I'm building a compiler for a language without a tracing GC, so a big chunk of the work is around memory management: functional in-place update, reuse analysis, and a Perceus-style reference-counting strategy similar to what Koka uses. The hard part was that my use case wasn't exactly covered by the Koka/Perceus paper. The prior art got me maybe 75% of the way there, but the remaining 25% was a cluster of bugs with very similar shapes and no obvious published solution.
With Opus, I kept getting stuck in this loop where it would fix one case, but break another case elsewhere in codegen. We ended up with something like 16 failed experiments just for one bug class. The workflow was: run an experiment, identify the shape of the bug, propose a fix, check whether it emitted the correct Zig, then see if the fix broke any previous memory-management cases. It was useful, but it kept choking on the parts where there wasn't clean prior art to lean on.
Fable was a different story for me. It one-shotted the Class A bug cluster, and then basically said "by the way, your previous attempts have these structural problems." More importantly, it identified the other related bug classes and came up with workable strategies for applying the Perceus-style memory management in those shapes too.
That's obviously anecdotal, and I'm not claiming Fable is universally better. But in my case, this was not a toy frontend wireframe. It was compiler work involving ownership, reuse, RC/drop behavior, and Zig codegen. The thing that surprised me was that Fable seemed better precisely where the problem wasn't just "reproduce known prior art", but required filling in a missing piece.
Also worth noting: I'm not using the API. I'm using the Max plan, so maybe there are product-path differences here. But I definitely did not have the "unpredictable beyond toy-scale" experience. For this particular compiler/memory-management problem, it probably saved me a ridiculous amount of time and money.
'by the way, your previous attempts have these structural problems."
Just to be clear, it did not have access to any previous work that opus did? Because they are pretty good at digging out relevant tmp files and making use of whatever is out there.
With my fable adventures I caught it hallucinating something and stating it as a fact in CLI twice. And it was something that I did not see opus do in such way, opus obviously many times stated some things that it did not verify but guessed, but fable said something like "the probe showed that ..." - but there was no probe, it was not about some past events it was about what it was doing right now. "I overstated"...
But boy does it know Chinese, so much better than any other english model, gemini used to be the king but fable clearly was trained on a decent amount of it. It has a deep cultural understanding.
I maintain a failure registry in the repo. Every failed attempt gets documented with the exact mechanism, the test that regressed, the revert SHA, and an instruction to start from that frontier. Fable read all of it.
But so did Opus.
Each of the 16 Opus failures ran in the same harness with the same accumulating registry. By attempt 15, it had disproofs 1–14 in context. By the end, Opus had basically the same corpus that Fable started with, and it still kept failing, sometimes by re-deriving an already-disproved approach in a slightly different shape.
So “it leveraged the previous work” doesn’t really separate them. Both had the leverage. Only one converted it.
What changed wasn’t more context. It was that Fable rejected a premise inside the context.
The registry’s standing framing was: “this needs whole-program borrow inference, which conflicts with per-module incrementality” (architecturally blocked.) Fable ran around 5 fresh attempts in-session, hit the same wall, and then noticed the framing was a red herring: the borrow analysis already runs module-wide, and for a single-module program, the module is the whole program.
Opus read that same framing for months and treated it as a constraint. Fable falsified it.
its the same repo, same rules, same disproof history, same workflow. The model was the only variable that changed, and the outcome flipped. Is it possible that attempt 17 by Opus could have figured it out? sure. but there's 16 previous attempts that say otherwise.
As fars as anecdotes go, that’s about as controlled as it gets.
Pointing out past suboptimal / failing behaviours to new opus sessions would almost always actually create a sort of "anchoring bias" that would drive the agents towards exhibiting the failure mode (often while mentioning how it wouldn’t fall for it).
As far as I can recall, Fable has been the first model to discover the documented failure modes, comment on them, and just… keep going, actually avoiding them. Quite a surprise.
You should consider doing the hard work yourself here. I sat down and reasoned through a Perceus-style RC mechanism a few years ago, made difficult by the presence of one-shot delimited continuations, and actually sorting it all out was not hard. Handing the correct semantics to Claude will produce the correct results if you take the time to understand the actual work you are attempting.
Similar. I gave it a really hard task, basically messy code in a complex domain that was bug-ridden from a mess previously created half manually and half by Opus. It cleaned things up beautifully, both the backend and the frontend.
Maybe the prompt was particularly well-suited for the model (I instructed it to put on a mathematician's hat, look at the mathematical substructure of the problem, identify invariants and general laws and verify them, then plan how to remediate).
It wrote a ca. 800 line in-depth analysis (at times spawning over 130 research agents...) with remediation plans, prioritized them and then implemented them. One issue was that this document was frankly over my head. Both the language it used and the mathematical parts were very terse, and in parts it felt like a post-C2-vocab exercise. The prose was much harder to understand than the code snippets / data models. As a non-native speaker, it lost me on the prose part, and had to ask it for a less elaborate version to actually understand it.
It burned the session limit four times, but it turned a huge mess of proof-of-concepts with patchy glueing into a coherent, stable application.
I'm also on the Max plan using Claude Code, and I have the feeling that the harness is much more important than the consensus expectation.
> and I have the feeling that the harness is much more important than the consensus expectation.
Is that really the consensus? There’s been a bit of literature lately on that. Can’t find the one about looking into whether or not the harness had a greater impact than the models (for comparable models), but there’s this one: https://arxiv.org/html/2605.23950
Zig is one of the worst targets for LLM generated code. It's nice that Fable has better support for Zig than Opus, but this anecdote is not representative as a general use case.
Slight misunderstanding. The LLM didn't generate Zig. My compiler does.
The model's work was in the Rust compiler internals, specifically the borrow-inference and refcount-insertion passes (Perceus-style ownership analysis). Zig is just the compiler's codegen target, the same way another compiler might emit LLVM IR or C.
The only Zig written by hand is the runtime: allocator code, RC primitives, list/string operations, etc. It's pure Zig, no libc, but it's small, stable, and was mostly untouched during this work.
The model only touched Zig indirectly, by reading the compiler's generated output to verify whether a fix worked. For example: checking that a drop was emitted before a parameter-slot reassignment. That's reading machine-generated code for correctness, not "the LLM writes Zig." Both models handled that part fine.
The 16 failures vs. 1 success were all in the ownership analysis, and that code is Rust.
Really depends of the ergonomics of the language. In erlang/elixir/beam langs etc, its incredibly ergonomic to write code that runs on distributed systems.
you have to try really hard to do the inverse. Java's ergonomics, even with Akka, lends its self to certain design patterns that don't lend itself to writing code for distributed systems.
We have a really talented engineer on our team (in the US), who has a green card and everything. He's taking a job in Brussels, he said very plainly hes not sticking around to find out what happens next. I don't blame him.
If you are able to make it work in Belgium it's a great move. Free education, free healthcare, 20 days PTO minimum, public transport, 15 weeks of maternity leave, labor protections, basically no crime, no guns, no weekly school shootings, total tax rate of at most 60%.
This person is leaving a regime where the physical safety and liberties of immigrants like him/her are in jeopardy, and HN starts a 50+ comment thread about high taxes. Peak commentary.
When we lived in Cal decades ago, taxes weren't all that much lower. Federal taxes and state taxes (higher) on a larger salary, lots of social security and other fiddly little taxes and 100-200/check for 20% of my health insurance. The only good thing about social security is that while you pay a more the benefits are larger 62+. I'm trying to remember, but like 25% for the feds, 8-10% for the state, 6% for social security and the health insurance. Call it high 40's or so, maybe 50%? Yeah, sales tax was lower vs. vat but I thought Cal had about the highest sales tax in the states?
It all depends on what the aggregate deductions that are outside your control sum to, and what you get for the money.
In Ontario, it would be about 38%, and that’d include healthcare. Canada is very efficient though. At least a decade ago, Canada’s non-defense spending per person was less than the US’s.
In Germany it would be about 44% total. Of course, in Germany, $200k is a top 2% income. In California it’s only a top 8% income.
That's what's directly taken out of your check right? But how much more do you pay after that in other taxes? And if you go even further, how much higher are the prices of everything that you purchase due to the various taxes involved in their production?
Does your Germany figure include healthcare and church tax? That could push it over 50%. Though church tax is optional and you can go private for healthcare.
Yes. Apart from the countries which live off of foreign direct investment, taxes are generally pretty high.
Also, in many EU states, companies contribute to social security. In some this is indexed to profits, but on others this is indexed directly to wages, so if you count that bit, taxes directly attributable to your income can easily exceed 60% of what a company pays out.
I don't know if Belgium is using that loophole when counting the 60%, though.
>Apart from the countries which live off of foreign direct investment, taxes are generally pretty high.
I have no idea about this. Can you explain what you mean and give some examples of such countries ?
>Also, in many EU states, companies contribute to social security. In some this is indexed to profits, but on others this is indexed into wages, so if you count that bit, taxes directly attributable to your income can easily exceed 60% of what a company pays out.
True. Some EU countries also tax the gross salary the employer has to give you before it gets to you, which is in bad faith not included in payslips. So when you negotiate your 60k gross wage, it's actually costing your employer something like 72k Euros. I hate this shady practice.
Their inward FDI stock to GDP ratio is around 250%, which is about 4× the EU average; and Ireland does this with a decently sized economy.
And then there's Luxembourg (1400%) and Malta (2000%) which arguably do much “worse” but are comparatively tiny.
I didn't do the math for every EU country. Those were just some of the few that came to mind. For instance, Cyprus has similar values to Ireland, but the Irish economy is 15× bigger.
When there's a lot of foreign money going through your economy and you can tax it to moderate amounts, you get to offer lower rates to your own citizens.
Which is great, but obviously doesn't scale if every country tries to do the same.
> I have no idea about this. Can you explain what you mean and give some examples of such countries ?
Probably countries like Ireland, Montenegro, Belize, etc which act as tax havens for foreign corporations. Or Singapore, while also a tax haven, acts as a center for regional trade.
They could also mean resource rich countries that sell mineral rights to foreign corporations, who made investments in infrastructure in order to facilitate their operations, and they pay back dividends to the state, which offset the tax burden of the local population.
If an American factored in the totality of their tax burden, it would be pretty high. The USA has the benefit of higher incomes and a gigantic population, so there's some economies of scale. But even so, add up all of income tax (federal, state, city, county), sales taxes, property taxes, tariffs, tolls, etc and the % is already pretty high. After factoring the cost of benefits that are free/subsidized in other countries, and the cost probably averages out to the same.
Of course, European countries can also have those same consumption taxes. But I'm not sure if OP factored that in.
> sales taxes, property taxes, tariffs, tolls, etc and the % is already pretty high.
These taxes you mentioned (ignoring income taxes) are even higher in many EU countries than the US, especially sales tax. Same for tolls, tariffs, etc. they're all higher here and they're increasing them and adding new taxes on top, because EU coffers are being bled dry right now with the economy, trade wars, and actual wars going on.
Also, commodity products and services are generally more expensive here than in the US too. Like, I see on youtube the hobby stuff Americans do in their garage with home labs, electronic measuring equipment, power tools and stuff, all gotten nearly for free on craigslist, but if I want to replicate their setups it would cost way more here(from a smaller wage too), not to mention buying a house with a garage in Europe is very much of out of budget to most working class in Europe to begin with.
All this stuff being so cheap and readily available is probably why Americans in their garages have been so much more inventive and entrepreneurial than Europeans.
>After factoring the cost of benefits that are free/subsidized in other countries, and the cost probably averages out to the same.
True, but a lot of free stuff you get back from the government is sometimes of low quality compared to what you pay for in taxes on a high income, due to never being enough money for everything everyone needs, and not being able to attract and keep qualified and motivated workers to stay in the public system when they can earn more privately, and it's only been getting worse and worse since Covid and Ukraine, with no signs of improving.
For example, I am now paying ~1000 Euros for private physiotherapy after my accident, since the free government one is abysmal, which I am forced to pay for anyway out of my salary even though it's useless.
Another example, after my jaw surgery at the public hospital here they just strap cold packs to your face like in WW2, while in the US, my ex-boss who went through a similar procedure at a hospital there they had specialized head cooling devices for your post-op recovery, instead of medieval ice packs, while also being free of charge from his employer insurance. So you might pay more in the US for health insurance, but you also get more in return.
Overall I think I'd still prefer living here than in the US, but there's valid reasons why immigration to the US, and especially the success of immigrants there from an integration and financial perspective, is so high compared to here despite all the issues the US has.
In Brussels? Over half your net income would go to rent. If you are frugal then maybe you can get it to work out. This is not the type of income where you eat out every week.
providing healthcare and education are costs easily overlooked by most americans. But the reality is, these are costs borne by americans as well. and likely at a higher rate: americans pay more per capitia on both of those versus most other nations.
The top marginal tax rate in 2024 (assessed 2025) was 50%. But that's not the total tax take - that's the marginal rate of tax for income above 48320 euro.
This is not easy to compare. When you take into account all the costs up to and after retirement this is another perspective. The cost of life is another factor.
And then of course quality of life, but that's very individual.
Absolutely not. What a depressing country. Soulless people. Literally.
Try Belgium in winter.
And today with the European white guilt it's full of Africans who hate not only their people , but European traditions and culture. This is the reality.
What you listed for the past 10 years in Belgium is an average week in Chicago.
And Chicago had 2853 gun violence incidents in 2024. On a population of 2.7 million. Belgium had 184 incidents on a population of 11.8 million. That is about 67 times more incidents.
>What you listed for the past 10 years in Belgium is an average week in Chicago.
Wait a second friend, first you claim "basically no crime, no guns", then when confronted with the facts, instead of taking accountability and correcting, you move the goalposts to some high-crime US city.
I'm sure Brussels is super safe if you use Mogadishu as the point of comparisons, but if we were to keep the discussion in good faith and stick to comparisons with EU cities, my eastern european city has literally zero crime and guns making Belgium look like a warzone by comparison.
We have literally zero people killed by suicide explosives, guns or machetes compared to Brussels. How can people look at those crimes and go like "yeah, it's not so bad, you only have a relatively small chance of being killed" ?
> Wait a second friend, first you claim "basically no crime, no guns", then when confronted with the facts, instead of taking accountability and correcting, you move the goalposts to some high-crime US city.
OP is right, if those are the worst things to happen in the past 12 years, that's effectively 0 crime.
Especially when you consider that so much of what you listed were actually terrorists attacks conducted by an organization that hasn't conducted a foreign terror attack since winning control of their own territory from foreign occupiers.
>OP is right, if those are the worst things to happen in the past 12 years, that's effectively 0 crime.
If that's "zero crime" from your frame of reference, then what are the cities that have actual zero crime? -1000 crime? NaN?
I'd also be curious to know, if for example you or a family member would have been a victim in one of those violent incidents that don't happen in other EU cities, if you'd still have considered it "zero crime".
Is it one of those cases that when people see so much violent crime it's just a statistic that they had waive it easily? Because I can't.
Then please argument using logic why it's obvious. I explained why it isn't oblivions, as per HN rules.
Subjectively sure, each to his own, it might be obvious to you if you're ideologically aligned with the poster, but for good faith debate, you'll need to add actual arguments to convince the other people of your take. Imagine telling the judge "it's obvious your honor" as your only argument to why you're in the right.
>As a passerby, I'm honestly not sure what pedantic hill you think you're dying on.
No hill dying here, I'm just pushing for facts over blind ideologies.
I meant compared to the US it has basically no crime. Total gun incidents in the US is 10x more than Belgium.
And yes obviously there are guns in Belgian society but with no guns I was referring to how regular people don't walk around with guns. If you play football and your ball enters someone yard you don't have to worry about getting shot.
> I'm sure Brussels is super safe if you use Mogadishu as the point of comparisons
I believe their point was that Brussels is “super safe” compared to Chicago. 67 times fewer gun incidents is quite a lot.
I live in Dublin, Ireland, which is a lot smaller than Brussels, and when there is a shooting it gets on the news. You can imagine how amused I was coming from São Paulo that a full-on gang war was going on when I arrived here and 4 people had been shot in the previous year.
A friend of mine who also came from São Paulo, a trauma surgeon, had to change specialty here because there simply isn’t enough work.
People need to take the name Chicago out of their mouths. If a message board thread is a poker game, bet the bank when someone tries to make a political argument using "Chicago" that they've never set foot here. Someone who grew up in Brussels would be approximately as safe in Chicago as they would anywhere in the United States --- less safe than in Brussels, because of overall automobile and firearms deaths in America, but no less safe than in any major city.
(In fact, your life expectancy in Cook County is several years higher than in the rural south.)
The gun violence in Chicago is tightly constrained to places and populations unfamiliar to the median Belgian. Chicago is a city of neighborhoods and structurally segregated by almost a century of redlining and "urban renewal" that created hyperconcentrated pockets of crime. It's a human tragedy and fully worth dunking on, but it has nothing whatsoever to do with how safe a visitor would be to the city.
(Chicago is also not even in the top 10 in US cities by index crimes, but whatever).
Like I argumented before, comparisons with random high crime cities across the pond are in bad faith, which is why i proposed sticking only to EU cities to make the comparison fair, and Brussels does fairly bad at that level.
If you insist to go this route, you can definitely find cities even in the US with less violent crime than Brussels.
Because it ignores the fact a white European, the vast vast majority of Belgium, won't actually experience that kind of homicide rate in the USA if you pluck them out of Belgium and dumped them there.
If you look at places in US with similar white European demographics (New Hampshire at <2 per 100k) the homicide rate isn't that much worse than Belgium (~1.2 per 100k).
The best predictor of being a victim of violence in the USA is to be black, the ~second best predictor is to be in a state with a high proportion of black people. If you are in a state that is ~as white as Belgium, the rate goes way way down.
Well said. After reading your comment and ruminating on it a bit, I think people handwaive the violent crime in Brussels since native Belgians (and possibly other well off immigrants) aren't very likely to live in Brussels city that's only 22% native Belgian and 78% foreign born.
I expect small cities and towns in the suburbs, where native Belgians are a majority, to have virtually zero violent crime, which would flatten out the crime spike of Brussels into good looking national averages.
>>he said very plainly hes not sticking around to find out what happens next.
Can't blame him, US residency really is a like a game of high attrition. Its your classic up or out scenario.
Some times even a passport doesn't guarantee a stay. Sooner or later, you fall ill, lose a home, have a divorce. Its a unique combination of extreme luck, work, health and many other factors several of which are totally outside of your control.
I agree. Having children does make ones priorities very cut and dry. I found it a lot easier to "adult" once I had children. My Friends, at the time often asked, "Is having children hard?" I often replied, in the beginning at least, "Children are easy, it's everything else that is hard."
Indeed, it is society's expectations that are hard.
I moved to the middle of nowhere after my kids were born. One day I let my child walk home "alone" from school, for the portion that is on our own property, and of course as soon as you do that a fucking Karen will randomly pop out of nowhere, and start interrogating the child. It is like clockwork. You could be 100 miles from civilization and as soon as you do something someone somewhere disagrees with, a fucking Karen (and even in a minivan, down rugged rural dirt roads, how the fuck did she get there?) will magically be there that exact second with a cell phone at the ready to call CPS. Thankfully I was able to stop her before that happened, as I was actually watching from behind the bushes, which in itself is shameful but saved my ass.
I've been using Alpha(as my main driver!) for a year now, there's been a few hiccups here and there but its been very good. I prefer this to Gnome.
It's my main driver for software development, it was initially a dual boot system with windows, but I found that I could use Steam with very little configuration and could do all my gaming in linux(Cosmic DE/PopOS, I have a Nvidia GPU) as well. Works out of the box with Bigwig Studio and my Soundcard (Ultralite mk5)
I use a mix of the Cosmic store and nix for packages and programs.
I don't need to use windows ever again for anything and it makes me very happy.
I live a 20 minute walk away. I never tire of looking at it. When friends come to visit I usually skip the touristy stuff but I will always accompany them to go see the Sagrada Família.
I'm building a compiler for a language without a tracing GC, so a big chunk of the work is around memory management: functional in-place update, reuse analysis, and a Perceus-style reference-counting strategy similar to what Koka uses. The hard part was that my use case wasn't exactly covered by the Koka/Perceus paper. The prior art got me maybe 75% of the way there, but the remaining 25% was a cluster of bugs with very similar shapes and no obvious published solution.
With Opus, I kept getting stuck in this loop where it would fix one case, but break another case elsewhere in codegen. We ended up with something like 16 failed experiments just for one bug class. The workflow was: run an experiment, identify the shape of the bug, propose a fix, check whether it emitted the correct Zig, then see if the fix broke any previous memory-management cases. It was useful, but it kept choking on the parts where there wasn't clean prior art to lean on.
Fable was a different story for me. It one-shotted the Class A bug cluster, and then basically said "by the way, your previous attempts have these structural problems." More importantly, it identified the other related bug classes and came up with workable strategies for applying the Perceus-style memory management in those shapes too.
That's obviously anecdotal, and I'm not claiming Fable is universally better. But in my case, this was not a toy frontend wireframe. It was compiler work involving ownership, reuse, RC/drop behavior, and Zig codegen. The thing that surprised me was that Fable seemed better precisely where the problem wasn't just "reproduce known prior art", but required filling in a missing piece.
Also worth noting: I'm not using the API. I'm using the Max plan, so maybe there are product-path differences here. But I definitely did not have the "unpredictable beyond toy-scale" experience. For this particular compiler/memory-management problem, it probably saved me a ridiculous amount of time and money.
reply