Oram?
research.kudelskisecurity.com/2020/04/22/a...
20.10.2024 00:54
👍 3
🔁 0
💬 0
📌 0
An image saying 614, and .1%
I feel like this undervalues the work I did before the company was founded.
I had intended it to map to int64 format
i = timestamp * 1024 + ClockID.
1970 + 2^53 microsecond = 2255
1970 - 2^53 microsecond = 1685
I think 1685 CE - 2255 CE is a pretty good range for computer events 🤷♂️
xkcd.com/2867/
Have you asked on
discord.com/invite/4T2ep...
hur.st/bloomfilter/... sounds right.
We have lots of wheels invented for lots of use cases. Some Authenticated Data eXchange protocol will win.
hi
https://www.nist.gov/pml/time-and-frequency-division/time-realization/leap-seconds
One of the upsides of Twitter rebranding to X is that maybe someday I’ll again be able to tell people that I worked at Twitter instead of clarifying that I worked at “pre-Elon Twitter.”
There's only one writer now. If writes move to the user agent it could pull down just the path to their new record.
Create the path back to create a new signed root.
The b#tree gives efficient client side signing without losing non-existence proofs. No keys on the server is a real use case.
Not responsible for every action of an AI, but if you know it causes harm and don't move to mitigate then yes.
So that repo can have control transferred.
Start with the PDS managing the keys and finally irrevocably transfer control to the user.
There are use cases where one repo controller may want to transfer a repo to a different controler with finality.
A triangle showing the trilemma between mutable identifiers. A name can only be two of exclusive, chosen, locally generatable.
did:plc needed to make a choice between consistent or available. By having a consensus group DIDs don't fork. If we just had delta gossip then a controller could fork their DID to two different states.
https://youtu.be/BCksQYqU5ok
https://youtu.be/BsV-4aj7PTo
https://youtu.be/O3QENUHs7yY
Not your keys, not your repo ... mostly just that you need control of the top rotation key.
That said, I would like to control the repo keys so that the PDS literally couldn't sign a commit without my user agent.
Image showing the list of PDSs by number of DIDs and the total length of all deltas.
It is just a little spammy.
bsky.social, 139,117 DIDs, 96.7 MB
uwu, 11,170 DIDs, 82.6 MB
I would also like to know what this DID's use case is
https://plc.directory/did:plc:m2o63tpnlwrfj43nrqzxyrry
still time to nullify the delta if you don't want to mutate the immutable log
```
"cid": "bafyreidaxmtdx6pb3up6tznwdbdse53uytfl7laql4cdlig22zhktkhfjy",
"createdAt": "2023-06-01T04:27:23.090Z",
"did": "did:plc:z72i7hdynmk6r22z27h6tvur",
"nullified": false,
```
Seems more like a bug in the PDS then the DID Directory. Being willing to sign an update where the DID String does not match the DID String for the account. Also the poor practice of reusing key material.
As long as the rotation key holder signs things they shouldn't the directory can't help you.
The Repo is a tree. If a PDS wanted to do a narrow checkout that only pulled some paths it could. e.g. Pull posts leave the long videos. Once you pull the root CID of the repo you can sync the rest at your leisure.
Feels like the lexicons should be more related to the records. I like it though.
Yup
ipfs://bafybeiddeohylecumwkpbhnyejtkt3rhnnpcbgqwjmlyqejuqok74ghrtq
think of it as an opportunity for those domain names that clutter up the everything drawer
just setting up my bsky
