[<< | Prev | Index | Next | >>]

Friday, May 09, 2008

Foresight in Hindsight

"People never remember what you say, and they never remember what you do, but they always remember how you make them feel." -Maya Angelou

Simon's corollary: When you disagree with someone about something and are proven right much later, they will not remember that you were right, they will remember that you were wrong.

I've been discouraged recently to find myself in the same debates from years back all over again, just with new items in place of the old. Enough evidence is in to resolve many of these old debates, but it turns out people don't remember that far back, at least not accurately, and as a consequence they are incapable of connecting how the world looked to them back then with what was actually true back then, and then relating this to how the world looks to them now, and, imo, how they are making the same mistakes now they made then.

Back in early 2006 when people were still quite gung-ho about the housing market, I was very much not. (In fact, I got quite a lot of flak--see above quote about being remembered for how you make people feel.) I was called out to put hard numbers on my cynicism, so I did:

Date: Fri, 24 Mar 2006 14:50:57 -0500
From: Simon Funk <simonfunk@gmail.com>
To: (various)
Subject: Re: down for the count...
Someone said (on Mar 24):
> Care to translate your gleeful doomsaying into an objective prediction of
> selected home values, perhaps based on easy-to-access zillow figures, over
> some time span?

Well, if one just extrapolates from this:

We could see a 50% decline over the next seven years (in SD, relative to incomes). If SD incomes continue to grow at the rate they were between 97 and 2000 (which is the data I happened to find just now), they'd increase by around 50% during that period, which means, based on these particular extrapolations, nominal house prices would drop 25% from where they are now. (Of course, this still represents a real drop of closer to 40%, masked by general inflation.)

However, considering most of SD's job growth in the last few years has been real estate and construction related... it seems likely the above estimates are optimistic. Also, the magnitude of the deviation is unprecedented; unclear whether it will take longer to correct as a result, or whether it will fall even faster than in the past due to a much larger % of forced sales (foreclosures, etc.) combined with a bigger panic and more publicized collapse. I'm expecting the latter-- it takes work to hold up a bubble, and this one's stretched further than ever, so when it starts imploding people are going to run for the hills rather than remaining optimistic. Also note that in the past there was a lot of room to lower interest rates during the declining cycles, which stretched them out. Not so much room for that now. (I seem to recall my home mortgage in 89, just before the last major burst, was around 11%!)

I would not be surprised to see some properties, e.g., some downtown condos, drop 50% in nominal value. I would be surprised if the SD average didn't drop at least 25%.


Here's the market today:

In my general dismay over people's lack of seeing a number of such patterns that have been unveiled to varying degrees in the last year or two, the housing market being the most clear and obvious, I asked if at least they remembered my above response. The person who asked me the question himself, to whom the above was primarily directed, replied that he didn't even recall me answering! (Somebody told me recently if you don't say "I told you so" people will not remember that you told them so. Now is her turn to say it to me.)

I used to believe that people would learn from these empirical challenges to their world models, but no longer. I am ever more enamored with Generational Dynamics which says that people today think the way they do because they can't think outside the illusions of the short time spans of their lives. Real estate can't go down because it never has in our lifetimes. Apply the same thinking to the financial crisis, to peak oil, to food and water shortages, to the national debt, to war, to Nazi America. The irony is that it's the optimism itself that creates these problems. And a little cynicism injected into the system unfortunately just gives the optimism more time to grow stronger. I had hoped that lessons learned in one domain would translate to others, but I'm not seeing the slightest indication of this. If anything, it is working the other way around, and what should be lessons to learn from continue instead to be cast into the everlasting light of optimism, and simply never seen for what they are. They move too slowly to be seen, and the human memory, as I am realizing, is a lousy anchor.

A related thought in a tangential email:

Orwell was a genius. I think one of his most astute observations is about people's tendency to remember the past as being basically like the present.


I just this moment solved a long standing (personal) problem with AI, and as usual the problem was that I was anthropomorphizing humans. Knowing that our perceptions, and hence memories, are built upon our hierarchical model of the world, I've long wondered how is it possible that we can remember anything accurately in light of bits of that model drifting with experience. The answer, of course(!), is that we don't! We remember a snapshot at some level of abstraction, and that memory is probably stable at that level of abstraction, but the connection between that level of abstraction and the rest of the world--in other words, what a given "statement" at that level of abstraction expands to in real terms--changes as we learn (and as our brain adapts our model to a changing environment). Hence, when we invoke the memory, the pattern it reconstructs--which may have been reliably accurate at the time--is now based on today's assumptions instead of yesterday's and thus potentially reconstructs something completely different, and yet the machinery of our mind gives it to us labeled as a slice of past reality.


[<< | Prev | Index | Next | >>]

Simon Funk / simonfunk@gmail.com