The title is obviously the wrong way around, exchanges turn distributed logs into order books. The distributed part is a resilience decision but not essential to the design (technically writing to a disk would give persistence with less ability to recover, or with some potential gaps in the case of failure (remember there is a sequence published on the other end too, the market data feed)). As noted in the article, the sequencer is a single-threaded, not parallelisable process. Distribution is just a configuration of that single threaded path. Parallelisation is feasible to some extent by sharding across order books themselves (dependencies between books may complicate this).

It would not surprise me at all if the sequencing step was done via FPGA processing many network inputs at line rate with a shared monotonic clock. This would give it some amount of parallelism.

good point, sequencing is very minimal, therefore some parallelism is feasible that way, but the pipeline is not that deep, at least ideally. Of course if people are chasing nano-seconds, it may make sense.

These inefficiencies are akin to having some “wrong” weights in a huge model. Corporations also average over their individual contributions, positive or negative. And negative feedback loops may be individually detrimental but collectively optimising.

It might involve US sending out trillions worth of stuff. The question is what stuff would that be, could be e.g. bombs.

The point is for it to be hard.

Borderline impossible! It's a much wider gap than modern English to Shakespearean English is.

There’s got to be a chronological point where older Greek gets much easier, though. Or at least, there was back when Greeks still had to work with katharevousa in school. I learned both Ancient Greek and then, though a year spent in Greece, Modern Greek. When I pick up late Hellenistic literature like Achilles Tatius, I feel like it is activating more the Modern Greek part of my brain than the 5th-century BC Athens part of my brain.

He was also working for Valve for a while.

Yes that’s why he and Gabe were speakers at the talk. In game economies were the topic.

And CCP (EVE Online)

A positive outcome of LLMs. Regardless if the specific article is AI generated or not, we become increasingly intolerant of shallowness. While in the past we would engage with the token effort of the source, we now draw conclusions and avoid the engagement much faster. I am expecting the quality of real articles to improve to avoid the more sensitive reader filters.

I now notice myself cringe internally whenever I say anything that has become a ChatGPT-ism, even if it's something I always used to say.

I used to write very formally and neutrally, and now I don't, because it comes across as LLM-ish. My sentences used to lack "humanity", so to speak. :(

Autistic here, and passionate about people’s experience.

I assume that’s why the original argument is that it’s not been a net positive. I.e. the assumption is that lots of work can be good and necessary, while even more that is evil and excessive can end up with a net negative.

I think you partially answer to yourself though. Is the value in the depreciating chips, or in the huge datacenters, with cooling, energy supply, at such scale etc. ?

Single point of failure?

A lot of old telephony systems were full of SPoFs.

A core design principle of the Internet was the ability to automatically route around damage. The requirement came from a desire to be robust against nuclear attack. Very few 20th century networking systems could do this. Star topology or ring topology all had intrinsic SPoF.