of what they experienced later as well as a simulation model people can download and explore. The model leads people through three stages: an initial stage-setting exercise, a more complete model to show added complexity in the problem at hand, and an exploration area where people can dig a bit deeper to try their hand at addressing the problem.

Pegasus Communications advertises each At Any Rate in their free Leverage Points newsletter that has a rather large circulation, and they set up a discussion area in their Pegasus Forums for each one.

In other words, that column seems to be planted in a fertile ground in which to talk about such things. The models are interactive. They tell a story. They are published on a high-visibility site and advertised in a high-distribution newsletter. There's a space established to enable discussion.

Yet I've gotten very few off-the-record comments (all favorable) about those columns. I've seen very few comments in the Pegasus Forum. I'm not sure anyone has contacted me about what they've seen there. I'm not complaining; I know that I don't write letters to the editor of the local newspaper, even if I strongly agree or disagree with what the newspaper has published. As a result, I don't necessarily expect (although I would welcome) lots of dialog about what I post online.

On Monday, April 9, I published a similar model (new URL) on Drew McManus's Adaptistration as part of his TAFTO 2007 (new URL) series. It was not interactive; rather, it contained diagrams, graphs, and a computer program (or a text-based model, which is the same thing). Admittedly, I tried hard to use literate programming ideas to intertwine the model and the story so that it would be more interesting and readable, and I let two others in the potential audience see an advance copy so I could find and fix any impenetrable sections.

Within a day, I had a thoughtful, lengthy comment added to that column. Two bloggers made quite favorable comments about the essay. I know of at least one person who had been telling me he'll get to the latest At Any Rate any day now who read and commented on my TAFTO column within a day.

What gives?

While I realize that the singular of data is anecdote, I think that this is showing me the barrier we erect when we ask people to download, run, and learn from an interactive model. While the barrier might be lower if I had used a simulator for At Any Rate that ran in a browser, I'm not sure; one would still need to take perhaps half an hour, perhaps more, to work through the model. It takes much less time to install the software, and i'ts a one-time action—a number of readers already have it.

The barrier may be more complex than simply the challenge of installing the isee Player required by the At Any Rate column. To explore and really learn from a simulation, someone needs to be willing to experiment. That means taking the time to understand the environment, to formulate hypotheses, to write those hypotheses down, to run various tests on the simulation model, to compare the results of the test with the hypotheses, and probably to try new experiments based on the learnings from initial experiments. That's far different than just opening the application, pressing a few buttons, and seeing what happens.

Needless to say, most of us who create such interactive simulations try hard to guide the user through the process. Most of us encourage people to form those hypotheses and to document them in writing or in graphs before starting a simulation. Yet I know (from personal experience—I'm not immune) that it's far too easy to treat a simulation as a video game: press the button, and see what happens. That's not often the path to deep learning.

With the non-interactive version, people can read just a paragraph or skim the entire article to see if it seems interesting. They can come back later to dig more deeply. They can print it out, if they wish, and read it on the bus on their commute. If a text version of the model is included, they can, if they want, copy it into the simulator and explore it themselves.

My lesson? Interactive simulation is no panacea, and it may be a disadvantage if I want to get my story told, especially if my audience consists of busy or high-level people. By telling a good story, I can help the reader learn something, most likely in less time.

Is there a role for exploration, experimentation, and interactive simulations? Certainly! But I need to be sure to consider the audience, their current interests, and what they know and want to learn.

I'd welcome others' insights and experiences. In an action research sense, I'll spend some time trying to disconfirm my conjecture; in the process, I might learn more.

TAFTO 2007

As promised, today is the day my article Is TAFTO a good idea? Really? (new URL) has been published at Adaptistration. If you're interested in classical music, I encourage you to take a look and see if you agree with my thinking. If you're more interested in making sense of tough problems, I also encourage you to take a look. I've used some graphs I don't normally see in such modeling work. In either case, I'd be interested in your reactions.

If you happen to be here because you found me on Adaptistration, welcome! You must be a lover of classical music (you may also be an orchestra administrator, a musician, a board member, or all of the above). In that case, you might also enjoy a recent conversation I had with Greg Sandow called Making musical sense by email, and you might like a short exploration I did of a statement in the recent Knight Foundation Magic Of Music Final Report called Making sense with numbers.

I'd enjoy having you as a regular reader of Making Sense With Facilitated Systems, and I'd welcome your comments either here or on the Adaptistration article.

More on factorial designs and simulation

Factorial designs and simulation: apparently Professor Barlas is teaching it in Istanbul!

For the analytically-minded among you, I'd note that MCSim, coupled with a bit of infrastructure you can develop, can make running small factorial experiments (up to a few hundred runs) a fairly quick and painless task. That includes their development, their execution, and their analysis. If a system dynamics model is deterministic, as many are, there's no need for replication.

You can see a simple example in my upcoming TAFTO contribution. I was pleased at the way in which a factorial design approach enabled me to generate and see useful results from a relatively simple model.

Making musical sense by email: the table of contents

For the past two weeks, I've been posting excerpts of an email conversation between Greg Sandow, composer, consultant to orchestras, author, professor, and music critic, and me about the future of classical music. I thought you might be interested in reading these excerpts, either because you're interested in the future of classical music or because you're interested in observing light-weight approaches to using systemic approaches as we seek to make sense of tough problems.

Because the conversation is spread across multiple postings, I'm providing this page as a table of contents for the series—a list of links to help you read the series in order from start to finish or to help you find the one section you're seeking.

1. Part 1: The introduction
   
2. Part 2: My first email to Greg
   
3. Part 3: Greg's reply
   
4. Part 4: My first email with higher-resolution graphics
   
5. Part 5: An augmented model
   
6. Part 6: What did you observe about the dialog?
   
7. Part 7: What did I observe about the dialog?
   
8. Appendix A: The first model
   
9. Appendix B: The second model
   
10. Special feature: An essay by Dr. Glenda H. Eoyang
    
    

If you're just discovering this series, welcome! It's a bit lengthy, so you may want to read it in several sittings. All of us involved in creating this series would appreciate any comments you might have.

Thanks to Greg for the dialog and for permission to post his words here and to Glenda for her contribution that indicates this approach is gathering momentum elsewhere, too.

Making musical sense by email, special feature

One of the primary reasons to publish this series was to suggest that system dynamics, one methodology in the larger systems thinking field, could be applied beneficially in a low-key way to help make sense of complex situations. Instead of making a big deal of talking about system dynamics as a useful approach, getting buy-in to try it on a particular case, and then doing it, we can just apply it naturally when and as the need arises. Sure, those of us doing the work need to educate those seeing such models for the first time so they know what they're getting and how to understand them, but I think this series has demonstrated that it's possible to do that as part of the process, not up front.

It turns out, not surprisingly, that I'm not alone in having these thoughts. Friend and colleague Dr. Glenda Eoyang has been having similar thoughts regarding a different systems thinking methodology. Glenda started exploring nonlinear dynamics and social systems in 1989, received her doctorate in Human Systems Dynamics in 2002, and founded the Human Systems Dynamics Institute in 2003. She teaches, consults, researches, and writes. She helps people see patterns that emerge from the chaos of human interactions and take adaptive action to increase coherence, health, and sustainability for individuals, teams, institutions, and communities. Her profound understanding of the many theoretical streams of complexity science and her gift for clarity make her an excellent guide into the world of human systems dynamics.

Because she's using similar approaches in a related field, I invited her to share her thoughts with us.

---

Technology matures and, thank goodness, we do, too. In Seattle there is an elegant hotel that was built by Ford Motor Company early in the last century. Gentlemen would live in the hotel while they learned the basic skills of car ownership, including driving and auto mechanics. Six months after the hotel was built, Ford found more efficient ways to meet their customers’ needs. I, too, have been seduced by the new. Fully one quarter of the first computer course I wrote covered binary arithmetic and the history of computing. Today, only the mathematicians and historians find that stuff interesting. In the dawning phases of a technology, bridges to the past are critical. As the technology emerges it integrates into our other intelligences, and we return our focus to the work at hand. The technology becomes a means rather than an end in itself. In future I hope we will be as amazed that people spent days learning “systems thinking” skills as we are with Ford’s hotel and the history of computing as a core competence for users.

When I discover a new technology it seems complicated and exotic. I want to understand its secrets and plumb its depths. For a short time, the technology itself is a preoccupation. I focus on it as if it were an end in itself. Over time, though, I become accustomed to the new ways to think and act. I absorb the new views and tools into my repertoire. They become a part of me, and I am able to see through them rather than focusing on them directly.

Today, my clients are more ready to think systemically than they are to learn about systems thinking. I believe this is the transition Bill and others are seeing in themselves and their clients. For example, a colleague who is a professional evaluator doesn’t design and implement “evaluation systems.” Rather, she works as part of the management team to generate and present meaningful data in response to specific strategic and tactical questions. Another Human Systems Dynamics Associate works in a school system, using the language of education and educational reform to spark conversation and action about complex human systems dynamics. I’m supporting a strategic planning process for a fast-growing international consulting firm. I introduce tools and techniques only in service of the conversation toward the organization’s business goals and improved performance. I add value not because I bring an exotic set of mysterious tools but because I use powerful tools to help them think and act with more insight, intention, and collaboration.

This transformation isn’t easy for me. I like binary arithmetic. I feel powerful when I hold the keys to a mysterious new discipline. On the other hand, my clients find the transition quite appealing. We work together on their concerns, leveraging their knowledge and expectations, rather than asking them to leave their world views behind and align with my arcane methods and visions of reality.

I am beginning to think of myself as a “praxis partner”—one who works with others to blend theory and practice in the service of effective action. As the technology of systems thinking matures, I hope my clients and I can, too.

Glenda H. Eoyang, Ph.D.
Executive Director, Human Systems Dynamics Institute

---

What are you seeing in your organizations? Are people doing more of this blending there, too? Is that helpful, or do you miss something in the process? Both Glenda and I would enjoy hearing your feedback. Comment here, or contact Glenda or me directly.

Making musical sense by email, appendix B

Here is the second of the two models I did in support of the dialog Greg Sandow and I had on the future of classical music. This one models the added effect that might occur if people stop attending concerts after they turn 30. Some stop forever, while others return when they turn 50. I left the model as close as I could to the previous one.

States = {
twenties, thirties, forties, fifties, sixties,
thirtiesNOGO, fortiesNOGO
 };

Inputs = {
newpa,  # Number of young people becoming concert-goers pa
};

Outputs = {
input,
avgage,
total
};

decade = 10.0;
bins = 5.0;   # no. of decades modeled
population = 1.0e6;  # total initial concertgoers

fractionNOGO = 0.50;  # fraction of twenties who leave for two decades
fractionQUIT = 0.20;  # fraction of twenties who leave forever

Initialize{
twenties = (population / bins) * (1 + fractionQUIT);
thirties = (population / bins) * (1 - fractionNOGO);
thirtiesNOGO = (population / bins) * fractionNOGO;
forties = (population / bins) * (1 - fractionNOGO);
fortiesNOGO = (population / bins) * fractionNOGO;
fifties = population / bins;
sixties = population / bins;
}

Dynamics{
from20spa = twenties / decade;
to40spa = thirties / decade;
to50spa = forties / decade;
to60spa = fifties / decade;
endingpa = sixties / decade;

quittingpa = fractionQUIT * from20spa;
hiatuspa = fractionNOGO * from20spa;
to30spa = from20spa - (quittingpa + hiatuspa);

to40NOGOpa = thirtiesNOGO / decade;

returningpa = fortiesNOGO / decade;

dt(twenties) = newpa - (to30spa + quittingpa + hiatuspa);
dt(thirties) = to30spa - to40spa;
dt(thirtiesNOGO) = (hiatuspa - to40NOGOpa);
dt(forties) = to40spa - to50spa;
dt(fortiesNOGO) = (to40NOGOpa - returningpa);
dt(fifties) = (to50spa + returningpa) - to60spa;
dt(sixties) = to60spa - endingpa;
}

CalcOutputs{
input = newpa;
total = twenties + thirties + forties + fifties + sixties;
avgage = (25.0 * twenties + 35 * thirties + 45 * forties
+ 55 * fifties + 65 * sixties) / total;
}

If the last model was quick and simple, perhaps this one is quick and dirty, for it had much less testing than the other one. I did it primarily to start a conversation we haven't finished yet, and I'll spruce up this model and test it more adequately as we go.

The reason to show this to you is to show how easy it is to change such a model to incorporate new conjectures. What's more, because these are text models, it's quite easy to use diff to see what's changed between the two. Because the entire model is in text, you can see the entire thing at one glance.

Here's the associated simulation file that describes the experiments. As you can see, it's almost like the other one; I just had to add the new parameters.

Integrate(Lsodes, 1e-6, 1e-6, 1);
OutputFile("simplerun02.out");
StartTime(0.0);
fractionNOGO = 0.50;
fractionQUIT = 0.20;

Simulation{
newpa = NDoses(2,20000.0,10000.0,0.0,10.0);
PrintStep(input,0.0,200.0,1.0);
PrintStep(total,0.0,200.0,1.0);
PrintStep(avgage,0.0,200.0,1.0);
}

Simulation{
newpa = PerExp(20000.0, 400.0, 0.0, 0.010);
PrintStep(input,0.0,200.0,1.0);
PrintStep(total,0.0,200.0,1.0);
PrintStep(avgage,0.0,200.0,1.0);
}

Simulation{
newpa = NDoses(2,20000.0,0.0,0.0,10.0);
PrintStep(input,0.0,200.0,1.0);
PrintStep(total,0.0,200.0,1.0);
PrintStep(avgage,0.0,200.0,1.0);
}
END.

For more on the syntax used in these files, see the documentation for MCSim.

Eagle-eyed readers will catch a spelling error and a mistake in the version I published previously. I've fixed those problems in the model shown here. Interestingly, the simulation outputs only change slightly, and their shape is the same, both because the dynamics are heavily controlled by the main aging chain and the cutoff of new attendees and because the mistake substituted a variable with a similar value. Thus I'm not providing new graphs.

In actually working on this model, I found it faster to print all the stocks and flows and then use the analysis program to select which variables to explore. In this appendix, I'm only showing print statements for the key variables used to produce the graphs in the report, so you have less text to read.

Making musical sense by email, appendix A

Because I believe in transparency as far as possible, here is the first of two models I did in support of the dialog Greg Sandow and I had on the future of classical music. Neither model is a paradigm of modeling artistry; they're simple, quick models to support our dialog and exploration. If we begin to use these more deeply, we'd certainly test and document them more seriously.

Here's the model I used to create the initial graphs I shared with Greg and Drew:

States = {
 twenties, thirties, forties, fifties, sixties
   };

Inputs = {
 newpa,  # Number of young people becoming concert-goers pa
 };

Outputs = {
 input,
 avgage,
 total
 };

decade = 10.0;
bins = 5.0;   # no. of decades modeled
population = 1.0e6;  # total initial concertgoers

Initialize{
 twenties = population / bins;
 thirties = population / bins;
 forties = population / bins;
 fifties = population / bins;
 sixties = population / bins;
}

Dynamics{
 to30spa = twenties / decade;
 to40spa = thirties / decade;
 to50spa = forties / decade;
 to60spa = fifties / decade;
 endingpa = sixties / decade;

 dt(twenties) = newpa - to30spa;
 dt(thirties) = to30spa - to40spa;
 dt(forties) = to40spa - to50spa;
 dt(fifties) = to50spa - to60spa;
 dt(sixties) = to60spa - endingpa;
}

CalcOutputs{
 input = newpa;
 total = twenties + thirties + forties + fifties + sixties;
 avgage = (25.0 * twenties + 35 * thirties + 45 * forties 
  + 55 * fifties + 65 * sixties) / total;
}

I've stripped out most of the comments to make it shorter; I suspect most of you who are familiar with system dynamics modeling (and many who aren't) can probably understand what's going on.

This model is written for the MCSim simulator created by Frédéric Bois and Don Maszle.

Here is the simulation file that runs the various experiments:

Integrate(Lsodes, 1e-6, 1e-6, 1);
OutputFile("simplerun01.out");
StartTime(0.0);

Simulation{
  newpa = NDoses(2,20000.0,10000.0,0.0,10.0);
  PrintStep(input,0.0,200.0,1.0);
  PrintStep(total,0.0,200.0,1.0);
  PrintStep(avgage,0.0,200.0,1.0);
}

Simulation{
  newpa = PerExp(20000.0, 400.0, 0.0, 0.010);
  PrintStep(input,0.0,200.0,1.0);
  PrintStep(total,0.0,200.0,1.0);
  PrintStep(avgage,0.0,200.0,1.0);
}

Simulation{
  newpa = NDoses(2,20000.0,0.0,0.0,10.0);
  PrintStep(input,0.0,200.0,1.0);
  PrintStep(total,0.0,200.0,1.0);
  PrintStep(avgage,0.0,200.0,1.0);
}

END.

For more on the syntax used in these files, see the documentation for MCSim.

Making musical sense by email, part 7

The conversation goes on, or at least I hope it does, but I thought this was likely enough to give you a flavor of what we have done. What do you draw out of that story?

I think there are a few key points to note about the process. Before I start, I should note, in case you missed it, that we've been applying a methodology called system dynamics to the problem Greg Sandow posed. I've only mentioned "system dynamics" twice so far in this series and never in my emails to Greg.

1. System dynamics can help us think more clearly. Part of that comes from the act of making our mental models more explicit and more precise, even when we're still viewing the world from 10,000 meters up. Part comes from the lessons we learn by considering the effect of feedback on the behavior of the system that's creating our problem. In this case, we used simulation to explore the effect of feedback.
   
   In other words, system dynamics can make us (seem) smarter.
   
   
2. Because system dynamics leads us to make our mental models very clear and explicit (even at the 10,000 meter level), we also clarify what it is we don't understand. That can help identify and resolve differences in understanding more quickly.
   
   Because system dynamics has the potential to make us look more unambiguously dumb, it can seem risky; fortunately, the benefit of thinking more clearly usually offsets the risk of clarifying our ignorance in front of others, and the two (clearer thinking and revealing our ignorance) together can help us learn and move forward more quickly.
   
   
3. System dynamics models and system dynamics "interventions" don't need to be big, overarching affairs that take the lead in organizational work; they can be inserted casually and naturally into the conversation as partners in an effort. As you saw, I never once told Greg that I wanted to do or was doing a system dynamics model nor why system dynamics was special. I think I was able to introduce enough information so he could follow what I was saying without inserting too much jargon. I think he was able to assess the utility of the model by what we did discuss.
   
   I think this is healthy. While big programs and up-front acceptance may be important in some cases, I think we respect our clients and our managers if we don't ask them to buy into a process before they see the results. Focus on the problem, not the tool.
   
   There's a benefit for those of us using system dynamics, too. System dynamics isn't the best tool for every problem. If we've sold a client or manager on system dynamics as the way to proceed and it then becomes evident there's a better way for this particular problem, we risk seeming to have failed. We can either change courses and make that risk real, or we can push ahead, staying with a less-than-optimum approach. By inserting system dynamics dynmaics more naturally, we can change courses as circumstances warrant, without having to eat too many of our words along the way.
   
   Of course, that requires that we have skills in (or connections to people with skills in) a variety of approaches.
   
   
4. It's possible to use system dynamics even by email. Text-mode email is quick (relatively), easy, and encourages interaction. Text-mode graphs make it easy to incorporate graphical data into the conversation, and text-mode drawings make it easy to show simple model diagrams. Text-mode models make it easy to convey the full detail of the model we're using to the degree and in the manner it's useful. I like that.
   
   Of course, some graphs and some graphics are too complex to show in text; for that we use other means. And some people, such as Greg, will prefer higher-quality graphs, while others will be quite happy with the text-mode graphs.
   
   
5. You don't have to simulate everything. Sure, our insights about nonlinear feedback systems aren't always very good, but there are times when we learn enough from a simple simulation so that we can carry useful lessons over into the more complex situation we really face. There are times we can draw on past experiences with simulations to understand the problem we're currently facing. There are times when we can glean enough insight out of the model we've created to do a useful, informed static analysis.
   
   In this case, I simulated a simple "aging chain" that simplified Greg's problem rather than building a more complex model that replicated the exact dynamics seen in US orchestra attendance. That model seemed to give insights useful enough to guide our discussion profitably.
   
   
6. You don't have to use system dynamics for every problem. System dynamics works superbly in addressing problems involving feedback, but it's not the only systemic approach we have. Pick the methodology that best suits your problem. Better yet, consider viewing your problem with multiple methodologies to see if that triangulation leads you to consistent solutions; if you get differing recommendations from multiple methodologies, perhaps you have more work to do.
   

There are a few lessons to be drawn from the musical content of this series, too:

1. In general, you won't change the average age of concert-hall audiences over the long haul by reducing the number of newcomers. While you may certainly affect that average age over the short term, the system will recover to its old equilibrium eventually.
   
2. If you cut the inflow of newcomers to zero, you will make the average age change to a new value.
   
3. Based on some results I shared with Greg but didn't show here, the average age recovers more slowly with more drastic drop-offs in newcomers. For sufficiently drastic drop-offs, it may take years or decades to distinguish between a drop to zero and just a drop to a drastically lower number.
   

Because Greg is working on a book on this subject, I encourage you to check out his blog to explore more lessons about the musical content of this series and to enter into a dialog with him. Check out his interests, too; I think you'll find he enjoys a broad spectrum of musical styles.

If you're more focused on orchestra management, see Drew McManus' Adaptistration.

You can also see another model I'm working on that addresses classical music in my upcoming TAFTO contribution; I'll announce that here when it's published.

I'd be remiss if I didn't thank Greg for the dialog we've had and for his willingness to share it here. Thank you, Greg. All of his words are published in this series with his permission.

Stay tuned for the final postings in this series next week, including a full listing of both models, a table of contents with links to each article in the series, and the possibility of a surprise guest essayist!

Making musical sense by email, part 6

The conversation goes on, or at least I hope it does. I hope these snippets have been sufficient to give you a flavor of how Greg Sandow and I have been talking.

Tomorrow, I'd like to give you my impressions of the meaning of all this. Before I do, I'd like to ask you to contribute. How do you interpret what you've read in this series? What value do you see in the approach? What did you see that you didn't expect to see? What didn't you see that you expected?

There's one particular question I asked earlier: do you think my text-mode graphs or the higher-quality graphs would have been better for facilitating dialog if this dialog had been between you and me, assuming the graphs weren't too complex?

I know you're largely a quiet group of readers, but I would like to learn from your impressions—perhaps we could all learn. Add a comment below, or, if you'd prefer, send me an email or give me a call.

I encourage you to take part in this conversation in other ways, too. The future of music is a project that Greg is working on for a future book. If you've got ideas to contribute to that effort, head over there or to his regular blog. If you're more interested in orchestra management, also check out Drew McManus' Adaptistration (you still need to think about the future of music).

If you'd like to talk more about approaches such as this for making sense of the business and organizational problems you face, keep following Making Sense With Facilitated Systems.

Then come back tomorrow to see my thoughts.

Making musical sense by email, part 5

So far, we've seen Greg Sandow's thoughts on the future of classical music and a dialog between Greg and me involving the use of system dynamics simulations to explore his ideas further. Hopefully this is showing you a way to use systems thinking that you may not have seen before; perhaps you'll see ways to apply it more directly in your work, too. We'll talk more about that later.

Based on my initial model, Greg had revealed a few more details of his thinking. In particular, he wasn't necessarily suggesting that young people were abandoning classical music forever; they might just be staying away for a decade or two.

Thanks to Greg's questions, I augmented the model to show what might happen if classical music concert-goers take a hiatus in their thirties and forties. I describe that in the lightly edited email below, including a quotation (with permission) from another email from Greg not yet quoted here:

"Greg Sandow"  writes:

> First, I believe your model posits that people start
> attending concerts, and then continue essentially
> throughout the rest of their lives. I don't know if that's
> true. That is, people might go occasionally when they're
> young, then not go (or not go very often) for many years,
> and then resume going, much more often, when they're
> older. This would be consistent with some of the data, for
> instance the preponderance of the audience in older age
> groups, and also survey results that show people most
> likely to attend regularly when they no longer have
> children at home. This may not have been the pattern in
> past generations, but it appears to be the pattern now.

Greg,

I've modified the model and run some quick and dirty tests
to let you see the first impressions.  If this happens to be
at all interesting, I'll need to do some model testing I
don't have time for right now.

I did think of one other factor that could lead to an aging
of audiences; see the end of the attachment.

> An impressive study done in Indianapolis a few years ago
> showed that people 40 and under learned about arts events
> they attended mainly by word of mouth. No other source of
> information came even close. Orchestras have little
> understanding of this, and have done very little work, as
> far as I know, to understand what makes people decide to
> go to concerts, or - really important - to go once, and
> not return.

Quite interesting.  That's consistent with the assumptions
Drew and I have been making, and it's consistent with the
purpose of TAFTO.

Perhaps the TAFTO model will be of use, when it's published.

Thanks,

Bill

The attachment I mention contains the results I shared with Greg (with one obvious typo fixed). That model is not very well developed, nor have we carried that part of the conversation much further yet, but I thought you might like to see that it was relatively easy to pursue alternative ideas.

TAFTO refers to Drew McManus's Take a Friend to Orchestra initiative. I'll be writing a simulation-supported column for his annual April TAFTO push. You'll have to wait until then to see more about that model.

Making musical sense by email, part 4

In response to Greg Sandow's request to see the actual graphics instead of text-mode graphics as shown in my first email to him, I sent him a PDF with the same text and with higher resolution graphics. I also made a few comments in response to his email (lightly edited here). Greg's comments are included with his permission.

> Thanks so very much for sending me your work. It's
> remarkable stuff, much more solid and promising than most
> of what I see. And your tentative conclusion is quite
> challenging (at least for the classical music business) -
> that the model that best predicts the observed decline in
> attendance (am I getting this right?) is the one that
> predicts no new people join the audience at all.

Greg,

Thanks for your response filled with ideas and insights.
Yes, you got that right; the simulation run that best
matches your data is one in which no more young people enter
a concert hall.

> I'd make one request. I'd much prefer to see the original
> graphics. I did look at your e-mail in Courier, but it
> would be a lot clearer if you could attach the charts as
> graphics, or perhaps send a spreadsheet with the charts
> generated inside it.

I'll regenerate the graphs in a higher resolution format.
I've got a few things on my schedule today, but I'll try to
get to it before the end of the day -- no promises until you
see them, though.

> Also, I mustn't forget to congratulate you on your careful
> work with that Knight Foundation tidbit. I've seen that
> too often used to support the idea that all we need is
> music education in our schools, and classical music
> attendance will go up again. As if, in other words,
> there's a causal connection between playing an instrument
> and classical music attendance. As you rightly point out,
> the Knight study makes no such assertion, but people jump
> to the conclusion anyway. And the statement, which I
> believe they make, that playing an instrument is a
> predictor of classical music attendance does muddy the
> waters, since it rather casually puts playing an
> instrument earlier than attendance in some sort of implied
> statistical food chain. That helps people to jump to the
> causal interpretation.

Thanks!  I agree with your assessment that they didn't
really claim that connection, but they didn't make it clear
that they didn't claim it, either.

> Anyhow, I think you debunked this all really nicely. My
> own predisposition is to believe that there's no causal
> connection, but that instrumental playing and classical
> music attendance are both characteristics of some yet to
> be defined demographic slice, the demographic in question
> being the one most likely to attend classical concerts. In
> other words, roughly speaking, playing an instrument and
> going to classical concerts have a common cause, rather
> than one causing the other.

That makes sense; I don't currently know what that factor
is.

> Going back to the analysis you sent me, I do have some
> suggestions for refinements.  > First, I believe your
> model posits that people start attending concerts, and
> then continue essentially throughout the rest of their
> lives. I don't know if that's true. That is, people might
> go occasionally when they're young, then not go (or not go
> very often) for many years, and then resume going, much
> more often, when they're older. This would be consistent
> with some of the data, for instance the preponderance of
> the audience in older age groups, and also survey results
> that show people most likely to attend regularly when they
> no longer have children at home. This may not have been
> the pattern in past generations, but it appears to be the
> pattern now.

You understood the model correctly.  As you note later about
your numbers, this model is a surmise, too -- a
simplification of reality to see what behavior the dynamics
of a basic "aging chain" would create.

Your causal explanation makes sense, though; I'll try to get
to add that to the model.  In particular, I could add a high
rate of dropping out in people's twenties, for example,
returning in their fifties.  I don't know if that fits the
pattern, but it would indicate what dynamics that might
create.  The real world, of course, is more complex than the
model.

> When I made so much of the NEA's statistic about the
> number of younger people dropping off, I didn't mean to
> say that this would lead to any immediate and directly
> proportional decrease in total attendance. I was using the
> data more impressionistically. Accepting the apparent
> truth that people in this era mostly go to classical
> concerts when they're older, I surmised that this younger
> generation, once it was 50 years old, would go in smaller
> numbers than previous generations did. I should stress
> that this is nothing more than a surmise - an
> assumption. It seems reasonable, according to common
> sense, but might turn out not to be true.

One of the things about the "operational thinking" this
modeling encourages is that it leads one to make one's
assumptions very explicit.  Sometimes we learn from that
exercise.

> Though when I combine it with certain data about people
> who are currently 50 and above, the conclusion seems even
> more reasonable. One big change during the past generation
> is that older people aren't as committed to high art as
> they used to be. They now "consume" (if that's the word)
> both high and popular art. This makes them less like the
> committed classical music audience of the past, and thus,
> in my view, less likely to attend classical performances
> in the numbers their predecessors did. Now we have younger
> people who go to classical performances far less often
> than their own predecessors did. That suggests that when
> these younger people are in their 50s, they'll be even
> more culturally omnivorous than the present older
> generation is, and thus less likely to be committed
> classical music attenders.

So we might modify the model to show some twenties dropping
out into a two-decade hiatus, rejoining as fifties, while
others never return.  We could vary the percentage that go
each way and see what happens.

> Second (returning to my suggestions for you), you might
> want to refine your model to reflect the fact that the
> total number of orchestral tickets sold includes many
> tickets sold to the same relatively few people - the
> subscribers. Currently, subscriptions amount to about 60%
> of all tickets sold. This number has declined sharply from
> the 80% or so reported a couple of decades ago, and is
> considered likely to decline still further. Still, this
> needs to be an important part of any model that predicts
> future ticket sales. It's not enough simply to predict the
> number of people likely to buy tickets. We need to know
> how many tickets these people are likely to buy, in order
> to get some idea of what the total ticket sales in the
> future are likely to be. Certainly, in the final analysis,
> orchestras are more concerned with the total number of
> ticket sales than with the number of individual people
> attending.

I'm working on a column for Drew's TAFTO series.  In that
somewhat more complex model and with Drew's consultation,
there are three categories (stocks) of people: NotYets,
Nows, and NoMores.  The NotYets are those who don't know if
they'd like classical music concerts because they've never
gone to one.  The NoMores are those who have attended but
have left, never to return for whatever reason.

To your point, the Nows are those who currently make up
audiences.  That's not the number of tickets, though;
depending upon various factors I'm not modeling, Nows may or
may not attend any one particular concert.  In the old days,
Nows would have bought season tickets; today, they may just
buy for concerts they like or that fall on convenient days.

  To be precise, Nows are not even the number of people.  I
  made the assumption that, for many of us, the decision to
  buy a ticket isn't independent of others.  If I want to go
  and my spouse, partner, or friend doesn't, I may not go --
  or we may both go; it's more of a joint decision.  Thus
  I'm modeling "decision-making entities", which lets me
  pick the nice, round number of some presumed 100 million
  total such entities in the USA out of some 300 million
  total people.

In the TAFTO model, I have three factors I'm varying: media
advertising (ADS), word-of-mouth advertising (WOM), and
audience retention (RET).  In other words, that model, while
only a bit more complex, does allow one to specify various
average retention rates for Nows, so it could cover what
you're describing at least in an aggregate sense.

According to my testing so far, WOM has the biggest impact
on the number of Nows, with RET in second place and ADS an
almost imperceptible third.

Now this model isn't well calibrated, so it may be wildly
off base.  It also doesn't say ADS aren't useful; I'm
beginning to conjecture that ADS have their primary role as
getting Nows to buy tickets by informing them of the who,
when, where, and what of a concert.

What it does suggest is that orchestras might fruitfully
spend some significant amount of their marketing on
understanding why people leave.

I've attached a really rough draft of some results (not even
the draft of the TAFTO column) with graphs so you can see
the sort of stuff I've gotten so far.  I'm suspicious of
some of the results, as I note in the text, and mindful that
I promised this article to Adaptistration first, so please
don't share these further without checking with me.  Drew
will have a final version on Adaptistration sometime in
April, I believe.

> I think I'll stop here. But thanks so much again, Bill,
> for doing this careful and important work, and for sending
> it to me. I'll be eager to see more.

Greg, thanks for taking the time to think through this and
come up with suggestions for improvement.  I will try to
make some of those changes and let you see what the results
are.

More later,

Bill

For the benefit of those of you who didn't see the start of this series, the "Drew" I mention above is Drew McManus, author of Adaptistration, a blog on orchestra management. And TAFTO? Well, you'll just have to read about TAFTO yourself.

You can read the PDF version I sent Greg with higher-quality graphs. In the interest of speed, I put that document together rather quickly; as a result, you'll see blank spaces in the text where the word processor forced a graph to a new page because it didn't quite fit on the current page. Had this been an official report, I would likely have typeset it for a more professional appearance.

Which do you prefer: this PDF version or the original text version? Which do you think better facilitates dialog?

Making musical sense by email, part 3

At the end of last week, I posted a link to Greg Sandow's thoughts about the future of classical music and an email I sent him describing my reaction to his thoughts.

Here is Greg's initial response, lightly edited and with his permission:

Bill,

Thanks so very much for sending me your work. It's
remarkable stuff, much more solid and promising than most of
what I see. And your tentative conclusion is quite
challenging (at least for the classical music business) -
that the model that best predicts the observed decline in
attendance (am I getting this right?) is the one that
predicts no new people join the audience at all.

I'd make one request. I'd much prefer to see the original
graphics. I did look at your e-mail in Courier, but it would
be a lot clearer if you could attach the charts as graphics,
or perhaps send a spreadsheet with the charts generated
inside it.

Also, I mustn't forget to congratulate you on your careful
work with that Knight Foundation tidbit. I've seen that too
often used to support the idea that all we need is music
education in our schools, and classical music attendance
will go up again. As if, in other words, there's a causal
connection between playing an instrument and classical music
attendance. As you rightly point out, the Knight study makes
no such assertion, but people jump to the conclusion
anyway. And the statement, which I believe they make, that
playing an instrument is a predictor of classical music
attendance does muddy the waters, since it rather casually
puts playing an instrument earlier than attendance in some
sort of implied statistical food chain. That helps people to
jump to the causal interpretation.

Anyhow, I think you debunked this all really nicely. My own
predisposition is to believe that there's no causal
connection, but that instrumental playing and classical
music attendance are both characteristics of some yet to be
defined demographic slice, the demographic in question being
the one most likely to attend classical concerts. In other
words, roughly speaking, playing an instrument and going to
classical concerts have a common cause, rather than one
causing the other.

Going back to the analysis you sent me, I do have some
suggestions for refinements.

First, I believe your model posits that people start
attending concerts, and then continue essentially throughout
the rest of their lives. I don't know if that's true. That
is, people might go occasionally when they're young, then
not go (or not go very often) for many years, and then
resume going, much more often, when they're older. This
would be consistent with some of the data, for instance the
preponderance of the audience in older age groups, and also
survey results that show people most likely to attend
regularly when they no longer have children at home. This
may not have been the pattern in past generations, but it
appears to be the pattern now.

When I made so much of the NEA's statistic about the number
of younger people dropping off, I didn't mean to say that
this would lead to any immediate and directly proportional
decrease in total attendance. I was using the data more
impressionistically. Accepting the apparent truth that
people in this era mostly go to classical concerts when
they're older, I surmised that this younger generation, once
it was 50 years old, would go in smaller numbers than
previous generations did. I should stress that this is
nothing more than a surmise - an assumption. It seems
reasonable, according to common sense, but might turn out
not to be true.

Though when I combine it with certain data about people who
are currently 50 and above, the conclusion seems even more
reasonable. One big change during the past generation is
that older people aren't as committed to high art as they
used to be. They now "consume" (if that's the word) both
high and popular art. This makes them less like the
committed classical music audience of the past, and thus, in
my view, less likely to attend classical performances in the
numbers their predecessors did. Now we have younger people
who go to classical performances far less often than their
own predecessors did. That suggests that when these younger
people are in their 50s, they'll be even more culturally
omnivorous than the present older generation is, and thus
less likely to be committed classical music attenders.

Second (returning to my suggestions for you), you might want
to refine your model to reflect the fact that the total
number of orchestral tickets sold includes many tickets sold
to the same relatively few people - the
subscribers. Currently, subscriptions amount to about 60% of
all tickets sold. This number has declined sharply from the
80% or so reported a couple of decades ago, and is
considered likely to decline still further. Still, this
needs to be an important part of any model that predicts
future ticket sales. It's not enough simply to predict the
number of people likely to buy tickets. We need to know how
many tickets these people are likely to buy, in order to get
some idea of what the total ticket sales in the future are
likely to be. Certainly, in the final analysis, orchestras
are more concerned with the total number of ticket sales
than with the number of individual people attending.

I think I'll stop here. But thanks so much again, Bill, for
doing this careful and important work, and for sending it to
me. I'll be eager to see more.

Best,

Greg

I'm curious: how did that compare with the reaction you thought I would have received? Did your thoughts about what changes you'd like to see me make align with his?

And what did you think of Greg's request to see "original graphics"? (I'll come back to that again later.)

Stay tuned for my response to Greg and to see those requested graphics!

Making musical sense by email, part 2

As I promised, here's the first installment in a dialog between Greg Sandow and me on the future of classical music. You might want to start by reading Greg's essay "The short version"; then read my first email to Greg (below). Drew McManus was included in this dialog, although the emails you'll see all involve only Greg and me.

Greg,

I've been following your thoughts for a while, and I've been
discussing a few ideas they've provoked with Drew McManus,
now that he and I have collaborated on a column
(http://pegasuscom.com/aar/model7.html) and associated
computer model.

I'd like to share those ideas with you and see what you
think; perhaps they'll be of use, or perhaps you'll be able
to educate me and help me refine my thinking.  I've copied
Drew on this, in case he has some comments he'd like to
make.  As you'll note, this email is a bit lengthy; I hope
it's useful to you.

By the way, I'll be showing some text-mode graphics below in
an attempt to make this an easier email to read.  Hopefully
you can view this email using a non-proportional font such
as Courier so that those graphs make sense.  If that's a
problem, let me know, and I'll create this in another
format.

One thing that caught my attention was your claim of aging
audiences.  In
http://www.artsjournal.com/sandow/2006/11/important_data.html,
you note that, among other facts, the average audience age
went from 45 in 1992 to 49 in 2002.  You think about the
potential causes and implications of such a development.

One way to think of such a problem is to see if we can
"operationalize" it: can we generate an operational
description of the events that we conjecture are playing out
in the real world, is that operational description similar
structurally to the real problem, and is that operational
description capable of generating similar behavior?  (See
the link near the end of
http://facilitatedsystems.com/weblog/2007/01/systems-language-for-business.html
for more on operational thinking.)

In simpler and more specific words, can I create a computer
model that captures your hypotheses, and does that model
behave as your data shows?  Success doesn't mean I've proven
anything, but failure might indicate a need for modified
hypotheses.  Enough success, combined with a bit of
triangulation, can strengthen our belief in those
hypotheses.

I like to start with really simple models, adding complexity
only when it becomes necessary.  Often we can learn the most
from those simple models.

Here's a simple model of an "aging chain" that might
represent classical music audiences.

   ..+- .                                                   . .--...m.-
  -+*.+..                +------------------+               ---% +-m++m-
.-.+m*++                 |                  |              . +###+#*#m.-.
-mm#*#%#.       +--+     |  Aging Chain of  |      +--+    --.#+mm%###++
 +.#%#*+. =====>| A|====>+                  +=====>| B|=====>.#*%-##+#m.-
.+%m*+++ .      +--+     |  Concert-Goers   |      +--+    ..m+%###m##..
..%-+% .                 |                  |               . --++m+%+--.
  ..+.                   +--------+---------+                  ..+-..%+ .
                                   \                           .  +-.-
                                    \
                                     \
                                      V

                                   Average
                                     Age

The "aging chain" in the middle is a series of "stocks," one
for each decade of age (arbitrarily twenties through
sixties; I don't think the dynamics change much if we add or
subtract a decade at one end or the other).  People move
through those stocks, taking ten years to move from one to
the other.  I've aggregated those stocks into one mega-stock
to simplify the graphics; if I had drawn the entire picture,
you'd have seen five rectangles, connected in a chain by
flows (pipes), instead of that one bigger rectangle.

The "clouds" at the left and right simply mean we don't care
where those people come from or where they go, at least for
the sake of understanding concert audiences; that's outside
the purview of this model.  There are two flows, shown here
as valves called "A" and "B" on pipes that flow from one
stock (or cloud) to another.  Flow A represents the number
of new concert-goers per year, and flow B represents the
number of people leaving the concert-going world each year.
In this simple model, 50-year-olds don't all of a sudden
decide to become concert-goers, and 30-year-olds don't all
of a sudden give up on classical music.  Those are
constraints we can lift later, if we want to.

Now it's easy to talk about what changes the number of
concert-goers: if A is greater than B, you get more
concert-goers, while if B is greater than A, you get fewer.
If A = B, the aggregate audience size stays static.

I've shown "Average Age" as a statistic we can calculate to
describe concert-goers, the same as we can describe their
total number.

I created a computer simulation of that model.  I set the
initial population of concert-goers to 1 million, spread
evenly in age across the five decades (stocks) I modeled.
If we have 20,000 new concert-goers each year, we'll exactly
replace the number of concert-goers who depart each year,
and the number of concert-goers will remain constant.

Because you hypothesized that young people had stopped
coming to concerts, I tested the model by running it over a
200-year period with 20,000 new concert-goers in the first
10 years and then half that thereafter.  Would that
replicate the data you were quoting?

Here's the graph of the number of new concert-goers per
year:

20000 AAAA-------------+----------------+----------------+---------------++
      +                +                New Concert-Goers per year   A    +
      |                                                                   |
      |                                                                   |
      |                                                                   |
15000 ++                                                                 ++
      |                                                                   |
      |                                                                   |
      |                                                                   |
      |                                                                   |
10000 ++ AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
      |                                                                   |
      |                                                                   |
      |                                                                   |
      |                                                                   |
 5000 ++                                                                 ++
      |                                                                   |
      |                                                                   |
      |                                                                   |
      +                +                +                +                +
    0 ++---------------+----------------+----------------+---------------++
      0                50              100              150              200

                                       Year

It stays constant at 20,000 per year until year 10,
whereupon it drops to 10,000 per year throughout the rest of
the 200 year time horizon of this simulation.

Here's a graph of the total concert-going population from
that model:

1e+006 AAAAA------------+----------------+---------------+---------------++
       +   AAA          +                +               +   Total   A    +
       |     AA                                                           |
       |       AA                                                         |
800000 ++       AAA                                                      ++
       |          AAA                                                     |
       |            AAA                                                   |
       |              AAA                                                 |
600000 ++                AAAA                                            ++
       |                    AAAAAAA                                       |
       |                           AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
       |                                                                  |
400000 ++                                                                ++
       |                                                                  |
       |                                                                  |
       |                                                                  |
200000 ++                                                                ++
       |                                                                  |
       |                                                                  |
       +                +                +               +                +
     0 ++---------------+----------------+---------------+---------------++
       0                50              100             150              200

As you can see, the number of concert-goers stays at 1
million for 10 years and then begins a steady decline,
stabilizing at 500,000 at about year 80.

Here's the average age:

  48 ++---------------+----------------+-----------------+---------------++
     +           AA   +                +               Average Age   A    +
     |          AA AAA                                                    |
47.5 ++        A     AA                                                  ++
     |        AA      AA                                                  |
     |        A        AA                                                 |
     |       A          AA                                                |
  47 ++      A           A                                               ++
     |      A             A                                               |
     |      A              A                                              |
46.5 ++     A              AA                                            ++
     |     A                AA                                            |
     |     A                 AA                                           |
  46 ++    A                  AA                                         ++
     |    A                    AA                                         |
     |    A                      AA                                       |
     |   A                        AA                                      |
45.5 ++  A                         AAA                                   ++
     |                               AAAA                                 |
     +   A            +                + AAAAAA          +                +
  45 AAAA-------------+----------------+------AAAAAAAAAAAAAAAAAAAAAAAAAAAAA
     0                50              100               150              200

Note that the vertical axis starts at 45, not 0.

At first glance, this seems intriguing.  Audience population
is declining and average age increases, at least for a bit.
Then, under these assumptions, average audience age drops
back to the same 45 years old.  Could that be?  Is it
possible that we're just on the front end of a declining
audience size, and audience age will correct itself
naturally?

There's something else a bit off here, though.  In "Where we
stand (2)," you provide a graph that indicates the mode of
age has gone up about a decade from 1992 to 2002, consistent
with your cohort theory, and you note elsewhere that the
average age rose by about 4 years.  This model only shows a
2.8 year increase in average age, and that's over about 30,
not 10, years.

What if the decline in the number of newcomers was a more
gradual and continuous decline?  Here's an
exponentially-declining number of new concert-goers each
year, starting at 20,000 and declining to half that in 69
years:

20000 AA---------------+----------------+----------------+---------------++
      +AA              +                New Concert-Goers per year   A    +
      |  AA                                                               |
      |    AA                                                             |
      |      AAA                                                          |
15000 ++       AAA                                                       ++
      |          AAA                                                      |
      |             AAA                                                   |
      |               AAAA                                                |
      |                  AAAA                                             |
10000 ++                    AAAA                                         ++
      |                         AAAA                                      |
      |                            AAAAA                                  |
      |                                 AAAAA                             |
      |                                     AAAAAAA                       |
 5000 ++                                          AAAAAAAA               ++
      |                                                  AAAAAAAAA        |
      |                                                           AAAAAAAAA
      |                                                                   |
      +                +                +                +                +
    0 ++---------------+----------------+----------------+---------------++
      0                50              100              150              200

Since the exponential doesn't drop as fast, you might expect
the number of total concert-goers to drop more slowly; since
the number of new concert-goers continues to drop forever,
you might expect the total concert-going population to
continue to decline.  You'd be right:

1e+006 AAAAAA-----------+----------------+---------------+---------------++
       +     AAAA       +                +               +   Total   A    +
       |         AAAA                                                     |
       |            AAA                                                   |
800000 ++              AAA                                               ++
       |                 AAAA                                             |
       |                    AAA                                           |
       |                       AAA                                        |
600000 ++                        AAAA                                    ++
       |                            AAAA                                  |
       |                               AAAAA                              |
       |                                   AAAAA                          |
400000 ++                                      AAAAA                     ++
       |                                           AAAAAA                 |
       |                                                AAAAAA            |
       |                                                     AAAAAAAA     |
200000 ++                                                           AAAAAAA
       |                                                                  |
       |                                                                  |
       +                +                +               +                +
     0 ++---------------+----------------+---------------+---------------++
       0                50              100             150              200

Remember that the drop-off starts in the first year in this
experiment, not the tenth year.

What about the average age of concert-goers?  Since the
decline in new, young concert-goers continues forever, you
might expect the age boost to last forever.  Since the
decline is less drastic, you might expect the age boost to
be less drastic.  Let's see:

47.5 ++---------------+----------------+-----------------+---------------++
     +                +                +               Average Age   A    +
     |                                                                    |
     |                                   AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
  47 ++                        AAAAAAAAAA                                ++
     |                     AAAAA                                          |
     |                   AAA                                              |
     |                 AA                                                 |
46.5 ++              AA                                                  ++
     |             AAA                                                    |
     |            AA                                                      |
     |           AA                                                       |
  46 ++         A                                                        ++
     |         A                                                          |
     |        A                                                           |
     |       A                                                            |
45.5 ++    AA                                                            ++
     |    AA                                                              |
     |   AA                                                               |
     + AA             +                +                 +                +
  45 AAA--------------+----------------+-----------------+---------------++
     0                50              100               150              200

You'd be right in both cases, although the difference in the
peak average age is miniscule: 47.0971 vs. 47.7867 years.

So neither explanation seems to account for the drastic
aging of the concert-going audience as reported in the data.

What if we take a drastic approach and cut off new
concert-goers totally after 10 years?  Under those
conditions, here is the total concert-going audience:

1e+006 *****------------+----------------+---------------+---------------++
      +   *            +                +               +   Total ****** +
900000 ++   *                                                            ++
      |     *                                                            |
800000 ++    **                                                          ++
      |      **                                                          |
700000 ++      **                                                        ++
      |        *                                                         |
600000 ++        *                                                       ++
      |          *                                                       |
500000 ++          *                                                     ++
      |            *                                                     |
400000 ++            *                                                   ++
      |              *                                                   |
300000 ++              **                                                ++
      |                **                                                |
200000 ++                ***                                             ++
      |                   ***                                            |
100000 ++                    ****                                        ++
      +                +       *******  +               +                +
    0 ++---------------+-------------*************************************
      0                50              100             150              200

As you can see, the concert-going public drops to nothing
(technically, the model shows fewer than 10,000 people by
year 108); it is already down to 494,703 by year 36 (26
years after young people stopped becoming concert-goers).

What about the average age?

64 ++----------------+----------------+-----------------+----------------++
  +                 +                +               Average Age'*********
62 ++                                           ******************       ++
  |                                   **********                         |
60 ++                           *******                                  ++
  |                        *****                                         |
58 ++                    ****                                            ++
  |                  ***                                                 |
56 ++               **                                                   ++
  |              **                                                      |
54 ++           ***                                                      ++
  |           **                                                         |
52 ++        **                                                          ++
  |        **                                                            |
50 ++      **                                                            ++
  |      **                                                              |
48 ++    **                                                              ++
  |    **                                                                |
46 ++  **                                                                ++
  *****             +                +                 +                 +
44 ++----------------+----------------+-----------------+----------------++
  0                 50              100               150               200

Finally, we're getting drastic changes in average ages.
According to
http://www.artsjournal.com/sandow/2006/11/important_data.html,
the age went from 45 in 1992 to 49 in 2002.  In this model,
it went from 45 in year 10 to 48.75 in year 20.  That's not
a bad match, and the model structure seems reasonable.

What's scary is that's a model of /no/ new concert-goers at
all!  In other words, the data you're showing /could/ be
consistent with a sudden change to essentially no new
audience members forever.  You came close to this same
conclusion in today's "The short version."

Now this model doesn't prove there are no new concert-goers.
There may be other ways to get similar results.  For
example, perhaps it's not true that "once a concert-goer,
always a concert-goer."  Perhaps younger people are starting
to attend concerts and then giving up in droves.  Perhaps
orchestra marketing is drawing in baby boomers who have
never attended concerts.  Perhaps multiple causes are at
work.  Perhaps you have other conjectures.  Any of these
hypotheses could be tested in such a model to see if they
are consistent with the reality you've been observing.

What I think is interesting is that a relatively simple
model can help shed light on the mental models we create to
explain the problems we face.  In this case, the first,
simple approach suggests things may be as you suggest, with
the caution that they /may/ be even more serious than you
indicate.  I'm curious in your thoughts on all this.  I do
apologize for the length of this email; I don't yet know how
to walk someone through a model such as this without taking
a little bit of time.

Drew, did I miss anything fundamental?

I'll be expanding on related ideas using a different model
in a column I'm doing for Drew shortly.  You can see some of
the blog postings I've made about music at
http://preview.tinyurl.com/2consf.  In particular,
http://facilitatedsystems.com/weblog/2006/11/making-sense-with-numbers.html
was a very popular posting about the recent Knight report.
Drew and I have exchanged other thoughts sparked by your
columns, but this note is long enough as it is.

Thank you for your time,

Bill

Think about what your response would have been, had you received such a message. Then come back next week to see Greg's response.

Postscript: When I initially posted this, the fixed-width email section overlapped Blogger's sidebar material. I reformatted the width of the email but left the text-mode graphics in the original size.

Making musical sense by email, part 1

If you've been reading this blog for a while, you probably know I like both system dynamics and classical music. Recently I've been in a conversation with Greg Sandow that involved both subjects, and I thought you might like to read about it:

• Those of you interested in classical music and the classical music business might find the subject matter of direct interest.
  
• Those of you charged with understanding and solving problems (I suppose that's all of you) might find our approach of interest.
  

Greg writes a blog on the future of classical music. He's also a composer, consultant to orchestras, author, and music critic. In addition, he teaches at Julliard and keeps himself busy in other ways.

He's been writing a series of blog essays called "Where we stand," which he's summarized and linked to in "The short version." I found those essays fascinating, for they paint his picture of what the classical music business faces over the next few years, and he offers his reasoning to understand his predictions.

As part of my making sense of his story, I wanted to see if I could "operationalize" his ideas: could I create a model that represented his hypotheses reasonably in both structure and behavior? The process might help me understand them better, it might help me test them, and it might find limits to their application.

Instead of telling you what we did, let me show you. Over the next several days, I'll post lightly edited copies of the emails Greg and I exchanged. That will give you the flavor of what we experienced. At the end, I'll provide my interpretation of what you read. I welcome your contribution to the dialog all along the way. While you're waiting for this to start, I encourage you to explore the links in this posting, for there's a wealth of information to be found.

Calling your shots before you make them

In straight pool, you have to call your shots before you make them.

That's a smart approach for working with system dynamics simulation models, too. Most people showing simulation models to others have likely noticed that you can show a person a simulation model result and often get the response, "Sure, that's what I expected. What's the big deal?" If you ask that same person to "call their shot" (draw a graph of the expected behavior of key variables) before you run the model, though, you and they will often discover they won't have a good record of predicting the outcome. That's not because they are dumb; it's because nonlinear feedback systems of the sort in which we usually live and work exhibit behavior most of us find rather unintuitive.

So do I suggest you do this to make people feel foolish? Not at all. I suggest this to help them (and me) learn. When any of us sees a result and says "What's the big deal?", that person likely hasn't learned from the experience. When we call our shots in advance, using our best insights, and then compare our prediction with the results of a simulation, we often learn one of three things:

• Perhaps our current insights are pretty decent after all, and we can be even more confident in our future predictions.
  
• Perhaps our insights aren't so good, and we can use the discrepancy between our insights and the simulation results to hone our intuition.
  
• Perhaps our simulation model is wrong, and we can use the discrepancy to build a better model of the problem we're facing.
  

There's more to this than just working with simulation models. As Bob Williams and I describe more fully in chapter 10 ("Learning Logs: Structured Journals That Work for Busy People") of Effective Change Management Using Action Research and Action Learning: Concepts, Frameworks, Processes and Applications, there are great benefits to be gained from calling our shots and then comparing those shots with what happens in real life. Done carefully, that becomes an action research approach to getting things done while simultaneously learning how to be more effective in the world we live.

Clark County Adaptive Management Program

Much of the time, we all keep certain details of our work private. Either we or our clients or customers don't want to tell others our secrets to success, lest others learn and take away our competitive advantage, or we don't want to expose our shortcomings, lest others find out which parts of our feet are made of clay.

I recently completed a project for the Clark County Adaptive Management Program. As a public program under the auspices of their Multi-Species Habitat Conservation Program (new URL) (MSHCP) (newer URL), it is subject to the Nevada Open Meetings Act, and thus their work and the work I did with them is public information.

If you're interested in what an early-stage dynamic modeling exercise might look like, take a look at their recently-published 2006 Biennial Adaptive Management Report (new URL) (newer URL). It talks about many things, including the work we did together to achieve three goals:

• Development of a system dynamics model(s) of conservation actions for implementation of the Multiple Species Habitat Conservation Plan.
  
• Use of the system dynamics model(s) to prioritize conservation actions to recommend for funding in the 2006 Biennial Adaptive Management Report.
  
• Use of the system dynamics model(s) to identify key uncertainties and information gaps to be recommended for funding in the 2006 Biennial Adaptive Management Report.
  

This effort was designed both to facilitate conversations among stakeholders who need to understand what the Adaptive Management Science Team is doing and how they make their recommendations to others involved in the MSHCP and to provide the Adaptive Management Science Team a new decision support tool.

Chapter 1 focuses on the model. You can download and explore a copy of the model we created together as Appendix B (new URL) (newer URL). If you don't own a copy of iThink™, you can download the free isee Player from isee Systems. I thank Sue Wainscott, Adaptive Management Coordinator, and the Adaptive Management Science Team for their support in this work. I also thank Ruth Siguenza, CPF, a long-term facilitator for the Clark County Desert Conservation Program (new URL) (newer URL), who first introduced Sue and me.

In many ways, this work resembles what Marjan van den Belt calls "mediated modeling," although I only discovered the term and her book Mediated Modeling: A System Dynamics Approach to Environmental Consensus Building part-way through this engagement.

By the way, if you happen to be in the desert near Clark County, Nevada and see a desert tortoise, please leave it alone; just touching it or picking it up can kill it. The Mojave Max Web site has more information on this amazing creature.

conservation
desert tortoise
endangered species
environment
system dynamics

Business cycles (no, not those cycles)

blink is Malcolm Gladwell's recent book on how we think "without thinking," as he says.

The Logic of Failure is Dietrich Dörner's explanation of the logic behind bad decisions we make.

Gladwell writes about improv groups that can put together seemingly unrehearsed sketches or plays on the spur of the moment. In fact, they rehearse the process intensively so they can react well on the fly. That sounds a bit like business; no matter how much we plan, reality always seems to have little surprises for us.

Dörner shows how our unaided mental processes let us down in certain situations and how the use of computer simulations can help us improve our understanding and ability to make good decisions over a wide range of situations. Simulations should improve our ability to react well when we don't have time for extended thinking.

Two books, similar story. There's a time for action, often without much time for reflection. There's a time for critical reflection, that period when we review what has happened and plan for the future.

The ability to practice can help in the critical reflection phase. Sometimes we can practice with others; actors and speakers do that. Sometimes we may find simulations effective. How often do we in business do that?