A cool crab wearing shades

@neckspike@indiepocalypse.social
175 Followers
302 Following
19.3K Posts
They | Lesbian | Transphobes love to guess my AGAB wrong | #KamenRider #tokusatsu #SuperSentai #Ultraman #Dogengers
PronounsThey/Them
Website:http://www.pineapplecove.org/
Signalneckspike.420

The Pig and her poster have been reunited. MissPiggy ran XENIX-11 at Microsoft from 1979 to 1987. These days, misspiggy runs Version 7 UNIX from which XENIX was derived and its disks are virtualized so they can run under simulation and the real hardware.

Please support our preservation efforts at https://icm.museum

We have a partial backup of MissPiggy and one day hope to run XENIX-11 on it once again.

#community #vintagecomputing #retrocomputing #unix #microsoft #history

"But Augie, if repeated #COVID19 infections were harming brains, wouldn't we see evidence of that?" Thanks for asking. Yes, we would. Diminished cognition, memory impairment, attention deficits, and damage to executive functions would lead to multiple growing problems, such as:

A doubling of unruly air passengers compared to before COVID: https://www.faa.gov/unruly

1/3

@EverBeyondReach
I love that movie. New Line fumbled the bag so hard on the sequels, they could've been printing money with a new Blade adventure every 2-3 years
@moonsea
Tumblr: it sucks and our owner desperately wants to be store brand Elon Musk
@moonsea tumblr loves doing that
@ai6yr it's quite nice when you get the rhythm correct! I think about Grammie's machine whenever these pop up, I don't remember it clearly enough to ID the model anymore but it was in beautiful working shape and I hope it ended up with someone who loves it very much.
@ai6yr One of my great regrets is I wasn't able to take my grammie's treadle powered Singer when she had to move into assisted living.
@ai6yr These old machines are insanely durable and a lot of the parts like motors and treadles are interchangeable or easily adaptable, so they last basically forever.

Been at the library a lot this summer holidays, just got back now and thinking about malls and flea markets and libraries

Y'know how you've got that little crow inside you that enjoys the act of Going Somewhere, getting Bags of Shinies, then coming home and examining those shinies, and sometimes you gotta give the crow a treat?

🐹 Man, you're so consumerist-minded, you gotta let go of the -
🦝 Shut up you have a crow as well, anyway as I was saying, the act of going out, getting a large amount of Things and bringing them back satisfies the crow. It gives the crow warm feelings of abundance and security, and the mall and the online shopping apps can feed that crow but at the expense of your wallet.

Daughter just filled up two big canvas bags of library books. Overflowing. Swollen. Our rule is she can take as many as she can carry, and we always end up breaking it. Bulging, heavy sacks of shinies.

If you've got a hole in you that could be filled by getting heavy bags of shinies, the library keeps that crow FED. The crow doesn't know the difference between shopping and borrowing, the heavy bags taste the same, it just cares about Bags Of Stuff, it's the same feeling.

A Little Adventure to Go Somewhere and Acquire Stuff. It fills the same hole.

Except the library ALSO feeds the beast that enjoys REMOVING stacks of things from the house and going "Ahh, that's better, bit more room in here now," because after a couple of weeks you've got to take them back

...and get more

×

A research study on using AI coding assistants has shown that while developers believed they got a 20% productivity gain from using the tools, measurement showed a 19% decrease in productivity.

METR recruited 16 open-source devs to complete 246 real coding tasks, randomly assigning each to “AI Allowed” or “AI Disallowed.”

Devs estimated time saved with AI before & after each task, while screen recording to measured real-world AI productivity gains.

https://secondthoughts.ai/p/ai-coding-slowdown

@carnage4life I can believe this. When you're vibecoding it feels like the LLM is doing a lot of good work for you. But when you go to examine what's actually been accomplished it's another story. They'll confidently tell you that some task has been completed but when you go to check the code it's hardcoded things in such a way that tests pass. (I've run into this recently)

LLM coding assistants are lazy - you need to be very careful about this if you're coding with them.

@carnage4life I am having to temper my reaction to this given how badly I want it to be true

@carnage4life

This jibes with my personal experience. When I try out AI for non-trivial tasks, it's a lot of prompt formulation & post-AI correction. Actually have to get a task done? I skip the AI stuff right now.

I really do think the "cookie cutter a basic unit test suite", "wire up a skeleton for this API definition", and similar use cases will have real value. But like the creative world is finding elsewhere, AI might help with the "How", but it still needs a human to provide the "What".

@jlbec @carnage4life I think that, in a mature dev-ops pipeline on a healthy code base with not too much tech debt AND very (very!) prescriptive user stories..... 20% productivity boost could be gained, exavtly for what you mentioned (backbone stuff).

.... it's just, is it worth killing the planet for?

@carnage4life I guess this really depends on the type of task. But finally a study and not some anecdotal evidence 🙂

@carnage4life

An excellent review of the process by someone who does software productivity studies for a living here
https://mastodon.social/@grimalkina/114837000070700830

@carnage4life I am glad some real world evidence for this has finally been presented, I've been saying this for at least 2 years now

I don't use AI because I'm measurably less productive, it slows me down to explain shit to an AI and make the output actually work, and then review the output, instead of just thinking real hard and writing the damn thing myself

any developer using a tool that measurably slows them down during their workflow is lost tbh

@carnage4life In Weinberg (The Psychology Of Computer Programming) it was suggested that actually nothing much is known about how real programmers behave, because nobody could afford to pay them to take part in studies or experiments, so all that was actually known was stuff about how computer science students behaved. Students either did what they were told or were very cheap to bribe into taking part in experiments (I got £5 for one, which was significant to me but trivial to the grant for the experiment).

Open source developers now appear to be an alternative source of relatively cheap study material.