glaslong

Yeah, if a human didn't write it, I might as well have generated it myself. And it'll be better tailored to the information I'm seeking.

I actually still REALLY value the human authorship though. Not for any sweat and blood nostalgia. I find MY generated docs still have lots of little errors, sometimes pretty critical even, which I can only recognize when I have the expertise and comb through carefully on edit.

The hand written author's name was a stronger guarantee that those wouldn't exist (as much), because you could trust in their expertise, and that they had to put down every word. It's much easier to miss the slop on edit and review than in draft, in my experience.

In general people always seem to want documentation, so they can pull it up exactly when they need it, but will do anything to avoid actually having to read it until that singular moment.

The number of meetings where I can answer all the other attendees' questions because I'm the only one who even skimmed the presenter's docs, I swear.....

Hilariously, I've been finding accidentally poisoning the well with LLM SEO like this works really well at my day job!

I have a toy project that replaces a large part of our systems (disclosure: VR/AR @Meta), so that I can learn by doing, no intent to ship.

But because I have it thoroughly document and tested, and because it touches such a wide swath of adjacent real functionality..... my toy project is CONSTANTLY getting tagged as a dependency and assigning me reviewer on unrelated work. Because when my coworkers yolo Claude at a task, and Claude searches the codebase, it generally finds my crap very well documented. :)

Tech giants upset to find themselves on the short end of the same T&C/EULA stick they've subjected their own customers to for years.

An impressive ouroboros of hypocrisy:

- From treating users like rich ground to be fracked for data.

- To scraping the entirety of the public and private (wherever possible) internet to sell it back.

- To complaining that their competitors do the same to them via distillation.

- Now whining that the few AI companies producing anything genuinely useful (such that they feel compelled to pay for their tokens) are data fracking them right back.

Yup. Model capabilities seem to keep converging quickly, not leaders breaking away for long.

Frontier labs are racing towards SaaS commoditization at incredible speed. And while there might possibly be $Trillions in productivity gained from their use, there's no reason to think those gains get captured by the model makers or inference providers at this point.

Maybe the Claude or ChatGPT desktop apps will dominate as the new MS Excel, but that's hard to do without already having locked the whole market into Windows.

There's virtually no platform play available to them.

Seems like a scam. Call your CC company and issue a chargeback :p

I love this for the popular sci-fi trope too, where you see some ship engineer swap one glowing crystal "compute core" for another.

We could have the photonic AI model ASICs for real!

(Dis)honorable mentions to the current waves of gleeful surveillance state enablement and gambling by any other name, the latter even directly preceded by desperately trying to get everyone's less techy relatives to bag-hold crypto pump n dumps

"Greasing the wheels" seems right in principle, but possibly putting the accelerant factor a bit... mildly. Like going from burning the turkey in the oven, to deep frying and burning your whole house down.

cybercrime losses across the United States rose 26 per cent in a single year

The FBI was candid that even these figures understate the problem. AI attribution in the report reflects only what victims recognised and reported, and most victims of a cloned-voice call never learn that a machine was involved at all.

INTERPOL found that AI-enhanced fraud is roughly four and a half times more profitable than its traditional equivalent, and that so-called agentic AI systems can now autonomously plan and execute entire fraud campaigns, from reconnaissance through to the ransom demand.

The DC3 gap is interesting. I wonder whether they deprecated it because the other EU server had plenty of capacity, or still keep it but only for... "special" account data flow.

Also, it looks like it's easy enough to ID your DC on their API, though I haven't tried it yet (more of a Matrix Stan personally): https://core.telegram.org/method/help.getConfig

Bet on Chinese tech sector to eat everyone's lunch with cheaper, faster, smaller, open-weight models?

That said, True Sickos who love this shit, hit me up, we need to stick together.

It is rare to be able to find a career that satisfies more than one or two quadrants of "Ikigai"

(en.wikipedia.org/wiki/Ikigai)

That is:

• what you can be paid for

• what you are good at

• what the world needs

• what you love

I don't fault people at all for only indexing on a job they're good at, and which pays, until we can figure out this whole post-capitalistic mortal coil thing.

There ARE many fortunate people for whom programming fulfills all 4, but they don't seem to last long before the profession crushes the love of it out of them. :)

Did the HOA Flock cams provide evidence for arrests? How did they help?

Or was this a case where a fake camera to scare people off may have been equivalent, Wile E Coyote style?

Time you have also directly bounded by how alive you are after the event

I'm curious how they could prevent taking them down if the local gov doesn't renew the contract? Presumably they're installed under some works dept land/pole/utility access permits that allow them the space and electrical, which all goes away and requires their removal.

Sorry if this is answered in the pod, don't have time for it immediately.

Directly equivalent to having outsourced it to a shop of remote consultants where you remained as onshore lead. And in all those past cases we would have said "No."

Is this a riddle where you happen to work for the CIA?

oh no, I hope the canal's ok

Luna@max is in a VERY interesting spot if their rankings are at all to be believed:

- Better than Opus4.8 in the coding agent index (doubt)

- Just below sonnet 5, even with glm5.2, in the overall intelligence index

- Cheaper than haiku4.5, glm5.2 and kimi2.6 on cost per intelligence task index

Technically I think there IS another org struct that helps, but it's impractical for a number of Big Corp reasons.

A broad studio-like portfolio, where you mostly stay hands-off and willing to burn a ton of money in directions that won't work, and accept that they are out of your control. Stochastically let the best 10% of isolated plays rise on their own while 90% fail.

I think the VC model tends to do this relatively well. Large orgs though are inevitably culturally, or even legally -- "shareholders! fiscal responsibility! optimal deployment of capital!" -- INCAPABLE of being that intentionally lossy and hands-off.

I have a theory of this based mostly in time working in Meta's RL org on VR and Wearables (disclosure)....

There's a fundamental comms issue, I think, where you can only broadly communicate the least common denominator of what every involved person will understand. The "imperative decrees lacking nuance."

This is doubly hard if the thing is New, so it isn't broadly understood by default in the collective professions of the workforce. For example, we had a HECK of a time getting many web+mobile designers to understand 3D for spatial design like the game designers did. And vise versa, the nuances of building reusable UI and interaction frameworks when maintaining a software platform, for the games folks.

It's virtually impossible (no pun intended) to do something both New and Nuanced with a large group of people. You'll spend all the effort trying to get them to mutually understand a shared, nuanced vision on one hand, and constantly scrambling with their different interpretations producing incompatible parts of work, on the other.

I think you can ONLY really manage it with small groups, and it's a big part of why startups will continue running circles around big orgs, even with shared talent pool.

The only organizational structure that I think mitigates it is having strong top down direction, with vision and expertise at basically every level of the hierarchy. Because it's just taking the small group expert model and delegating it down to expanding levels of detail in execution. Apple seems to be wholly structured around this, and has occasionally done it very well.

What options did they go through before the current cross-industry trend became massive recurring layoffs then I wonder

A related thing that took me a while to accept when I started woodworking is that wood moves, a lot.

If you built the bookshelf in wood, it will be expanding, contracting and shifting over time with temperature and humidity variation throughout the day and season. And asymmetrically depending on the grain.

The straight right angles won't stay that way, and it's better to design such that they change in complementary ways, rather than remain perfect.

Principal-SWE-Bench will take some time to run, because the LLM needs to wait for a crisis to present its solution, having correctly identified that the same solution would have been organizationally impossible to propose until that moment.

Hmmm maybe I'll try this, since I have a newer Frame (which I'd love to get rid of).

I don't like that anything would have to go through their services, but it would be fun to try!

Agreed! Usually this came up because someone wanted to add a popup, but call it a modal to sneak past the bad connotation. When really they shouldn't have been doing either and they knew that; just laundering a UX they knew was crappy.

Mostly through Steve Bannon as far as I'm aware, yes. Though apparently there were connections through moot, and later Peter Thiel adjacency as well.

https://www.garbageday.email/p/here-s-how-epstein-broke-the-...

at least until Flock reports that you're reading a book you bought 2nd hand, and Amazon shuts down your AWS account for the audacity.

There is already typically enough information to contact people's family, employers, or (as was popular / horrific for a moment) send SWAT to your house.

Real names and contact info don't stop people, and worse, they expose everyone to these tactics. If I read a comment of yours that I don't like, I can asymmetrically punish you by signing you up for spam or worse, and you'd never know who did it or why.

We do need solutions where people stake value (like their real identity) and have to contend with real repercussions. But I don't think identity like this actually gets us there.

The reason Old Internet felt better is probably still the solution: smaller online communities, where you have a sustained presence, and getting account banned would meaningfully disconnect you from your social group.

Algorithmic feed / Front Page effect seems to be the main cause of the problem, since you get a flood of people disconnected from the community or topic, having no context, piling on, then leaving.

YouTube seems to have recovered from being a toxic pit and now manages this well. But I haven't found a good writeup of how they made the shift.