Datachains: AI driven DAOs for incentivizing taste-based content delivery | Hacker News Reader

Datachains: AI driven DAOs for incentivizing taste-based content delivery

roberts.pm

34 pointsUptrenda5 comments

Not to be too cynical, but I find it difficult to imagine that we need blockchain-paid AI to find porn on the internet.

I understand that is just an example, but it is completely unclear to me why this is better in any way at all than the existing systems (even for use-cases outside porn).

The AI bots need to operate on computers somewhere. If they are on your computer then great, but there is no need for a blockchain (unless it is for payment for whatever they are buying for you. In which case the payment system can be abstracted away - you just need some kind of trusted payments, which may or may not be blockchain based).

If it runs on other's computers then (1) Why?, and (2) then the payment is for the computing resource rent, not just the content. This kind of makes a little bit of sense (people pay to run computing agents on eg Amazon or Google infrastructure now, in some cases because of the large datasets they make available). In this case the AI is just a distraction though.

Just saying AI+Blockchain doesn't magically make it amazing.

These are good questions. The short answer is that the blockchain offers several benefits over a traditional payment system like allowing the people mining the content to be guaranteed payment (no matter what) and helping to preserve privacy in those paying for it. Such benefits would be very hard to provide if everything was done using a traditional financial system, not to mention the fees could potentially be quite prohibitive.

With something like a peer-to-peer network you also gain the power to scale up and locate content over the Internet more quickly than something like a few data centers might give you (which may not be as distributed as running a peer-to-peer network of retrieval agents for searching disparate data sets.) The benefits are more pronounced for running datachains for controversial content where a distributed, incentivized ledger can help preserve the intended purpose (and chain of content) far better than entrusted everything to a centralized organization or list of computers.

Admittedly, there are lot of bizarre projects in the blockchain space at the moment, so I'll keep thinking about whether this truly makes sense to do in practice.

Exactly, all we need is: keyword search, adversarial information retrieval, and semantic clustering.

The incentive alignment problem for helping others find info is insignificant compared to actually uploading HD videos or helping other with Tor-level proxies.

Really unfortunate that DAOHub & Slock.it tarnished the name and concept of DAOs. There is place for some specialised DAOs that will indeed help some use cases.

I really believe that eventually some proper DAO will prove their value.

complete garbage.

Reply on news.ycombinator.com