Another accidental advantage

I found David Cheng’s thoughts about his recent China trip fascinating — textured and surprising.

These two linked assess­ment are notable. First:

The Chi­nese tech ecosystem is closed, for two com­pounding rea­sons. First, B2B SaaS never worked in China, so there is very little shared sur­face for work in the first place. Second, every com­pany com­petes with every other com­pany, and there is a real fear that if any com­pany touches another’s data it will be “eaten.” The con­se­quence is that there is nothing for an agent to prac­tice on.

And then, the other side of the coin:

This mat­ters to anyone building here: the Amer­ican inter­op­er­ability we treat as a con­ve­nience is actu­ally [an AI] training asset. Open APIs made inte­gra­tions easier; they also create the envi­ron­ment for agents to learn in. That is not a gap that closes by spending more on post-training. It closes only if the under­lying ecosystem opens, and the under­lying ecosystem is closed for rea­sons that are com­pet­i­tive and struc­tural rather than technical.

(Emphasis his.)

I think often about the con­tin­gent devel­op­ments that made AI agents of the LLM variety pos­sible at all. Among them: the devel­op­ment of an open web, which is to say a scra­pable web. The internet absolutely didn’t have to be this way — it could easily have been a giant AOL, a mas­sive Minitel. In many ways, those shapes would have been more con­sis­tent with tech history. Also: the suc­cess of open source, which larded that open web with huge piles of code to study. Again, this wasn’t destiny, and as recently as the 1990s would have seemed improbable.

Based on David’s note, I might now add a third lucky stroke: the rel­a­tive ubiq­uity of open APIs. Really, what it means is, “Web 2.0” came home to roost at last — its ulti­mate value totally unfore­seen in the 2000s, bigger than anyone could have imagined.