Yes - the promise of these critters is to really comprehend and empower human intention - which requires context. Shared team and community context is also important to help us work together, and i don't see that done well yet (except the $25/seat but that is not inclusive for communities).
Thank you for posting this. Context management, crystalization and summaries, interfacing with normal formal tools, on behalf of the user's intention, this is the current front line.
see below. i think your comment is mostly disingenuous so not responding in detail. but yes i spent some years fighting scammers/bots/fraudsters with some success, this is what gave rise to the thoughts in this article. "social" problems are solved by tech all the time, or do you not have a house with doors that close?
right, so when we ask it who it is, we ask it to prove it. signing with a private key is a kind of one-way street, a proof that anyone can verify but only the key holder can make
and yes it has vulnerabilities - quantum computing or more commonly lost keys or stolen keys
but, its a tool - its like we don't abolish ids because sometimes they are fake. it takes effort to fake them and it is relatively hard in a good DID system- some would say impossible, but anyway you have to grab the private keys
yeah these things happen, you can call them account takeovers, or in the model you give, purchases. And yes, it will happen. every strategy gives rise to counter-strategies. but in general, a game with long-lived actors and some transparency has the cooperators winning a bit more.
so bit of background, Ayush and I were on the Postmates Risk team a few years back, and have experience building models and features to block fraudsters at scale.
One of the key learnings in risk is that long-lived assets - ip address, device id, login, anything that can be tracked over time - these are key to distinguishing fraudsters from real users.
and this article is talkinga bout that in the more generic world of AI bots and humans who may not be distinguishable, and how to hold AIs responsible when they can more easily invade spaces that used to be limited to humans who could navigate complex interfaces, and interact convincingly.
So its not a new problem, but the threat models are adapting, and pointing out that making use of DIDs more of an expected standard is important to respond to them.
I think i said that in the article! but clearly some folks in the comments thought i've never blocked a bot before...
Techdirt has not posted any stories submitted by Golda Velez.
Intention is all you need
Yes - the promise of these critters is to really comprehend and empower human intention - which requires context. Shared team and community context is also important to help us work together, and i don't see that done well yet (except the $25/seat but that is not inclusive for communities). Thank you for posting this. Context management, crystalization and summaries, interfacing with normal formal tools, on behalf of the user's intention, this is the current front line.
uh, this is from real world experience dude
see below. i think your comment is mostly disingenuous so not responding in detail. but yes i spent some years fighting scammers/bots/fraudsters with some success, this is what gave rise to the thoughts in this article. "social" problems are solved by tech all the time, or do you not have a house with doors that close?
cryptography proofs
right, so when we ask it who it is, we ask it to prove it. signing with a private key is a kind of one-way street, a proof that anyone can verify but only the key holder can make and yes it has vulnerabilities - quantum computing or more commonly lost keys or stolen keys but, its a tool - its like we don't abolish ids because sometimes they are fake. it takes effort to fake them and it is relatively hard in a good DID system- some would say impossible, but anyway you have to grab the private keys
yep. that's all part of the game we are playing
yeah these things happen, you can call them account takeovers, or in the model you give, purchases. And yes, it will happen. every strategy gives rise to counter-strategies. but in general, a game with long-lived actors and some transparency has the cooperators winning a bit more.
lessons from risk/fraud
so bit of background, Ayush and I were on the Postmates Risk team a few years back, and have experience building models and features to block fraudsters at scale. One of the key learnings in risk is that long-lived assets - ip address, device id, login, anything that can be tracked over time - these are key to distinguishing fraudsters from real users. and this article is talkinga bout that in the more generic world of AI bots and humans who may not be distinguishable, and how to hold AIs responsible when they can more easily invade spaces that used to be limited to humans who could navigate complex interfaces, and interact convincingly. So its not a new problem, but the threat models are adapting, and pointing out that making use of DIDs more of an expected standard is important to respond to them. I think i said that in the article! but clearly some folks in the comments thought i've never blocked a bot before...