HNHacker News
TopNewBestAskShowJobs

everforward

3,011 karma · joined December 10, 2021

submissionscomments
everforward··on GLM 5.2 Is Out
The Anthropic news is demonstrating much the same; fall in line or eat export controls.

There was a time I would have agreed with you, but these days even as an American I fail to see a difference. China is probably less likely to try to disenfranchise or imprison me, to be honest.

everforward··on /architect: Reduce Fable tokens by 80%, Fable orchestrates/reviews, Codex builds
Most LLM tasks can be expressed as a DAG, but the odds of it succeeding go way, way up if you drop the acyclic requirement (eg a “run tests, if they fail, fix it and loop back to running the tests” stage).

And it gets delegated to context because it’s either to have another session and tell it to double check and critique the first LLM than it is to write a deterministic test for every prompt. Like if I want a new form that sends a REST request on submit, I can have two LLMs duking it out in 5 minutes. If I have to write Selenium tests then I might as well just write the feature. Or I can have an LLM write the tests, but that’s more or less the same as letting a second LLM judge the first.

everforward··on Noise infusion banned from statistical products published by Census Bureau
Census data provides a reliable source to build off of, which makes joining between data sets more reliable. A lot of what you're talking about would be partial prints of an identity that have to be joined up with others to give reliable data.

Eg

> Cell tower data

That's just going to get you a subscriber and device ID, unless you're talking about going deep packet inspection and parsing the contents of the packets. You could, but that's a lot of effort to get something the census can hand you for free.

> credit bureau integration

Notoriously unreliable and identities for the purpose of credit get stolen constantly. The easiest way to clean that is against known-good info, like the census.

> social media scraping

Half the profiles are fake, also not reliable data unless you clean it up. Again, census data makes it very easy to cut out profiles that don't match a real person.

> tax, payroll

These are probably fairly reliable, although they usually won't tell you about a person's demographics.

> passport, visa, medicare/medicaid, immigrations and customs databases

There's an enormous part of the population that won't appear in these at all. The huge part of the country that's "working poor" but not poor enough for Medicaid probably aren't traveling internationally. I wouldn't be surprised if half the country doesn't appear in any of these.

The census has value in that it contains a huge depth of information, is tied with your identity, citizens are compelled by law to answer so even the privacy folks have to respond and lying on it is a crime (enforcement is probably non-existent, though).

I'm sure that can all be reconstructed to some level of accuracy given sufficient effort, but that's a lot harder and requires a ton more coordination than "SELECT * FROM census_data WHERE ..."

everforward··on Show HN: Paca – Lightweight Jira alternative for human-AI collaboration
> I'm not convinced companies always need software tailored to their workflows, and could benefit from adopting worn-path workflows instead.

I’m dubious, because for an established company the question is whether the software adapts to the org, or if the org adapts to the software. It’s a lot harder to change the workflow of a whole company than to buy software that enables your current workflow. There’s months of retraining and figuring out where compliance goes in the new workflow, and things that get done wrong along the way because it’s new, and etc.

You need a pretty big efficiency win to offset the dead weight of time spent just changing workflows.

everforward··on /architect: Reduce Fable tokens by 80%, Fable orchestrates/reviews, Codex builds
Agents are in a wacky state, which makes projects like this fall into a weird spot. Eg I vaguely expect my agent to do two disparate things: manage dependency injection for tools, prompt modifications, etc, but also be the sort of “brain trust” that controls the flow of execution (can we stop now, do we keep going, etc).

This project is meant to be the latter, but there’s not a clean way to integrate that into Claude Code or Codex because they expect to do both.

Pi can do it, but then your users can’t use their Claude subscriptions, so you have to cludgily try to do the same thing via LLM prompts.

everforward··on πFS
It does, but only vaguely unless you already know how it works and can work backwards to Newton's laws. Eg Newtonian mechanics can explain how flying works, but if you don't already know then it's hard to go from Newton's 3 laws to a functional explanation of why planes don't fall out of the sky.

Some of that is also the domain. It's less that science is an extreme form of compression, and more that natural phenomenon are highly compressible. They're a small number of kinds of interactions repeated a bajillion times. How many equations does it take to explain electricity (ignoring equations that are derivatives of ones already included)? I think it's less than 5.

On some level, you could probably reduce all of the Standard Model down to models of atoms, their motion, and the basic subatomic particles (the non-quantum ones). That would explain almost everything that happens on Earth in a very short form, though few people would be able to go from that to explaining how lightning works.

everforward··on Cheap Iranian drone downed $25M US Army helicopter–maybe by chance
Assembly lines are much faster but less flexible, which is probably a bad fit for something evolving as quickly as drone weapons.

A basement lab can start making new drones as soon as you can get them the components. An assembly line will take months to get any new machinery it needs, set it up, ensure production works, figure out how to QA it at scale, etc.

It doesn’t matter if you can make 8 million drones if the enemy already has a way to counter that particular kind of drone. Eg I think Ukraine has been through a bunch of iterations as they adapt to Russian tank armor, jammers, extended range, etc.

everforward··on Cheap Iranian drone downed $25M US Army helicopter–maybe by chance
This is all wishy washy without an actual court, but I’m dubious this would rise to that standard.

From what I can tell, international law requires attacks to be distinct and proportional. Distinct meaning they distinguish between civilian and military targets (you cannot intentionally attack civilian targets), and proportional meaning that the collateral damage to civilians is proportional to the military value of the strike.

This probably doesn’t meet that bar. Cutting off water is bad, but 20k people is a fairly small population for this kind of thing and presumably the US won’t stop them from repairing it (that probably would be a clear cut war crime). Presuming Irans government is functioning, this should be extremely hard for those people but probably not lethal. Trucking water around is an option, even if it’s not great.

everforward··on Dopamine Fracking
It doesn’t. I also think the fake truffle tastes weirdly like gasoline, the real ones don’t at all.

They are dramatically more expensive, though. Last time I had it, I think I paid a $50 upcharge to have it added to my pasta. Might have been more, it’s been a couple years.

everforward··on Judge Learns Both Sides Used AI, Cancels Trial, Kicks Everyone Off the Case
That part actually seems possible, and I’m surprised it didn’t work. Last I looked, legal citations had an exacting format I would think you could parse out and check if the case exists at least. Won’t work if the AI hallucinates the citation format though, I guess.

You could maybe even verify quotes if the citation was attached (treat ellipsis as a regex “.*” and check the case text). That part would probably be imperfect, but I’d think useful at least.

everforward··on Ask HN: Are you still using a Vision Pro?
I tried a bit on the original Vive and text was awful. I didn’t get nauseated, but games didn’t make me sick either (I did get some “sea legs” when I took it off).

Haven’t tried on the AVP, I think it has way better displays than the OG Vive did.

everforward··on Show HN: Performative-UI – A react component library of design tropes
I think they sort of do, it’s a form of signaling.

It’s the same game we play verbally with slang. Slang word gets made up, people use it so others know they’re in the group that uses the slang, usage spreads until it’s no longer a group indicator. You see it all the time as an age grouping. You can almost guess someone’s age by the slang they use for “good” (cool, lit, bussin, etc).

This is the same. Startups invent a new UI style to separate themselves from the incumbents, incumbents eventually copy the style, cycle repeats.

everforward··on Life is too short for a slow terminal
This is going to vary wildly by what you do and how much CLI you use.

I’m an SRE and the answer is “constantly”. I get pulled in a lot of directions, it’s way easier to maintain context if I open a new terminal or tab for each thing.

Joe asks about something, I open a terminal. Teammate asks about something, new terminal. Joe replies, I swap back to his terminal to look at the scrollback buffer.

I’m closer to you when doing more dev work. One tab running a watcher for builds and restarting the app, one tab to run tests or whatever, a couple for poking around if I need to grep or curl or whatever.

everforward··on A Matter Wi-Fi Light Bulb in Rust on the Raspberry Pi Pico 2 W
That likely increases your latency by a fair bit. Most of my devices take a second-ish to connect, configure the WiFi stack and start sending data.

I’m not sure I’d want a full second of latency on button presses. I would have figured Amazon used BTLE to the Alexa and used it as a BT to WiFi gateway

everforward··on A Farmer Donated Land to Turn into a Park. The City Is Building a Data Center
The guy donated it, so they didn’t buy it per se.

This is purely from a legal perspective. Morally it’s abhorrent (at least as presented).

everforward··on A Farmer Donated Land to Turn into a Park. The City Is Building a Data Center
Go look at their page for “covenant”, because “real covenant” is a subtype that only specifies the ways it’s different from a non-property covenant https://www.law.cornell.edu/wex/covenant

Quoting from that page:

“The party capable of enforcing the covenant depends on whether the burden or the benefit runs with the land. In other words, only the party who the covenant is designed to help can enforce it.”

Your page spells out the other relevant bits. Real covenants must benefit one party at the expense of another (horizontal privity), so the heirs of the man who donated the land are the benefactors. That it helps (or at least doesn’t harm) the neighbor does not make them the benefactor here because their benefit was incidental (ie they aren’t legally “the benefactor”).

Covenants being centrally registered is a matter of convenience when house shopping, not a declaration that the state will enforce them.

I’d actually bet there are a lot of houses that have racial segregation covenants on them still because the benefactors quit trying to enforce them. I know my city has a bunch of racist laws on the books still because the city quit enforcing them ages ago, city council doesn’t want to spend time revoking laws that haven’t been used in 50 years, and no one has standing to sue to revoke them unless they get arrested for them.

everforward··on Sam Bankman-Fried applies for a pardon from Trump
From what I recall last time I looked into it, sort of by current legal theory.

A pardon does not automatically require a confession (though it could be part of the terms), however blanket pardons remove the ability to plead the 5th on the topic if subpoenaed as a witness. The 5th precludes being forced to testify against yourself, but if the government has declared it not a crime then you can’t testify against yourself even if you wanted to.

From what I recall, non-global pardons still maintained some level of 5th Amendment protections. Ie you were pardoned for A, but talking about A might reveal you did B, so you could still potentially testify against yourself.

Iirc, non-global pardons usually come with a stipulation like “must testify that or about…” because of this.

everforward··on A Farmer Donated Land to Turn into a Park. The City Is Building a Data Center
> If it is something that could have been objected to by the original seller (they would have had standing to sue) and they have not agreed to change the covenant (because they are dead), it seems as if anyone affected should be able to sue.

They don’t because it’s a private agreement, so only the involved parties can sue. In this case, if the original seller died then standing to sue would be inherited (I believe). If the inheritor doesn’t care, then neither does the government.

There’s also a bunch of weird edges. Like if the land just isn’t usable as a park because it’s too out of the way to be worth maintenance or building it would be insane because it’s a literal swamp, what is the city supposed to do with that? Own it and just do nothing with it in perpetuity?

> If there is a covenant that says that the land will be a park, that's the "zoning" by the seller being stricter than local zoning, which means that it also conforms to local zoning.

Sellers don’t get to do any zoning, the city does. You can add a covenant that says a single family home in San Francisco can only be used for fracking, despite the fact that there’s no oil and zoning wouldn’t allow it.

> I can sue Facebook if they allow another user to violate their terms of service.

No, you can’t. Or rather, you can file it, but it will be tossed out immediately. There is no tort for failing to enforce your own ToS. You might be able to sue Facebook for negligently failing to stop a user from breaking an actual law.

It’s against Facebook ToS to use a name other than your legal name on your account. How confident are you that you could win a lawsuit against Facebook because Post Malone’s account isn’t named “Austin Post”?

everforward··on Anti-social: It's fads, not friends, which now dominate social media feeds
> It seems bizarre to me that you only limit yourself to these people. I regularly try to meet up with people I don't know super well but want to get them or their city better. Social media has absolutely helped facilitate this.

It’s bizarre to me that you can’t find enough people locally to be inundated in events (unless you live somewhere remote, which is an aspect of social media I hadn’t considered).

I don’t even consider myself particularly social and I’m inundated with events and people I need to text because it’s been a while. I had to cut back because it was too much. Magic on 3 separate work nights with different groups, an event every Sunday with the locals from the bar, a family event most Saturday’s and friends if not.

And then trying to weave in the new acquaintances into existing stuff, because I’m a lush and 3 beers in I’m everybody’s friend and am setting up a grill out with a stranger to see if his jackfruit tacos actually taste like chicken so I can tell if he’s just a vegetarian or a vegetarian _and_ a liar!

There are something like a couple million people within a half hour drive of me, I really don’t have to use Instagram to find someone doing cool stuff around me.

everforward··on A Farmer Donated Land to Turn into a Park. The City Is Building a Data Center
It would generally be the opposite, what law gives them standing to sue?

My knowledge as a non-lawyer generally agrees with above, most states won’t allow you to sue for neighbors doing something legal that decreases your property value (CA is the exception I’m aware of, and even then it’s a “sometimes” kind of thing).

I’m not even sure who they’d sue. Presuming the land is zoned for a datacenter, the datacenter is allowed to do datacenter things. You could sue the city to try to prevent the zoning, but sovereign immunity would preclude suing them for doing their job (zoning, in this case).

everforward··on Harness engineering: Leveraging Codex in an agent-first world
Oh there absolutely are, I don’t mean to imply any certainty in that attribution.

Quality of data is totally one. Immutability may be another (it’s easier to reason about if you don’t have to track mutations to a variable). Another interesting one is Clojures emphasis on composability using basic primitives that are sort of hard to grok initially but unlock really cool stuff.

You can do some incredible stuff with recursive map and arrow functions in a few dozen characters.

everforward··on Anti-social: It's fads, not friends, which now dominate social media feeds
This is bizarre to me, because anyone I know well enough to link up with on a trip also knows where I live and vice versa.

That’s just a really odd relationship to me. Maybe it’s a social media thing.

FWIW, I hear things like this but have never heard of any of my friends that use social media actually doing it. In the same way that you could use an Emmy as hammer, but nobody does.

everforward··on Anti-social: It's fads, not friends, which now dominate social media feeds
There’s also an end to the feed. I forget the limit, but at one point I think I hit the maximum number of pages at like 50 (quality drops pretty dramatically, by like page 10 there was almost nothing interesting or notable).
everforward··on Harness engineering: Leveraging Codex in an agent-first world
Seconded on smaller files. I feel like I tend to get better responses faster.

A notable flaw here is that I’ve not tried large vs small files in a large codebase. Most of my experimentation there has been on personal projects where even a small file contains a significant part of the project. I could see degradation when it has to load 5 files to figure out how something works.

Total LOC (tokens, really, literal lines probably don’t matter) is interesting as a factor. That might go some way towards explaining why LLMs are weirdly good at Clojure.

Eg last I checked Anthropics one-shot performance on Clojure was about the same as Python or Go despite almost certainly being less represented in training data. The combination of density and simple primitives might be easier for an LLM to wrangle, ameliorating the impact of a less popular language.

everforward··on The perils of UUID primary keys in SQLite
This has edge cases that are infrequent but messy because it depends presumes timezone changes are transitive across time and they often aren’t.

Eg I save a date 18 months in the future in US Western time at 0600. 3 months from now the US gets rid of Daylight Savings. You saved it as 1400 UTC (which was correct), which is now an hour off because the local time zone is different.

Encoding it in local time fixes that, because it delays the TZ conversion until you need it and your times stay correct as long as you update tzdb like once a month.

everforward··on S&P 500 rejects SpaceX, also blocking entry for OpenAI and Anthropic
Broad strokes, you would expect to see a withdrawal of funds from the market by anyone following equal weight funds. New entrant means you have to pull money out of existing stocks to re-allocate to the new entrant to maintain equal weights.

It would be visible at a macro level, you’d see a higher sell volume and probably a drop in price as all the equal weight funds rebalance.

There’s a heavy motivation to be the first mover here, because those sells will cause a supply spike and price drop. By being the first mover, you can rebalance before prices drop.

I don’t have sell volume data, but we didn’t see price drops so either a) the market did not believe and no one rebalanced, or b) few funds rebalanced, and the other funds disbelieved enough that they thought the risk premium was so small they could buy at a slight discount and profit, balancing supply and demand.

everforward··on S&P 500 rejects SpaceX, also blocking entry for OpenAI and Anthropic
“Average” is also doing terrible things there. The “average GPU” is probably the integrated graphics on the CPU of a laptop.

If you scoped it to “average gaming desktop”, double digit VRAM is pretty normal at this point. If costs came down, I imagine the higher end GPUs would start including enough VRAM for 30B-ish models.

everforward··on S&P 500 rejects SpaceX, also blocking entry for OpenAI and Anthropic
Have they even talked about how they’re going to handle the heat?

I’m willing to believe the construction is plausible (maybe not cost-effective, but possible), and robotic operation of it is a stretch but with some optimism I could be convinced.

I don’t see how they can get rid of the heat, at least in a realistic way that isn’t “a square kilometer of black body radiation surfaces” or something equally “works on paper, so uneconomical it will never actually happen”.

everforward··on Failing grades soar with AI usage, dwindling math skills in Berkeley CS classes
My SO does the same and it saves my butt sometimes, but I don't like it. It feels like I'm not really solving the issue and just making it her problem instead of mine, but she has a less strong response to the interruptions.

She's certainly better at it though. I can't manage the interruptions, nor am I good about remembering to create calendar events and what not when events come up.

everforward··on Failing grades soar with AI usage, dwindling math skills in Berkeley CS classes
I think some people are okay with communication that’s less involved. Like meme-y BSing where everyone involved knows everyone else is putting like 12% of their thinking power into sending a response.

I don’t really enjoy that, so I find having that many threads stressful and annoying.

I just take a hard line and will unilaterally downgrade communications (while politely letting the other party know). I have all my family group chats muted because my mom uses “Send” the way you’d use Enter on a desktop. End of a sentence? Send text. Next bullet point in a list? Send text.

I muted the chats and told her that I want my ringer on in case there’s an emergency, but I got 30 something notifications in 5 minutes during an interview and it’s unfair to the candidate or other people in the meeting. Internally I rationalize it as revoking someone’s ability to make noises on my phone at whim. They can still text me, they just can’t interrupt me anymore.

It helps a lot, even if only temporary. I’ve muted people for a few hours or a couple days before when I’m already stressed and they’re really chatty.

← PreviousPage 8 of 34Next →