fnordpiglet

I’m going to need a fable reset to explain to me the rules of fable resets and usage. I read it a few times and it’s about as perplexingly convoluted as I’ve ever seen, and that’s including AWS pricing!

It feels like there’s some maximalist move here where you use fable for 50% of your weekly usage then reset it, then use 5.5 for the rest, but I can’t decide if I’ve read it wrong or not.

I much prefer the open ai “it’s raining resets” model for obvious reasons.

Yah I’m also interested in the implications here. Varda is an awesome company and I would have hoped they saw a great exit for such great engineering.

Yes, but my question wasn’t the utility of automated ontology building but the fact this seems to use pretty naive ontological and semantic graph tooling, making it less practically useful than if it used the industry standards of 30 years.

There are plenty of graph databases, and RDF is expressed in triple stores. OWL is just for validation.

I’ve used AWS Neptune as a pretty reasonable managed sparql query engine but for most my needs actually just local jsonld mastered in git is plenty for almost APO knowledge graphs I have.

I’ve found LLMs are exceptionally proficient at authoring sparql, and quite clever. This isn’t surprising given there’s plenty of sparql in the corpus, and it’s semantics aren’t HUGELY different than SQL. Likewise they do fine writing datalog etc.

I found LLMs are pretty good at extracting ontologies through a) researching ontologies that exist for adjacent fields, there are a lot published after decades of work, b) doing multi pass ontological classifications using existing ontologies as baselines and refining, c) given ontologies attempting to represent knowledge in them and querying them along various paths with agents tasked with building recipes and techniques and evaluating them across synthetic benchmarks developed along side.

There’s also a lot of books on how to develop ontologies as well as online resources. They are all in the corpus and tasking a frontier pro level model with building a process and mental model document with references inline for grounding works really well as well for priming.

I’m struggling to see why this is better than OWL/RDF and SparQL. Given the breadth and depth of semantic web knowledge in the corpus they do an amazing job discovering well established ontologies and building new ones on presented data and querying it with sparql. In fact better than almost anything I’ve seen. What an irony that the semantic web after all these years is likely to find a real use - it’s original one in fact.

Are there a lot of software jobs inland? Moving to Montana is nice from a cost of living POV but unless your goal is to become a rancher or miner, there’s not a hugely booming software engineering market.

I think it’s completely inverted actually. The big companies have a much smaller moat now. This is almost certainly the best time to do a startup. You can build a decent product very fast, GTM in a month or so, and pivot rapidly. But it requires a lot of skill to do it with AI tooling in a way that sustainable and the techniques and tools are very immature and shifting around rapidly. Large companies are struggling very hard to integrate these into their flows and workforce while a small company, with the right small set of people, can achieve as much or more than a large one with thousands of developers.

People will take food without paying for it before dying. You don’t have to forage when you can loot grocery stores. At a point when a large percentage are going to die and the food is there, it will be bedlam and the police won’t be stopping it because their families will be part of the riot.

The notion that people will simply roll over and literally die for profit margins is absurd. The system will break and be rebuilt in that situation.

I actually feel like it’s really premature for a relatively new technology to be showing profit improvements. We are talking on the order of a few years of investment and “figuring out” how to use it productively in business. Most new technologies take a decade or longer to really yield tangible value.

This is the first cycle in my (relatively) long career where the demand for return felt instant. The tools and techniques are almost comically immature and the average person interfacing with the tools is largely bewildered.

For word processing for instance people spent years upon years coming up to speed on wordstar, word perfect, word. Decades later they’re a common part of life and the tangible benefit over secretarial pools is obvious - there is no such thing as a secretarial pool.

AI it feels like almost a blink of an eye and the demand is “well, it’s been a few years, where’a the post scarcity society I was promised?”

Slightly different take - use the Claude / OpenAI / muse desktop navigation integration to actually go into your YouTube account and like all the videos and watch them for a bit to seed your algorithm.

See Conways law for how to mitigate this

He’s never achieved a majority vote in any election, FWIW. His claim of a mandate was just as brazen as all the other litany of lies - he’s a lame duck lame duck. The true shame is on the congressional republicans for letting him sully them so deeply and follow him in his self immolation of the entire government and America in the world. The rebuilding will be a major test of the successor, and it’ll be fascinating to see if the Goldwater republicans can become the party of reason again rather than this reality TV idiocracy.

And won via a plurality after losing the popular vote in the last election. The least popular president in history can’t win a majority of popular vote. It’s not shocking he’s deeply unpopular, he never was popular. He’s a won twice on a technicality of the republic created to ensure the rural states wouldn’t be dominated by Virginia in the original establishment of the union.

He’s a pariah everywhere on earth, most especially at home. The sad part is the democrats couldn’t mount a successful campaign against someone so unfit to lead. Now there’s nothing left to do but clean up the mess and undo the terrible harm he has caused and repair what can be once he’s finally out of our hair.

I think no one wants this personally, but there is so much money - and employment - behind the private pay system that it’s virtually impossible to reform it. There is also a real aspect that the system was designed by those in power to further a necessity to work, an overriding fear that people might idle or not work with maximum productivity. Once the system was laid down. With dozens of huge insurance companies and even more smaller ones, with billing departments, coding clerks, adjusters, claim processors, duplicated at every clinic, hospital, insurance company, and the constellation of other roles, partners, providers, it’s now employing millions of people just to operate this hideous machine of national suffering. Talk of reforming it into a single payer system or a socialized system is wildly unpopular with everyone who depends on the current system - millions upon millions of people, and deep spending companies supporting candidates. In a country where either politicians either win by a landslide due to partisan politics, in which case it doesn’t matter what constituents think, or the margins of victory are so narrow that you can’t afford to screw over millions and millions of people through direct action like legislation, the system is metastasizing. There are solutions to the health care problem from a consumer point of view, but the churn it would cause on the employment side would be profound and structural. My view is probably the best strategy is make money young here and emigrate to a place worth living, which is painful to say.

There’s very little I can see truly breaking American society but I think the stark stupidity of health care here is up on the top.

I’ve been in the industry a long time and benefits at FAANG are not great. Deductibles are high, subsidy pretty average, overall package weak. Earlier in my career I paid nothing for insurance, had a lot of fringe benefits, and I was genuinely grateful for the perks. I’ve looked at ACA alternatives compared to my FAANG benefits and it wasn’t a huge difference frankly - better on the margin but nothing to write home about, at least in my local market.

The fact people have to find a job not for money but for health insurance is about as dystopian as I can imagine. Toil just to not get sick and go broke, not because you need the money. What a sad world we built for ourselves.

As much as I appreciate your intention might be to be helpful it was in fact quite rude and arrogant. I think telling someone in a technical forum to not be so technical and lecturing them on clarity of communication is pretty offensive - you don’t know me, know what I do or who I am, what my background is, and clearly my message wasn’t intended for your consumption since you found it not for your consumption. Some advice since we are dishing it out, this sort of stuff is as annoying as correcting grammar on Usenet posts or all the other forms of things that disingenuously begin with “not to be rude” - if you are aware it is likely to be read as rude, maybe just not type it next time? Just trying to help you be effective and not show such gross colors next time.

PS, it’s about builds on the same tree but about edits on the same set of files in the tree that would cause block COWs in brtfs. If you read the original links docs you will see that’s the use case they’re targeting, and it’s a lot different than build artifacts. Their proposal isn’t just assisting in deduping but in merging rapidly files that have interleaved changes.

It’s like almost yesterday it was CRDTs and Rust!

Perhaps but it wouldn’t have been explicit about why I have that problem. I don’t think it’s a normal problem to have, whether it’s elite or not, it’s IMO not common. The speculative pipeline and the merge train is the key part because they overlap with changes in the same files. This is why block based doesn’t help me as much. I’d prefer to not censor myself with respect to my problem statements for some internet humility virtue signaling.

I have this pattern but because I run pipelined agents working on a speculative merge topologically sorted task list merging into a merge train. This leads to lots and lots of work trees that largely overlap except for slight variants. Their articulated use case is mine, and I can have up to a hundred work trees at a time. The storage and related back pressures can be complicated.

It is more content aware and tuned towards versioned data vs block aware among other things. There’s a benchmark as to how it impacts things in the site

https://layerfs.ai/benchmarks/dsh-dedup/

I think it’s actually a fairly useful thing but I worry about something so obviously vibe coded hitting 1.0.0 and without a track record but the ideas seem pretty useful. I’ve certainly found the work tree semantics most agents and harnesses follow at scale to be obnoxious at best and it seems to tackle issues head on.

My involvement in inpatient drug abuse clinics has been benzo addiction is one of the hardest to treat and accounted for nearly half the patients, the rest split between alcohol and opioids, even during the height of the oxy crisis. It’s much less reported on because it’s widely prescribed and it’s nearly impossible to OD on. The effects aren’t as dramatic as alcohol, but are no less debilitating. Furthermore because benzos and alcohol have similar pathways they are often abused together and they amplify each others effect. You can satisfy the addiction with either - so people withdrawing from alcohol are often given benzos and people addicted to benzos often drink to relieve their addiction. People describe klonopin and similar long lasting benzos as a six pack that lasts 18 hours - which is a long time to have than level of declined inhibition. This further leads to more drinking, but also to other out of character behavior like overspending - some patients have bankrupted their families through excessive and bizarre spending.

Many people receive opioid treatment and do not become addicted. In fact most do. Most people can drink alcohol without becoming an alcoholic. This isn’t a proof that “anecdotal experience” that opioid addiction and alcoholism aren’t serious problems. In fact it is a demonstration that they are insidious for that exact reason - it’s fairly easy to slip through the cracks in the system into a serious problem that is life altering and extraordinarily difficult to emerge from.

Finally, the hardest withdrawals I’ve seen are benzo withdrawals. Relapse is also extremely high because the profound anxiety that led to the addiction is amplified in the years of subsequent withdrawal with very little as effective as a benzo to mitigate it. Paralyzing irrational fear is in many ways worse than pain, and pain is often a symptom to boot.

I feel that perhaps saying “well not everyone, take my anecdotal experience” isn’t the most empathetic response, and the statistics bear this out. The prevalence of benzodiazepine dependency is roughly on par with alcoholism in the US and several times that of opioids. The distinction is that crisis level addiction is roughly the same as opioid, while alcohol is roughly 7x worse. This likely because acute alcoholic behavior is rapid and very readily observable, while opioid and benzo dependency can be masked behind an otherwise normalish appearing life. The decline is fairly slow often unless you trip over into fentanyl. Further the prevalence of common use of both opioids and alcohol is much higher because surgery and recreational alcohol use is fairly prevalent while anxiety disorders less so.

Stevie Nicks is a popular icon who has been very public about how hard her withdrawal from Klonopin was, including how her skin sloughed off among other horrible effects. She lost a large part of her life span to it. But otherwise it’s also a relatively under reported dependency given its prevalence.

https://time.com/3482462/rolling-stone-stevie-nicks-drugs-mu...

I’d also note in many countries benzodiazepines are highly controlled with much higher bars than US, including the UK, where they’re only allowed in crisis, not general anxiety disorders less.

And they do it in powershell

IMO the most useful thing a mathematician can do is not solve “practical problems” but to come up with conjectures and riddles that look around several dark corners of math at once. The proving of a conjecture is a bit of a sport done to impress people with your cleverness; but the clever person was the one who made the conjecture to begin with.

Given the trajectory of LLMs to date I see them helping construct proofs for conjectures but not devising new conjectures of any material substance any time soon. The challenge is, as Tao has said elsewhere, part of being able to formulate a conjecture is by exploring the landscape around it by doing proof research. Seeking answers finds useful questions. The fear is the tools will take away the exploration of the landscape. But that’s silly IMO. It takes a special type of lazy to let a tool take away your understanding that’s important for what you’re goal is, and a special type of passion to learn with a powerful tool. The passionate will adapt and progress will accelerate, not slow down.

For instance Tao has posted some of his interrogations of proofs with AI assisted reasoning and it didn’t seem to diminish his ability to establish insight rapidly, but to shut down avenues rapidly and to pivot and expand his understanding rapidly. The insight was all his own, as was the path of inquiry. The understanding of the output was his as well. He carries it into tomorrow and uses it to formulate new more interesting conjectures the next day. The AI ceased to “exist” as soon as the inference turn was over.

The hand wringing is pointless and lame. John Henry moves aren’t necessary or useful, and when I see a tunnel boring project now I see tons of people well employed there, just the holes are bigger and better, and they’re engineers instead of strong backed hammer slingers.

And seen the disproportionate use of military style force to run down naturalized citizens in the parking lot with tanks and tear gas?

I for one welcome our magnetic millirobot overlords into my beating heart.

It would almost be as big a monstrosity as Amazon

Wait, we’ve already given away our privacy. Where’s our crime free society?

I used to go to GHC to recruit every year on their floor, I enjoyed it greatly and hired a lot of bright people from it. It’s sad to see it so humbled. There was to my knowledge nothing quite like it. I hope it can recover.

I think honestly they should focus on their core competency. Microsoft’s biggest problem is the total lack of focus. GitHub is just a great example of that total lack of coherent direction or product vision. I hope Anthropic can avoid becoming everything and nothing at once as Microsoft has done.