3 ms·
Speaking specifically to the certification point, it is a complete non-starter. There are multiple reasons why it won't work: 1. I don't want an identity certi
by jerf 4y ago
Speaking specifically to the certification point, it is a complete non-starter. There are multiple reasons why it won't work:
1. I don't want an identity certified. I'm sure there's a human somewhere. I want the content verified as not having been autogenerated. Scammers can trivially get "verified", there are plenty of humans that will be able to be "verified", but then generate arbitrary amounts of spam on these identities.
2. I can't even remotely conceive of a way for the incentives of the putative "certifier" to line up correctly. They will be incentivized to take everyone's money and mark them "certified" with as little (expensive) effort as possible. It's the same problem TLS certs had, before we basically collectively admitted that was the case and fell back to the Let's Encrypt model, only orders of magnitude worse.
3. Plus even if the incentive structure was correct, there would still be enormous motivations to cheat. It isn't just "spammers" who want to use this tech. Some of the users will have the political firepower to throw around and get themselves "certified" regardless of the underlying situation, further diluting the marker.
To be honest, what this may well mark is the end of the international internet as it stands now, and not much less, and possibly sooner than you think. The only solution is some sort of web of trust, which I use in a broad sense of a solution class, not the exact GPG web of trust or something. But you're going to have to meet people in person and make your own decisions.
Though we'll probably pass through a decade in which we try the centralized authority thing, before people generally realize the central authority simply does not and can not have their best interests at heart and no possible central authority can ever be trusted with the assertion "This person is worth listening to and this person is not", no matter what form it takes. Too many predatory entities standing by and watching out for any such accumulation of authority/power and standing by to take it over and drain it of all its value. Too much incentive to cheat and not enough that can be done about it.
- irq-1 4y agoI think there's something wrong in the way were looking at this. Free services are made to get humans attention for advertising or eventual payment. If a service like reddit is overrun with bots, won't the services shutdown? There won't be any "public" places like now, except for forums like this where monetizing isn't the goal. I'm not sure it's a problem for a place like hackernews -- spam would be a problem, but we know that. Voting and verifying that content is 'good' could be an issue, but why would the bot/AIs care about hackernews internet points? Email was open to everyone, and when the spam came we filtered it. Comments on news articles, youtube and amazon reviews are already mixed with degrees of 'bad' and uselessness, and we mostly ignore it. Only the 1% comment, 10% vote, or something like that. Generated content made by us and for us seems more likely as the future, and confirmed identity doesn't matter much for that.
- dsign 4y agoI agree with most of what you say, except for this bit: > Too many predatory entities standing by and watching out for any such accumulation of authority/power and standing by to take it over and drain it of all its value. > Too much incentive to cheat and not enough that can be done about it. That assumes that people will stand by and watch the too many predatory entities and do nothing, and under "current assumptions", that's exactly what will happen. But current assumptions can be broken. For example, there could be a revolution and certain/a few/some governments may lose the monopoly in violence. Such an idea, alien as it looks to us now, was the way of things through most of history and it is the way of things--sadly--in too many places still today. If citizens have a relatively efficient mechanism to keep the certification authority trustable, then the certification authority may solve 70% of the problem. Web of trusts are also a possibility, but yes they will be regional and in need of strong baking by face-to-face meetings. Probably all trust mechanisms will like that: local and patchy. I've got to add my own note of pessimism: relatively trivial exchanges of information won't survive the AI age. Sure, business people will find ways of negotiating prices for goods and services across borders, scientists will meet at conferences and exchange data and results, and the Interpol will talk to trusted authorities in the member states using cryptographically-sound channels and base agreements and methodologies. But the general public will trust very little of what they see in the Internet. We will disconnect.
- pixl97 4y ago>We will disconnect. And replace it with? I'm sure you'll say "A local group of people we trust", but in general in history the locals will have been dumb as hell too, and only trustworthy because you had no other means of validating what they said. And that would only cover the people disconnecting, not the converse... The AI won't disconnect from you. You still have to buy goods, you'll still need services. AI will be there watching everything you do all the time. Where your car drives. The people you meet up with. Always calculating, always optimizing. Too much power for the greed of humans to ever put back in the box.