Comment by btown
1 year ago
I have a slightly different view on this. ZIRP allowed companies to do crazy things when it came to tech investment. For instance, would Kafka (and by extension, Confluent, and much of the distributed-systems landscape) exist if LinkedIn hadn't decided: "you know, let's not think too hard about whether building a custom queue system is fully economically rational, let's just let our engineers do it, and we'll be able to spin it to investors/shareholders as a good use of capital." It's no coincidence that this happened in 2010, the first year that the interest rate dropped below 1% (and substantially so).
https://www.rtinsights.com/the-technical-evolution-of-apache...
https://fred.stlouisfed.org/series/FEDFUNDS
Were the people who worked on that project value-optimal for LinkedIn? Maybe, maybe not. Were they "lazy" and a "waste" in the value they brought to society overall, especially including the spin-offs and startups that members of these teams would go on to build? I don't think that's an accurate characterization.
Zero-interest was an effective way to incentivize public companies, VC-backed companies, and their LPs to spend large amounts of money on speculative engineering work - speculative enough that one might even call it research, but the fact that we could call it something other than research is the entire point.
It's far from the only such incentive, and it came with many, many downsides, including the creation of many over-funded startups that over-promised and caused harm in their under-delivery.
To be sure, startups aren't going anywhere. There are surprisingly fun challenges when profitability and sustainable growth are core requirements for every project, both macro and micro, and it's just an evolution of our hacker mentality to have to deal with these new constraints.
But I fear that we've lost the incentive to have companies invest speculatively in crazy projects without having a sufficient replacement for ZIRP, and it may take a generation before we understand how that will have affected the rate of innovative output for the entire world.
It's a fair perspective, but I'd counter by bringing up the opportunity cost we spent on the incredible amount of trash startups that ZIRP produced. Those Juicero & crypto & NFT people could've been doing something productive with their lives instead of whatever that mess was. Was Kafka worth that cost? I dunno. But seeing the obvious waste and the clowns who profited from it made me, personally, really depressed at the state of our industry. A system that so blatantly does not distribute rewards for real, valuable work really discourages one from bothering to even try putting in the work.
It kinda feels like raising the Temperature of a LLM too high, if you'll forgive the analogy. Yes, there might be occasional-seemingly brilliant-bursts of creativity but most of the output is just going to be nonsense.
I suspect that Kafka paid off for linkedin long before the authors left to start a Kafka company.
RabbitMQ et al were so hard to scale/operate at volume.