On Discoverability and the Internet

Some thoughts on where energy might most usefully be expended to make the internet ‘better’. It seems likely that I might write some more in this vein sometime, so consider this Part 1 of n on Discoverability

The chief problem with the internet is (I’ve long thought) the problem of discoverability. And it’s one of the hardest things to solve. It’s also worth noting that, although it is almost never explicitly spelled out, the money thinks the same — and the money is just following where the masses lead. The most valued internet sites — Google, Facebook, Twitter, Reddit — are all to do with discoverability one way or another. And on the whole they do a pretty piss-poor job of it. I think that anything that does discoverability better will be worth a whole pile of money. Even better, it might actually serve humanity better if it avoids the obvious and toxic biases that inhere in existing platforms, and even better still, if it works for federated and distributed systems.

It is, though, a fucking hard problem, and may well be (probably is) one of those isomorphs to P=NP, so it might be impossible/infeasible to solve perfectly, but there may exist approximate solutions that are much easier to solve in reasonable time and are ‘good enough’. My own instinct is to look to nature and human behaviour for clues as to how to tackle the problem, but I harbour a sneaking suspicion that Applied Math might have some powerful tools to bring to bear on it.

The natterbox engine I have in mind’s eye (something like Twitter but with significant differences) may serve as a useful test/exploration vehicle in this space.


Observation: Trouble starts when these discovery platforms try/need to make money. To “monetize”. The only way anyone has come up with to fund a Commons Of Chatter is by advertising. This can be the blatant sort we’re all too familiar with, or it can be the slightly more subtle sort: “Enable Brands to Engage with their Market.” The first is just kak. The second is just kakker. Companies simply don’t have the faintest fucking clue how to engage in a conversation or dialogue. “Two-way” is something that’s simply not built in to the old-slow AIs that are companies. They only know how to shout at us. So advertising/marketing is simply a very kak way to monetise. Nobody gets good value (except perhaps the platform itself.)


(Offtopic: Anybody who comes up with a platform that improves/helps/facilitates real dialogue between corporate entities/old-slow-AIs and Real Human Beings™ is on a hiding to nothing. Companies really don’t want to hear from their customers lest they hear things they’d rather not know, things that run contrary to their leaders’ cherished ego-projections. Companies really only want systems that give a convincing illusion of dialogue.)


Another challenge: It is human nature to cheat… to game the system. If government/king/despot comes up with (say) a basis for taxing the populace, the populace will (quite quickly) find a way to avoid those taxes. (Numerous examples throughout history, ranging from canal-frontage taxes in Amsterdam resulting in very narrow houses, to taxes on malt resulting in people using sundry other crap to replace malt in beers, to taxes on the floor-area of buildings resulting in those buildings with precariously cantilevered upper stories you see in old European towns, to…)


Larry and Sergei came up with a slightly-more-clever way to gauge the relative reliability of pages on the ‘net called Page Rank. At the time it was better at estimating reliability than anything else anyone had ever done yet and led to dramatically better discoverability. It took about two weeks for people to figure out ways to game that system, to have their craptastic web pages rank high in Google’s algorithmic estimation. Just make thousands of shite webstuff that links to your ‘content’ and get the Google Machine to index it. The remainder of history since has been Google implementing tricks for sniffing out that sort of gaming and ignoring/penalising it (to the extent this doesn’t conflict with their advertising revenue), and of SEO types figuring out how to game the resulting “new” algorithm. The resulting caucus race is still ongoing, and nobody is winning. Least of all the poor schmuck who just wants discoverability.

What I’m suggesting here is that any scheme for facilitating discovery had better be anti-fragile in the face of human nature — in the face of attempts to game the system, even when the workings of the discovery system are known. (Sounds eerily isomorphic to the contrasts between how modern cryptosystems work compared to WW2-era cryptosystems.)