RIP to a friend, 33 is too young.
jonny (nonvenomous)
@jonny@neuromatch.social
known or reasonably foreseeable hazard
Digital infrastructure 4 a cooperative internet. social/technological systems & systems neuro as a side gig. writin bout the surveillance state n makin some p2p.
information is political, science is labor.
science/work-oriented alt of @jonny
This is a public account, quotes/boosts/links are always ok <3.
- manifesto (web)
- https://jon-e.net/infrastructure/
- pronouns
- they/them
- personal
- https://social.coop/@jonny
- other git
- https://github.com/sneakers-the-rat/
- radicle
- did:key:z6Mkfon2LQyRcsGnHxGz6sKu2RvEj6c76u7eF4aNsuQo1E41
- pgp
- 5AD4A79EB302DFB716302F896DCB96EF1E4D232D
Replying to @jonny@neuromatch.social
Spider Baby suffered from poor marketing as well as a series of title changes, being billed alternatively as The Liver Eaters, Attack of the Liver Eaters, Cannibal Orgy, and The Maddest Story Ever Told. Although these alternate titles have little or no relation to the plot, the latter two appear in the opening narration by Chaney: "This cannibal orgy is strange to behold in the maddest story ever told."
movie titles, as always a complete free-for-all
Replying to @jonny@neuromatch.social
i like seeing what @ryan picks for the double feature - like this isn't a #monsterdon movie because it intends to be a comedy, but it sure looks like a nice complement #monsterdon
Replying to @jonny@neuromatch.social
I'm gonna stick around on the #miru stream for the double feature, i likea spider #monsterdon
Replying to @jonny@neuromatch.social
Earth vs. the Spider (1958) - 3/5
a perfectly average #monsterdon movie, representative of the middle tier of the genre
Replying to @jonny@neuromatch.social
ONE MINUTE REMAINING FOR A MARRIAGE #monsterdon
sure, sure, plasma arc, that's what spiders are vulnerable to #monsterdon
tried the latest and greatest "AI" against our open issues, "fable" that's supposed to be the end of programming forever. if i am being extremely generous, on first glance, 1/5. on review, 0/5. attempted issues were extremely easy, i think with reading the library these would take about 15 minutes each, tops. These are issues written for other human beings and are relatively sparse on details, in this case a quiet project with me and another person as main authors, but on review, these should all be doable with a brief skim of the code and docs - a human could parse the details here and implement these with minimal discussion or time. I timed myself, 5-15m each, as the primary author.
- 147 - everything in our package is run by events, so nodes expiring should be an event. the "AI" took the issue literally and just suppressed errors in 2/3 runners. incomplete, undesireable. fail.
- 239 - no output was emitted, claimed issue was solved, fail. This one in particular is minimal and easy, a three line fix.
- 200 - closest to success, initially ruled success, but on further review failed. Need to turn a scalar-valued reference into a set. This has the literal solution in the issue text. "AI" implementation increased algorithmic complexity of O(1) to O(n). unacceptable, fail on review.
- 110 - handled in the laziest way imaginable that latched on to the issue title but failed to read the actual problem within the issue text, which is relatively verbose and poses several possible solutions. "AI" chose none, did the wrong thing. fail.
- 166 - this one actually was already solved and i wanted to see if the "AI" recognized that. it added an additional unnecessary mechanism that made the relevant code less efficient. fail.
model took ~60 minutes, used my entire token allotment, roughly the time it took me to do all of them correctly. i'd say it would take someone unfamiliar with the library 2x that but that's non-empirical, just a guess. if i had to sit and coach the "AI" through all of these it would take longer, i'm sure, even assuming i had unlimited tokens. easy issues. each of them is a few lines. all it takes is being able to understand the context of what's being done.
@jdp23
Its been said to death, but the thing that LLMs do that's useful for programming is basically serve as an extremely lossy compressed index to all the code in the training set while making it look like a miracle that you are creating. If we instead had an actual search index where I can search for related code in an abstract AST space, see prior implementations in their context with proper attribution, compare where mine and their differ, that would be about a billion times more useful than an LLM for code. Something like the Wikimedia algorithm library would be a better tool than an LLM if it had the same amount of money poured into it, but it wouldn't get that money because it doesn't launder authorship or make you feel like a special genius.
Replying to @jonny@neuromatch.social
(This is an invitation to show me oddbody furbies or other monstrosities, should anyone have any to spare)
disregard holiday, go hunting for oddbody furby tumblrs