To The Modelers Of The World

I wrote this as a comment in this thread, decided to post it myself.

————————

Models are not the thing modeled. It is a category error to equate them.

The future is forever closed to us, every single thing we think we know about the world 10 minutes from now could be changed in such a myriad of ways. And are being changed*, we just aren’t paying attention.

It simply astonishes me to see people claim to know the future. If they did, they would be making $ from that fact. To do so after knowing the history of such claims is Hubris To The Power Of Dunning-Kruger.  But more likely they were intentionally ignorant, limiting their view of the world to a pinhole so that their truths are clear, the limits imposed by their field of vision are easy to ignore, as well as the swirling currents outside it.

You can model components of our reality to a level of precision that allows an edge in making decisions about things in the future, with different probabilities for classes of events. High probabilities of useful precision in airframes and electronics and many areas of science and technology, given a lot of experimenting.

You cannot ignore that last ‘given a lot of experimenting’.  Very few new things work the first time, not even minor modifications of old things.  I think there are few scientists who get their experimental setups working the first time, the proper statement of they hypothesis correct the first time. In my world, the best teams turn their circuit boards the fastest, their EE designs the most often with the best test set, and their processing setups algorithms and tools are tested and changed continuously.  The progress I see everywhere is fastest where the philosophy is ‘fail fast and often and systematically and learn the maximum from each failure’.  Fast compiles changed how software engineering is done, a hardworking engineer can now make many 1000s of errors in a day, as good a measure of progress as any other in thoughts being tested.

But that progress requires a hard standard to measure against, a way to know what our efforts have achieved and measurements to correlate different aspects.  We can do that for hard science and technology.  But above the level of hard science, I don’t think any government can claim any measure of cost-effectiveness at anything, the scene through history is people making ever-bigger bets and losing ever-more devastatingly. The progression seems so clear to me, the analyses I see agree, and if any gov-equivalent entity had such evidence, we would all know it, every tiny bit of evidence supporting any government program is replayed often in things I read.

You cannot allow this ‘we can predict the future’ kind of thinking in government, because it is the basis for the Progressives of both Left and Right in their programs to improve the world with welfare programs and war, respectively. Neither have worked, and they can’t work for the same reason that climate models can’t do more than tell us what our current understandings say about what might be, with the discussion of error bars including the many predictions in the past contradicting each other in every detail and how that is normal as scientific understanding progress, so don’t believe this one either.

Nobody would allow a foreign policy if the full explanation of past decisions and outcomes were explained, nobody would allow a federal poverty program seeing the wasted money and lives that resulted, and nobody will allow major costs based on this class of predictions of a far future. But the people on the payroll are all big defenders of each one, and most anxious for me to accept their model of reality and its prediction of the future.

The same nature of nature denies them all. I am a complete agnostic as to what will be because I do not think that it is possible to ever know enough or compute well enough based on initial conditions fine enough to have more than ‘things my models suggested’ knowledge. Maybe useful, but only the actual, real, happening-in-the-future future will allow us to know. Until then, it is the results of a model. One of many.

One of many different predictions that it made itself, over its development, in fact.  That development was not correct, at first, just like hard-science projects are rarely correct, first time.  The difference is that reality tells us when we get the hard-science projects right, but the future is not a thing we can measure against, so there is no hard standard for the model, and we can expect no hard information to be produced from it.

So risks to be covered, maybe, and assessments of risk and what must be protected to what extent will vary. OK, that complexity the world can deal with, just another risk to consider and be hedged.

But you gain no confidence in me by claiming you know the future. That is always BS.

*One of the reasons power points are lousy is that they do not convey the dynamic system information you normally need.  That is assumed by the author-speaker, and much of the difficulty of teaching is in getting those dynamic models into people’s head.  Those are a hidden communication channel between experts, and simultaneously prominent in the lists of the reasons we outsiders “don’t get it” and why they so often get it wrong, as dynamic models are hard to construct and hard to evaluate, it isn’t something you do via idle contemplation.

 

Advertisements

3 thoughts on “To The Modelers Of The World

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s