A real-world high stakes game of experimentation with externalities:

Last night at 10, my car was at the front of several miles of cars on the Garden State Parkway all stuck behind a segment of road 3-5 feet underwater. You could try to drive through if you wanted,
but most people were dissuaded by the half dozen stalled/flooded cars in the water.

For about three hours, one vehicle every 15 minutes or so would go for it. Whether it succeeded depended on its type, the path it took, and the water depth.
The interesting thing is that a failed experiment (trying and having your car stall in 3 feet of water) has considerable private cost: deeply flooded cars are totaled, and the cost of even a lucky recovery in such a case is more than a few thousand dollars.
The lone state trooper there did not project any artificial confidence about this situation; he said up front that they had never seen a major highway flood like this.

But he also didn't discourage experiments. He sort of gave advice and watched with interest what happened.
And so did the few hundred drivers with a view of it, the braver ones gauging their own strategies and prospects. A big F-150 making it through was followed by a few others, whereas small SUV's failing dissuaded similar attempts.
One more interesting thing is that when the water began going down visibly, experiments became, if anything, less frequent, since it was pretty clear you could go soon at low risk.
@marketsensei points out that drivers may be too eager to experiment if insurance covers the loss. On the other hand, the classic theory of experimentation with externalities suggests that there were too few experiments!
One last, psychological note. The jam started at 9:45 or so and ended four hours later. That time flew by weirdly fast - the constant small news was absorbing.

Overall very lucky - could have easily been much worse. Take flood warnings seriously!

n/n

• • •

Missing some Tweet in this thread? You can try to force a refresh
 

Keep Current with Ben Golub

Ben Golub Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @ben_golub

29 Aug
Sometimes faculty complain about the stubborn Ph.D. student, who seems unaffected by advice. Talent and energy are risk factors for this disease, and, worse, is closely related to personality traits of many successful academics.

A few random thoughts.

1/
What "bad stubborn" looks like from the advisor perspective is that you thoughtfully engage with the work, repeatedly say something (that you feel is) REALLY IMPORTANT that should affect the project, and perceive it not to be affecting the project or the student's thought.

2/
A friend wishes they could tell students one cheat code for success. When faculty say, "This seems like a question you can answer in your project and people would really care about the answer," *actually try to do that*, or at least have serious conversations about it.

3/
Read 11 tweets
14 Jun
An applied mathematician I know thinks it's hilarious that economists care about formal rigor so much more than, e.g., applied physicists do.

Rigor, he says, is valuable, but other inputs currently seem to have a much higher return for advancing economic theory.

1/
For example, if our theorizing about long-run outcomes of social learning falls short of our potential, it's not because we forgot to check a subtle condition in applying the martingale convergence theorem in our model of their Bayesian behavior.

2/
(their = the agents').

"His people" (applied mathematicians, applied physicists) would not worry about that. Instead, they would quickly work through much more "theory," but without great rigor, and use the results to refine the collective decision about how to continue.

3/
Read 10 tweets
10 Jun
A few simple facts that some people find surprising the first time they hear them.

Imagine $100 is behind door A or B and I give you independent hints about which. The hint says either A or B but is right only 55% of the time.

First hint is worth $5, second hint is worth... $0!
Why? Because the second hint never makes you *want* to change your decision. (Think about the four possible hint combinations.)

This is a key idea behind a beautiful paper by Meg Meyer, here:

2/
If you want the second hint to be useful, you need to make it biased, "favoring" the leading option, so that if it comes back a surprising negative against the leader, you might actually change your decision.

Meyer uses this to derive implications about organizations.

3/
Read 7 tweets
28 Apr
I've been playing around with a virtual talk format that's different from traditional slides, which deals with my biggest complaint about slides: lack of persistence of information.

Almost always, I want to see "setup" again during the first result/example, but it's gone.

1/ Image
In the format here, each panel is basically a slide, and I reveal these panels one by one.

That might be too small on a projector screen. But when everyone is in front of a monitor anyway, this seems to be better for giving the "lookback opportunities" I would want.

2/
Overall, I've never learned/understood better than during college math classes where professors slowly filled up six nice sliding boards.

In an ideal world, we could have this but with prepared content.

Probably not happening soon, but virtual talks allow an approximation!
3/3
Read 8 tweets
24 Apr
David Blackwell would be turning 102 today.

He's best known for the Blackwell information ordering, the way to formalize when some signals give you more information than other signals.

A thread on Blackwell's lovely theorem and a simple proof you might not have seen.

1/
Blackwell was interested in how a rational decision-maker uses information to make decisions, in a very general sense. Here's a standard formalization of a single-agent decision and an information structure.

2/
One way to formalize that one info structure, φ, dominates another, φ', is that ANY decision-maker, no matter what their actions A and payoffs u, prefers to have the better information structure.

While φ seems clearly better, is it definitely MORE information?

3/
Read 18 tweets
8 Apr
Our perceptions of some of the things we experience are deeply inaccurate. 🧵

Case 1: The vast majority of restaurants get few visits and go out of business quickly. This seems surprising because the typical restaurant you experience is busy and long-lived.

1/
The gap between reality and perception happens because few people experience any given unpopular, short-lived restaurant. Precisely because it is unpopular and short-lived!

The brilliant @CFCamerer, who gave this example, notes that it's not just curious but consequential.

2/
We, including aspiring restauranteurs, undersample unsuccessful restaurants so badly that it can make the restaurant business intuitively feel easy.

So too many people start restaurants who should have done other things instead.

3/
Read 15 tweets

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Too expensive? Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal Become our Patreon

Thank you for your support!

Follow Us on Twitter!

:(