Interesting points. The below is a friendly disagreement from someone passingly interested in this field. > First, people should have multiple keys/identities

Interesting points. The below is a friendly disagreement from someone passingly interested in this field. > First, people should have multiple keys/identities by default I feel like you can do this already, no? Just create different Keybase accounts for each identity. I don't understand how having a single keypair prevents you from managing your contact groups however you like. They seem like orthogonal problems to me. > Second, matching a name with social media is the wrong way to lookup others. It only works if people put enough personally identifiable information (PII) on their social media accounts Key verification is the problem in crypto. Like it or not, many people already put tons of PII online. It's a perfectly valid way to verify lots and lots of people. Perhaps it doesn't work for everybody, but it's clear that keychains don't work. This is a different strategy to solve the verification problem, and I think it's got legs. > Third, leaving people to manage their own private keys is worse for security than having them managed by others. I understand your point here, but I disagree. Given all the server hacking going on these days, I'd much rather not have a single point of failure. Or several large points of failure, in the case of your distributed system. The odds of someone hacking keybase.io or keyexchange.google.com seem much higher than the odds of someone attacking my phone or my desktop. I think it's a mistake that keybase allows uploading and storing private keys. > I think the winning solution will be: a distributed Your idea doesn't seem to solve the verification problem. Keybase's strength is that in order to subsume someone's identify, you have to hack several of their accounts. That's much more difficult than hacking a single keyserver. How do you verify that the key I get when I search your crypto-DNS system for irq-1 actually belongs to you?