Yep, Pangram says they think false positives are way worse than false negatives, so if their AI isn't sure enough they prefer a "not ai" output to an "is ai" output.

Are you sure? If you're household income is <$340k (depending on details) you'll still get some scholarship, and as long as your household income <$200,000 tuition is free.

[1]: https://college.harvard.edu/financial-aid/net-price-calculat...

> All told, the governor said that her major subsidy projects would create 20,595 jobs in Michigan

Even using these numbers that works out to $135k/job, which is bonkers!

Creating personal copies of copyrighted works are allowed. (Also, libraries really don't mind if you take pictures of the content of works they have.)

> There was a thread yesterday

said thread was: CT scans of BYD car parts (https://news.ycombinator.com/item?id=48375824)

Small nitpick, but it's also possible to communicate by changing the location.anchor property (by either the iframe or its parent window.)

Checking this shows that the top 2 performers in this graph lost more money (~8%) in 2008 than the bottom 2 (~2%)

> Signal ships safety numbers because the platform might one day be compelled or compromised, and the architecture is meant to let you catch that. But almost nobody verifies

We have a solution to this! Wa and Signal both have key transparency. This uses cryptography to make it possible to verify that everyone is getting the same data[1]. Now your phone can check the keys listed under your username are all keys you made (and your contacts can check this too!)

Edit 2 (quick note): if you don't trust the app on your phone to verify your keys, then you also can't trust it to show you a valid security code, or do what the author proposed in their product spaces.

Edit:

It's also striking how similar (in essence) the current solution is to the solution the author is working on/proposing:

> Spaces takes this shape. (Disclosure: I work on it.) Issued names live in a binary Merkle trie. The root of that trie is committed to Bitcoin’s chain, used here as a widely-replicated, hard-to-rewrite timestamp service

Fundamentally the same: the name is your phone number (or alternatively in signal your username), key transparency also uses a merkle tree based structure. Instead of using the bitcoin chain as a consensus mechanism, key transparency implementations generally use trusted witnesses: simple servers that promise to only sign consistent versions of the merkle tree. This is better! Because essentially no clients (phones) have a local copy of the bitcoin chain, so you still have to trust a server to tell you what was posted in the bitcoin block.

For the rest current key transparency systems also have verifiers, which verify that the append only merkle tree is transformed into a dictionary legitimately (this is pretty compute intensive, and needs to be done by a trusted server too. WA currently contracts cloudflare as their only verifier). Spaces would also have to do this to be secure if they reach any scale, but this isn't mentioned in TFA.

Also a message for the author: Key transparency is cool tech, but you shouldn't reinvent the wheel! I hope you research current solutions more! You can ask questions in the transparency.dev slack (https://transparency.dev/slack/)

[1]: There are a bunch of details here. You need to check that everyone _is_ actually getting the same data. There are multiple ways to do this. The transparency ecosystem has generally stabilized on a system where you have trusted verifiers. But anyone (yes you!) can setup a server that can help monitor the chat app and trusted verifiers.