Mastodon Feed
Boosted by dysfun@treehouse.systems ("gaytabase"):
gsuberland@chaos.social ("Graham Sutherland / Polynomial") wrote:
squirrels are so good at climbing trees because they have branch prediction
Mastodon FeedMastodon FeedBoosted by glyph ("Glyph"):
dpnash@c.im ("David Nash") wrote:
@glyph @ildiavolorosso @Firesphere @pluralistic
Case in point follows.
For context, bear in mind I have been, and remain, extremely skeptical of the long-term utility of this tech ever since ChatGPT gave me a giant wall of bullshit in response to a question I knew the answer to, back in ChatGPT 3.5.
I was testing out a then-current release of Claude Sonnet to be sure I wasn’t creating a strawman argument about the state of the art based on older impressions.I handed it a simple programming task with a graphical component (plot some data in 2D or 3D). A program like Claude Sonnet might be able to generate code for the plot, but being a large *language* model, it can’t actually analyze the plots at all competently. After getting the “base” program from Sonnet, and having it come up with an expectedly garbagepants analysis of the plot, I had it tweak the parameters of the plot a few times so it was easier for a fellow human to see that the analysis was consistently garbage. I had Claude do this a few times before collecting enough plots to make the case clear: even with a syntactically clean program, it wasn’t able to comment intelligently on the output.
The biggest problem here was *not* that Claude produced crappy analyses of graphical data, though of course, that is indeed a problem.
The biggest problem here is that I didn’t read the program itself for a while and tinker with the output it gave — something I have done many times with programs I’ve been developing — but instructed Claude to do the tinkering itself.
I had gotten sucked into an approach to debugging things that I not only did not intend to at the start, it’s one I know I generally dislike. I’d let my hands come off the wheel and let Claude take over much of the actual interaction with my source code, *even though I actively believe this is a bad idea*.
I take this as very strong evidence that personal assessments of how you are *actually* using apps like these are fundamentally unreliable, and only non-subjective impressions, agreed on in advance by the people involved, are appropriate for making judgements about how useful and productive working with gen-“AI” coding tools actually is.
Mastodon Feedabbreviations can be useful for saving on typing but in documentation or formal writing, always remember to use long-form versions of technical terms to make them more readable.
for example,
ghost in the init .
ghost in the help
ghost in the clone
ghost in the fetch
ghost in the push
Mastodon Feedjsonstein@masto.deoan.org ("Jeff Sonstein") wrote:
getting set up for weekly local voice net on 50.250MHz USB
Mastodon FeedBoosted by jonny@neuromatch.social ("jonny (nonvenomous)"):
GrapheneOS@grapheneos.social wrote:
They have their tool pretend to be a certain USB peripheral to target specific Linux USB drivers. These exploits are widely used including against journalists and political activists. Here's a technical article about a specific case of this in Serbia:
Mastodon FeedBoosted by glyph ("Glyph"):
selectric@tacobelllabs.net ("liz ten eleven") wrote:
they call it single sign-on because YOU HAVE TO SIGN ON EVERY SINGLE GODDAMN TIME
Mastodon Feeddysfun@treehouse.systems ("gaytabase") wrote:
so i see there is a position open to work on element.
do you think i'd gain or lose points for adding a joke about unable to decrypt message?
Mastodon Feeddysfun@treehouse.systems ("gaytabase") wrote:
how can people say linux isn't enterprise ready when it's just as full of slop as enterprise stuff?
Mastodon FeedBoosted by dysfun@treehouse.systems ("gaytabase"):
dysfun@treehouse.systems ("gaytabase") wrote:
@alina @gsuberland no, matrix is a distributed denial of service technology
Mastodon Feeddysfun@treehouse.systems ("gaytabase") wrote:
the JSON performance crowd
ah yes, you know you're in the very top tier of performance when you elect to use JSON.
Mastodon Feedapparently people are out here rocking 30g switches. i assume to do that you first have to replace all your fingers with carbon fiber wireframes first
Mastodon FeedBoosted by jsonstein@masto.deoan.org ("Jeff Sonstein"):
antlerboy ("Benjamin P. Taylor") wrote:
Teaching with the wrong ends of the constructivist stick – Christian Moore-Anderson
https://christianmooreanderson.com/teaching-with-the-wrong-ends-of-the-constructivist-stick/
Mastodon Feeddysfun@treehouse.systems ("gaytabase") wrote:
oh goody, the ECB has a youtube channel and they have a video out about "taking care of your money".
which is good, because i'd ordinarily assume they were spaffing it up the wall on AI datacenters or something.
Mastodon Feeddysfun@treehouse.systems ("gaytabase") wrote:
so which egyptian god is it that's reborn tomorrow? i've lost track.
Mastodon Feeddysfun@treehouse.systems ("gaytabase") wrote:
guess which silly spattered boiling hot fat all over their chest, legs and face.
it fucking hurts.
Mastodon FeedMastodon Feedjsonstein@masto.deoan.org ("Jeff Sonstein") wrote:
testing backyard setup for tonight’s local voice net on 50.250 MHz USB. going to use a rigid dipole this time around
Mastodon Feedkeyboard update: 43g is _real_ light actuation force, I might have overdone it with these "fast" switches
Mastodon Feeddysfun@treehouse.systems ("gaytabase") wrote:
the first example was scottish soldiers' chest measurements
Mastodon FeedBoosted by kornel ("Kornel"):
ellecordova@kilogram.makeup ("Elle Cordova") wrote:
Massive new data centers are being built across the country whether we’re ready for them or not. This is our moment to decide the terms at zoning meetings and utility hearings, in written testimony, and at the ballot box.
As hyperscalers come into communities to build these data centers, we should be asking: what tech will it be using, and how is it powered? Who pays for any necessary grid upgrades? What benefits does the surrounding community receive, and are those promises legally binding?
I have another video that will go into the details on all of this coming soon, so watch this space for more info!
Attachments:
- video: cff592da6c0feac2.mp4
Mastodon Feedit's that time again! upgrade your #python on macOS with `mopup`:
Mastodon FeedBoosted by glyph ("Glyph"):
ronald@blog.ronaldoussoren.net ("Ronald Oussoren") wrote:
I’ve released #PyObjC 12.2.2. This release can be build with Xcode 27 and the Python 3.15 wheels where build with 3.15rc1.
Mastodon FeedBoosted by brib@bribstodon.xyz ("brib :neofox_floof: :Nonbinary:"):
joepie91@slightly.tech ("Sven Slootweg, ("still kinky and horny anyway")") wrote:
reminder: all those 'homemade hacks' for looking at an eclipse (looking through CDs, sunglasses, etc.) are going to damage your eyes permanently, and you won't notice until much later. they do not work.
you have exactly three valid options:
- look through certified eclipse glasses from a known-good manufacturer for <30s at a time at most
- go to your local observatory, they probably have equipment set up to do this safely
- project the eclipse on the ground using small holes but DO NOT look at the eclipse directly, also not through small holesanything else will leave you with eye damage for the rest of your life. you do not want eye damage.
Mastodon FeedBoosted by brib@bribstodon.xyz ("brib :neofox_floof: :Nonbinary:"):
adamgreenfield@social.coop ("Adam Greenfield") wrote:
As someone who regularly helps elderly, low-vision, low-English-literacy & other folks who are struggling to understand them manage their phones (I know @jessamyn does this too), I cannot begin to explain to you the rage I feel at the designers & development organizations who foist these UI choices on a public not remotely equipped to navigate them.
The controls are overwhelming, still. They’re couched in jargon, still. They’re hidden somewhere non-obvious, down the end of a four-node decision tree. Fifteen years after I left the industry things have only gotten worse. And the worst bit of all – as I always insisted then, based on my own extended observation – is that *users internalize the blame for their inability to comprehend the UI*.
They wind up feeling stupid and incompetent. They wind up bewildered and scammed. They feel terrified to hit the wrong button, and then not be able to find it again. It’s so, so hard to watch. It breaks my heart, and it makes me furious.
Mastodon FeedBoosted by glyph ("Glyph"):
nedbat@hachyderm.io ("Ned Batchelder") wrote:
Coverage․py can track what tests ran each line of your code! The HTML report can have a list of tests for each line. Did you know this was possible?
https://coverage.readthedocs.io/en/7.15.4/contexts.html#dynamic-contexts
Mastodon FeedRE: https://tacobelllabs.net/@flamingspork/117077640724510728
I have family in Canada and I have spent a fair amount of time there, and there's a lot of stuff which just feels like "the US, but like you're subscribed to the premium version". But there are subtle differences in the way a few things work that are always a bit surprising for folks who haven't visited before
Mastodon FeedBoosted by glyph ("Glyph"):
dreid@wandering.shop wrote:
RE: https://hachyderm.io/@petrillic/117073813371435539
If not scam, why scam shaped?
Mastodon FeedBoosted by brib@bribstodon.xyz ("brib :neofox_floof: :Nonbinary:"):
soraya.press@bsky.brid.gy ("Soraya Nadia McDonald") wrote:
KEEP BULLYING THEM IT WORKS
Mastodon Feedcstanhope@social.coop ("Your weary 'net denizen") wrote:
Anyway, I don't use my personal feelings to measure productivity. I use my feelings to determine if I'm enjoying something, which is a worthwhile question on its own.
Mastodon Feedcstanhope@social.coop ("Your weary 'net denizen") wrote:
I long ago became skeptical of my peer's self-reports (and my own) of how a new technology made them "more productive" as I watched them become excited about the next shiny new tech thing. Something new would come along and make them *feel* like they were more productive (and often they would push the whole team to adopt whatever it was they were excited by), but by any measure that I could come up with, their work had not measurably improved.
(I obviously can't run rigorous experiments, but there are things you can check via source code control and bug trackers and other available metrics.)
There are things that have helped with quality over the years. Source code control (even the ones before DVCS were better than nothing), static analysis, unit tests, documentation, training. And by helping with quality, we get an indirect boost to productivity because we spend less time tracking down bugs and putting out fires in production that could've been caught earlier.