I have long espoused the works of Gwern Branson, and his essay Choosing Software is no exception. If I were to sum up the vibe of it, it would be: It is not enough for software to be good - it must also be lindy. There isn’t much point in eating the learning curve for, say, Emacs if Emacs isn’t going to be around long enough to see a good ROI on it.
But how can we actually know that a given piece of software X is going to be around, much less actively maintained and updated, say, 10 years from now? Certainly there are things software developers can do to make this easier; but outside of the FOSS community, which I’m sorry to say is a very small portion of the software that has actually actively eaten the world, most financial incentives seem to run orthogonal-at-best and counter-at-worst to this. Don’t shoot the messenger here: I have a long, long history of thinking about the topic of how to make the software I write, use and love usable for the long term, in every sense of the word. Even for open source stuff it’s an inherently toilsome problem. What hope is there for software I can’t even compile from scratch?
Every software industry appears to have converged upon some subtly different norms for hedging against this kind of thing, and it is fascinating to see all the different ways this work. I have spent my whole software career working in various highly-regulated industries, first software for doctors, then software for people who manage cold hard dinero, and currently software for trains. They all have their own ways of ensuring that, in case of emergency, their customers can still make use of the bits they hawk and schlep. But focusing too much on e.g. how cool those “dead man’s switch” escrow companies are where the whole job is being really good at reviving ancient software in case one of their clients goes kaput seems misaimed to me, for the simple reason that I have never worked at a company which has actually had to use any of those mechanisms. I have not yet worked for any company which has gone bankrupt. Indeed, I have never worked for any company which has not already existed for at least 30 years. Generalizing from my experience would be like trying to find life lessons from a randomized sample of elite ultramarathoners! Survivor bias, ahoy!
No… What I’d really like is some kind of independent market, whose whole bag is taking an unbiased view at the software provided by Company X, or perhaps of FOSS Project Y, and which can make even a rough claim, like: We estimate an 80 percent chance Emacs will continue to have at least one active maintainer in the year 2035. We estimate a 5 percent chance Widgets Inc.’s Widgetr application shuts down with no data egress possible for its consumers by the summer solstice. Et cetera.
You know what this sounds to me like? This sounds to me like what traders do.
Financial trading is one of those things that is much, much more interesting than you would ever suspect based on a dry description of what it actually is. At the highest levels - the levels pretty much anyone who actually makes money trading has to be at, because of how ridiculously competitive it is - trading is all about coming up with stories as to why a certain thing is more or less likely to happen than other people think it is. This surprises many people who get a whiff of Wall Street and turn their noses away in disgust at the prospect.
But I’m aware of at least one person who, back in February 2020, woke up one day to realize that, oh shit, COVID-19 is probably going to mess everything up for a while. And they had enough knowledge of financial trading to realize: They could put their money to the test and bet on looming disaster. Three months later they were able to walk away with $65 for every dollar they bet because they were right.
This may seem immoral to you. Consider a world where, say, Goldman Sachs made the same kind of bet: 100% of their portfolio against COVID-19 crashing literally the entire world economy. What would happen? Well, a lot of similarly very smart people would realize this means that they could take all of Goldman Sachs’s money for the paltry sum of averting this horrible catastrophe, and so we probably would have seen much, much more pressure to ramp up mask production, roll out the Moderna vaccine ASAP, etc. If they succeeded, then Goldman Sachs would have been called a laughingstock for betting and losing the house like that, and the COVID-19 vaccine would not have systematically impoverished the entire world for many orders of magnitude more dollars of loss. I won’t belabor the point here, though: My goal is to simply convince you that, yes, there are in fact ways to make money off of saying “This bad thing is going to happen”, because you can bet against all of the people who are saying “This bad thing is not going to happen.”
So it could be in the world of software as well. In fact a specialized kind of financial market, known aptly as a prediction market, already exists to do this very thing! Manifold Markets is my personal favorite example, primarily because I myself suck at it, which is a really good sign that actually smart people are already using this platform to actually make pretend Monopoly money. People already use sites like this to bet on how much money Bitcoin will be worth at a given date and whether the Indiana Pacers will emerge from behind to take the NBA this year - there’s no reason to believe we can’t get useful information out of polls like “Will OpenAI continue to offer API endpoints for gpt-4o-mini-2024-08-13 by the end of 2030”. The fact that the question requires specialized knowledge only means a well informed trader has more to gain by collecting the evidence, weighing it all skillfully, and betting on their beliefs.
I suspect prediction markets like this might be a kind of “free lunch” for the software ecosystem as a whole, counterintuitively making it both more stable and more dynamic. If the prediction market bellweathers forecast, for example, that e.g. Google Maps will be dishonourably discharged by 2026, that is a really big deal for a lot of people, software developers and ordinary users alike, and it could spur people to start migrating their apps to use a different GPS service even ahead of any official announcement from Google themselves, if it’s important enough to them. Or someone could see this and say to themselves, “It’s now time to build”, and create the next latest and greatest Google Maps, hopefully with an actually sane revenue model this time.
And, in its own way, that decentralized information-gathering process which we call capitalism might just wrap back around to make the chore of deciding whether to learn Vim or Emacs or some secret third thing that little bit easier. “Emacs has the edge for now, but the actuaries say all of its maintainers are pushing 60 with no replacments. Sub-90% chance it will be maintained by 2225. Team Vim all the way”, etc.