4 ms·

To me, this feels like a precursor to the actual issue that's going to happen. Eventually, we won't be able to distinguish between real humans and bots online.

To me, this feels like a precursor to the actual issue that's going to happen. Eventually, we won't be able to distinguish between real humans and bots online. No CAPTCHA will work, no Turing test, none of that stuff. Message authentication (e.g. keysigning, or a centralised provider you trust) works for individual identities that you're already aware of, but the general case of trusting comments online? Even without all of that, you've never had any idea whether the "person" replying to your comment is paid, or has a bias, or whatever else. If, say, 30% of Hacker News commenters were "fake" in that future scenario, would we know? How would we know? Once we develop a method, can't that be gamed?

rawrmaan 8y ago
It's one of the biggest problems that I worry about with the internet. Yes, public key cryptography can prove that "message x came from source y". But nobody's figured out a way to prove "source y is a person" or even better "source y is definitely the person they say they are". I like the approach projects like Keybase and Stellar are taking: Building a decentralized network of trust based on existing, real-world trust-based relationships.
ryandrake 8y ago
As for HN, do you think it’s less than 30% today? If so/not,how do you know? What do you count as “fake”? Actual bots, sure, but what about a real person running an automation that spams overnight? What about many real people mass-typing “opinions” on behalf of someone who pays them? What about a single businessman typin a falsehood that helps his business? Which of these count as fake?
- esotericn 8y ago
  a) I don't know! b) All of the ones you've mentioned, at least. I deliberately didn't define it because I think it's difficult.
opwieurposiu 8y ago
If you had to pay a fee to post a comment, the bots would not scale. Even 1 cent would work.
glglwty 8y ago
What if people actually like those "fake" comments more than comments generated by humans?

rawrmaan 8y ago

It's one of the biggest problems that I worry about with the internet. Yes, public key cryptography can prove that "message x came from source y". But nobody's figured out a way to prove "source y is a person" or even better "source y is definitely the person they say they are". I like the approach projects like Keybase and Stellar are taking: Building a decentralized network of trust based on existing, real-world trust-based relationships.

ryandrake 8y ago

As for HN, do you think it’s less than 30% today? If so/not,how do you know? What do you count as “fake”? Actual bots, sure, but what about a real person running an automation that spams overnight? What about many real people mass-typing “opinions” on behalf of someone who pays them? What about a single businessman typin a falsehood that helps his business? Which of these count as fake?

esotericn 8y ago
a) I don't know! b) All of the ones you've mentioned, at least. I deliberately didn't define it because I think it's difficult.

opwieurposiu 8y ago

If you had to pay a fee to post a comment, the bots would not scale. Even 1 cent would work.

glglwty 8y ago

What if people actually like those "fake" comments more than comments generated by humans?