4 ms·

> Manual human labor requirement creates a rate limiting which is removed by automated content creation in the name of the human. What if we impose artificial

> Manual human labor requirement creates a rate limiting which is removed by automated content creation in the name of the human. What if we impose artificial rate limiting with posting fees or some kind of proof-of-work requirement? > I wonder if we will look back and see that we have witnessed the creation of the nuclear weapons of information warfare. That's a terrifying thought.

pixl97 6y ago
Proof of work can easily be farmed off on botnets.
- cosmojg 6y ago
  Oh, definitely, but that will get very expensive very quickly, especially with a self-updating difficulty mechanism. I don't see botnets being much of a problem outside of one-off posts (or maybe a few more for state actors).
Reelin 6y ago
> > Manual human labor requirement creates a rate limiting which is removed by automated content creation in the name of the human. > What if we impose artificial rate limiting with posting fees or some kind of proof-of-work requirement? These hypothetical anonymized but authenticated interactions are backed (via ZKP or otherwise) by a trusted identity (which itself might be pseudonymous, government verified, whatever). As such, why not track and publish the (approximate) rate at which attestation interactions are undertaken by a given identity (per some very coarse grained unit of time)? This rate could then be linked to the otherwise anonymized actions upon attestation. Then people or platform moderators would be free to make their own judgments about what constitutes a reasonable (ie trustworthy) rate of interaction within a given context. (This is a form of information leakage, so in certain scenarios it could allow deanonymization or at least linking anonymous interactions together. Extreme care would have to be taken when selecting the degree of approximation and the duration of unit time.)
Sebb767 6y ago
> What if we impose artificial rate limiting with posting fees or some kind of proof-of-work requirement? This is going to be a tough one. Someone with access to GPT-3 is going to have access to enough funds to post extensively. Also, muting poorer people is not a good idea. Proof of work seems more promising, but you'll have trouble finding something that can't be automated. Even with captchas - when you're solving them, you're effectively training Googles network to solve their own challenges.

pixl97 6y ago

Proof of work can easily be farmed off on botnets.

cosmojg 6y ago
Oh, definitely, but that will get very expensive very quickly, especially with a self-updating difficulty mechanism. I don't see botnets being much of a problem outside of one-off posts (or maybe a few more for state actors).

Reelin 6y ago

> > Manual human labor requirement creates a rate limiting which is removed by automated content creation in the name of the human. > What if we impose artificial rate limiting with posting fees or some kind of proof-of-work requirement? These hypothetical anonymized but authenticated interactions are backed (via ZKP or otherwise) by a trusted identity (which itself might be pseudonymous, government verified, whatever). As such, why not track and publish the (approximate) rate at which attestation interactions are undertaken by a given identity (per some very coarse grained unit of time)? This rate could then be linked to the otherwise anonymized actions upon attestation. Then people or platform moderators would be free to make their own judgments about what constitutes a reasonable (ie trustworthy) rate of interaction within a given context. (This is a form of information leakage, so in certain scenarios it could allow deanonymization or at least linking anonymous interactions together. Extreme care would have to be taken when selecting the degree of approximation and the duration of unit time.)

Sebb767 6y ago

> What if we impose artificial rate limiting with posting fees or some kind of proof-of-work requirement? This is going to be a tough one. Someone with access to GPT-3 is going to have access to enough funds to post extensively. Also, muting poorer people is not a good idea. Proof of work seems more promising, but you'll have trouble finding something that can't be automated. Even with captchas - when you're solving them, you're effectively training Googles network to solve their own challenges.