[jandrewrogers]

It isn't just about AI. Some R&D domains started disappearing from literature and the public internet a decade before the first LLMs. The incentives to go dark emerged even when the adversary was other humans. AI is just accelerating a trend that was already there. Some areas of frontier computer science research have largely been dark for decades.

The strategy is to quietly do several years of iterated hardcore R&D. The cumulative advances are such a step change when seen by would-be fast-followers that it obscures the insights that allowed individual advances to occur. As an exaggerated case, imagine if the public history of powered flight skipped from the Wright Brothers to the Boeing 737.

In practice, this strategy has a major failure mode that people overlook. The sharp discontinuity in capability means that almost nothing that exists in the market is prepared to integrate with it. This is a large impediment to adoption even if the technology is objectively incredible and the market will inevitably get on board.

In short, it looks a lot like being too early to market. This is surmountable with clever execution but with this strategy you've traded one problem for a different one.

[rtrgrd]

Also similar to block withholding [1]

[1] https://bitcoin.stackexchange.com/questions/4943/what-is-a-b...

[sigbottle]

Interesting, any examples of companies that followed this model?

[bitmasher9]

You get a time advantage for doing this strategy, but your talent will be pouched and your competitors will be able to catch up fairly quickly.

[jandrewrogers]

I used to think this but it only seems to be true for a shallow tech advantage, which isn’t this scenario. A sufficiently deep stack of compounded tech is robust against even aggressive talent poaching. The knowledge is embedded in the network, not the random individual.

We see this in jet engines, silicon fab, et al.

[bitmasher9]

With these very deep tech stacks, does it really matter if you publish or not? Execution is still very hard for manufacturing these items, and will be for awhile.

We’re very very far from prompting to a silicon fab

[asdff]

I mean, even north korea has figured out the nuclear bomb, the original greatest secret deep stack of compounded tech. Seems like anyone can figure out anything if they are hell bent on it on this earth. Engineers seem to be more fungible than people anticipate I guess, and no one really comes up with unprecedented unique ideas. The whole research process incentivizes incremental work on known concepts to justify receiving funding at all, since it is in high demand and short supply.

[adrianN]

Korea had the advantage of like seventy years of technological advancement from the first nuclear bomb.

[asdff]

Soviets figured it out in a couple years after we did, very much planning to keep it out of the soviets hands the whole time.