[elseweather]

This is a fascinating article that winds through a whole raft of problems that feel super relevant in the software world today: end-to-end encryption, distributed messaging, what we would today call "containerization" (in 1984!), many other things besides. I often dislike this kind of abstract general nonsense but I think there's a really interesting through-line here I hadn't noticed before.

One thing that struck home with me was this line:

> Some of these source files were corrupted by byte exchanges, and their owners were forced to the ultimate end-to-end error check: manual comparison with and correction from old listings.

At work I sometimes joke that us SWEs are the last line of retry-and-checksum defense, but it's not a joke! This paper argues correctly that engineering extreme reliability into lower level common components is uneconomical, but on the other hand pushing reliability all the way up the stack doesn't work either. If you surface the jank all the way up to your customers you just have a buggy, crappy product and your business will fail. One level down, if you rely on talented engineers to fight operational fires all the time you have to pay a lot of expensive engineers to not make progress delivering features.