The rational decision, when you know you cannot control how others will act, is to minimize the pain. If you saw the HN thread on Thursday, there were founders posting that they were told it will be fine. How do you think they were feeling on Friday?
And it is not a personal decision, continuity of business is a pretty big deal. Nobody in the company would forgive you for missing their payroll, because you thought it was the ethical thing to do.
But part of what makes that whole bit of game theory interesting is that it has a lot more complexity then just the simple version. Yes in a plain prisoner's dilemma betray can make sense, but in an iterated prisoner's dilemma that's no longer true. That's been part of the debate around the whole debacle, for much of SV's history it was very much iterated, it wasn't just about any one company for VCs or even founders, there was some interest in the overall ecosystem. Some very successful people only found success after multiple failures, but success could pay for it all in the end. VCs too, maintaining the ecosystem mattered. In that sense, killing the goose laying the golden eggs, even if there is some short term payoff, is still clearly a poor idea.
There was an HN piece sometime in the last few days I think talking about this and speculating/arguing that the rise of unicorns changed the community from iterated towards the more singular game. If it only took a single megahit to cash out forever, then incentives might be more towards selfishness vs thinking about the next company and the next after that.
>If you saw the HN thread on Thursday, there were founders posting that they were told it will be fine. How do you think they were feeling on Friday?
Sure, but how do you think they were feeling Monday? Or now? Zero threat to deposits after all for anyone including those who sat tight, but they lost a valuable service and icon, invited other problems and scrutiny, etc. While I'm sure lots will confidently say they did the right thing and that purely looking out for #1 was "responsible" I suspect there are at least a few who have some real regrets, even if SVB absolutely made very stupid choices as well.
He first made this argument in 2017 in an article about Uber: https://stratechery.com/2017/the-uber-dilemma/
https://stratechery.com/2023/the-death-of-silicon-valley-ban...
I’m not in the field but I’ve never heard a theory that a generous strategy works in a game between an arbitrary number of (mostly) anonymous players.
And to call this event part of an iterated game at all seems… unusual.
If you are a founder who withdrew all your company’s money, then your reputation is neutral or perhaps positive. You saved your company. You did “whatever it takes”. These are good qualities in a founder.
If you are a founder who left your company’s money in SVB, nobody is going to give you a medal for that. Instead it exposes you to questions such as “Why did you have so much cash in one place?”. If depositors hadn’t gotten bailed out then you also face the collapse of your startup, your coworkers losing their jobs, etc. So this is definitely a much more negative outcome iterated or not.
I'm not sure they deserve quite so much vitriol as that website directs at them, but it really does seem to be a lose/lose situation for all the people who claimed to be fans of that bank. No one seems to be better off.
And while it's not hard to explain this outcome as 'the rational thing', it's also easy to imagine a different outcome where some more courageous leaders got together and held things together.
>I don’t understand how the iterated prisoner’s dilemma makes a difference.
SVB isn't fungible, this isn't the case of "a bank is a bank", it was heavily used for a reason and that was it had developed specialized services, expertise, and culture around supporting startups. Losing it may mean that SV as a whole is permanently, or at least temporarily but significantly, impoverished. New startups may become harder to start and operate. This doesn't mean the end for existing businesses, certainly not already grown successful ones with stable income and sufficient size for in-house expertise, so in a singular game it's not a big deal. But to the extent VCs or founders care about the next game and the one after, the startups to come that don't yet exist even on napkins, they may now be worse off than if through cooperation SVB had been saved and reformed.
You're missing the entire point of iteration, that it's not just about this one game. Startups fail for all sorts of reasons, what has made SV successful long term wasn't the success or failure of any single startup but that there was an overall environment where it was straight forward to have new ones try again and again.
tobyjsullivan sibling wrote:
>Any research recommending alternative strategies for iterated games is almost certainly looking at repeating games between a consistent pair of players.
Historically though SV has indeed been relatively consistent players, from VCs to founders to employees to, well, services like SVB! Silicon Valley Bank was founded in 1983, not yesterday. What is that to Silicon Valley itself if not consistent?
The iterated game here is all the startups after the current batch. Where do VCs and founders go to and deal with in Fall 2023? In March 2024? Will it be exactly and perfectly as good as SVB, or better? If so then sure, no problem. If not, they've sacrificed advantage in future iterations for the short term.
RyanGWU82 wrote:
>The Prisoners' Dilemma analogy was covered in Ben Thompson's Stratechery article this week
Thank you! There it is, glad I can read it now. Much better than anything I wrote. And to add on for the future here is the HN discussion on it:
It is, but also it's even worse. The usual prisoner's dilemma is explained with 2 parties. In this case, it was thousands of parties with a choice of: a) do what's best for your business, b) bet that almost all the other parties with also choose b.
2-person prisoner's dilemma is a fun game theory exercise. "All the customers" prisoner's dilemma is a social "have you met other business people" exercise.
In this case, starting from the perspective I have a choice (and thus the bank is still liquid), I can only do at best the same as I would have done if I pulled out.
From their article:
https://www.google.com/amp/s/www.circle.com/blog/an-update-o...
> We have reason to believe that under applicable FDIC policy, transfers initiated prior to a bank entering receivership would have otherwise been processed normally.
In other words it shouldn't be entirely dependent on whether they were in the bank run. Earlier transfers, yeah.
In a bank run situation people were rushing to the bank and tying up the phone lines. Panicking shouldn't be encouraged to such a degree as to make millions of dollars depend on it.
The amount recoverable ended up not depending on whether an SVB customer attempted a late withdrawal. And that's good. I would have expected it even if depositors had to settle for a portion of their funds.
What does Friday matter? Friday was when the cynics would rejoice in righteous indignation. Monday was when responsible people were redeemed; when society through its government recognized that the real moral dilemma here was whether or not to vindicate the cynics.
There is a solution to the Prisoner's Dilemma, and it's called civilization, known in the modern era as institutional government. Like their communist cousins, die-hard libertarians should be forced to live in truly libertarian countries (i.e. those with extremely weak governmental institutions where wealth is safeguarded primarily through a web of private arrangements) and only then be allowed to opine on what's rational. For some reason libertarians can't often be found in those countries; not by choice, at least; and certainly they don't park their wealth there.
I mean, it was fine for them by Monday.
