HNHacker News
TopNewBestAskShowJobs

afiodorov

784 karma · joined June 21, 2019

Likes Computer Science, Mathematics, AI, ML
submissionscomments
afiodorov··on Is being bilingual good for your brain?
When you spend some time transcribing live, impromptu speech, you'll notice that it often doesn't follow the rules of written grammar; speakers frequently abandon sentences midway through.

For example, in the linked clip[^1], the speaker says:

  "uh the European Union uh that's not a US creation that's a you guys creation so don't ex..[abandoned word] the strength of the west [abandoned sentence] and the west is a really I don't know what"
For a moment, she struggles to express herself. Yet, there's a qualitative difference between not knowing what to say because a thought is not fully formed, and knowing what you want to say but realizing you've forgotten the specific word you need. For instance, you might be about to say "cherry," only to find you've forgotten the word and instead say something more general, like "forest fruit (fruta de bosque)," which is still correct but less precise.

[^1]: https://youtu.be/_hBd8w-Hlm4?si=7-kvpUoeYo5ODPiI&t=787

afiodorov··on Is being bilingual good for your brain?
The phonetic similarity between Russian and Spanish is a huge relief. As a Russian speaker, pronouncing English has always felt like a workout for my mouth; the sounds are completely alien. Spanish, on the other hand, is effortless. It just flows, since I'm using the same phonetic toolkit I grew up with.
afiodorov··on Is being bilingual good for your brain?
There's another level after fluency (C1), which is near-native fluency (C2). At the level of such mastery you don't feel the need to simplify just to be understood, your utterances now define the language itself as you've achieved the level of the crowd whom the language belongs to in the first place.

P.S. I've typed this out in English after having achieved such unlock.

afiodorov··on Is being bilingual good for your brain?
I use my third language, Spanish, every day, and my second, English, for work. On top of that, my partner is a native Portuguese speaker, so I'm passively soaking up a fourth. (I usually reply to her in Spanish, but we watch everything in Portuguese—though this month it's been all Italian, just for fun).

To this day, I still find Spanish a bit more challenging than my native language or even English. I think it's because even though I moved to Spain over seven years ago, I never fully immersed myself in the culture. I'm pretty sure I haven't read a single book in Spanish.

I still do that classic thing non-fluent speakers do: I'll get halfway through a sentence, realize I don't know a specific word, and have to rephrase my thought more simply. To be clear, I'm far from a beginner, just not yet fluent.

Anyway, I can attest that grappling with a language you haven't quite mastered is a daily mini-puzzle that definitely keeps the brain working a bit harder than it otherwise would.

On a side note, I love that LLMs can handle so many languages now. After 17 years of living abroad, I still feel most at ease speaking my native language, Russian, even though my vocabulary is a bit lacking these days for more complex topics. It makes me completely understand why people prefer to receive medical care in their native tongue.

afiodorov··on Andrej Karpathy's YC AI SUS talk on the future of the industry
I am not saying rewrites are always warranted, but I think LLMs change the cost-benefit balance considerably.
afiodorov··on Andrej Karpathy's YC AI SUS talk on the future of the industry
> So, it was really fascinating that I had the menu gem basically demo working on my laptop in a few hours, and then it took me a week because I was trying to make it do it

Reminds me of work where I spend more time figuring out how to run repos than actually modifying code. A lot of my work is focused on figuring out the development environment and deployment process - all with very locked down permissions.

I do think LLMs are likely to change industry considerably, as LLM-guided rewrites are sometimes easier than adding a new feature or fixing a bug - especially if the rewrite is something into more LLM-friendly (i.e., a popular framework). Each rewrite makes the code further Claude-codeable or Cursor-codeable; ready to iterate even faster.

afiodorov··on It's the end of observability as we know it (and I feel fine)
I like this reply because it shows the shifting paradigm of hireability. Years ago, software engineers used to be hired just for their coding talent and would be put up with even if they were arrogant and had poor hygiene. With time, we all rebranded as problem-solvers, now required to be good communicators and team players in addition to writing good code. The next evolution is that we will be the ones responsible for a certain area, and how we get it done will be abstracted away.
afiodorov··on If you are useful, it doesn't mean you are valued
The article's 'useful vs. valued' distinction is spot on, and for me, it largely hinges on perceived replaceability.

Being a diligent workhorse makes you 'useful' - you're reliably closing tickets. But 'valued' often means bringing innovation or strategic foresight that's harder to replicate.

Sometimes, too much visible grind on routine tasks can almost cap your perceived value. It reminds me of the senior dev whose Jira updates might be terse — like 'Continuing research on core problem' — for days. They're not judged on daily ticket volume, but on the eventual breakthrough or critical insight that unblocks everyone or defines the next big thing. That signals a different kind of leverage and indispensability than just high output.

afiodorov··on Avoiding skill atrophy in the age of AI
Regarding the common analogy about GPS atrophying map skills, I have a slightly different take based on observation and experience. My dad, who learned to drive pre-GPS, struggles to simultaneously drive and follow navigation – it's too much input, too fast. He needs a co-pilot or pre-planning.

For those of us who learned to drive with GPS, however, it wasn't simply about foregoing maps. It was about developing the distinct skill of processing navigation prompts while simultaneously managing the primary task of driving. This integration required practice; like many, I took plenty of wrong roundabout exits before it became second nature. Indeed, this combined skill is arguably so fundamental now that driving professionally without the ability to effectively follow GPS might be disqualifying – it's hard to imagine any modern taxi or ride-share company hiring someone who lacks this capability. So, rather than deskilling, this technology has effectively raised the bar, adding a complex, necessary layer to the definition of a competent driver today.

I see a parallel with AI and programming. The focus is often on what might be lost, but I think we should also recognise the new skill emerging: effectively guiding, interpreting, and integrating AI into the development process. It's not just 'programming' anymore, it's 'programming-with-AI', and mastering that interaction is the next challenge.

afiodorov··on Vibe Coding is not an excuse for low-quality work
NASA* :D
afiodorov··on Vibe Coding is not an excuse for low-quality work
Running AI itself is not low effort though, far from it at current efficiency levels it burns way more energy to produce outputs than humans. Setting latest advanced models to think about the problem on your behalf doesn't necessarily make it low effort overall.

Vibe coding is first and foremost moving some of your problems to the AI and that can be suffice even in professional setting where there's a lot of mundane requests, e.g. setting up slack alerts, running simple data transforms or figuring out one-off data analysis. Maintainability is less of an issue on tasks where it can be done from scratch if need be.

If your situation is result-driven and easily verifiable AI-driven solutions are absolutely appropriate.

afiodorov··on Vibe Coding is not an excuse for low-quality work
I think it is a losing battle. People are energy preserving creatures and we skip paying attention if we can. Because paying attention is effort. Vibe coding is exactly this, low effort development and thus is enticing. Now if we can get away with low effort why shouldn’t we? I am not advocating serious NATO missions to be vibe coded for trajectories, no. When billions are at stake no way. But what if billions are not at stake? What if nothing is? This weekend I added a way to maximise my QApplication app. I just asked Claude code to do it and tested it. It worked. That is all I need for my own app that I use. It’s not that I don’t have any other users but works on my machine is the motto of this specific free and open source app.
afiodorov··on Claude Code: Best practices for agentic coding
I agree but optimisation often reveals implementation details helping to understand limits of current tech more. It might not be worth the time but part of engineering is optimisation and another part is deep understanding of tech. It is sometimes worth optimising anyway if you want to take the engineering discipline to the next level within yourself.

I myself didn’t think about not running linters however it makes obvious sense now and gives me the insight about how Claude Code works allowing me to use this insight in related engineering work.

afiodorov··on Everyone knows your location: tracking myself down through in-app ads
I wonder if Apple will extend their iCloud+ private relay service to apps as a result of this.
afiodorov··on Understanding gRPC, OpenAPI and REST and when to use them in API design (2020)
I’ve generally regarded gRPC as a high-performance protocol mainly suited for connecting microservices—something you’d keep internal rather than expose publicly. But it shines in use cases like live captioning, where a transcription service has to stay in sync with a video feed and can’t afford dropped messages. In my experience, using plain WebSockets for high-throughput internal communication was a mistake because while WebSockets use TCP underneath, they don’t inherently handle reconnection or message acknowledgments. With gRPC, those features come built-in, saving you from implementing them yourself.
afiodorov··on Ask HN: Organize local communities without Facebook?
Why don’t you build it? Just the features people need of course. Seems like the kind of thing LLMs are quite good at (giving you prototypes).
afiodorov··on Ask HN: How can I realistically change careers?
Overall, I like the sentiment. However, there’s a common pitfall: as experts transition into decision-making roles, they often rely on their older technical knowledge. Over time, this once-valuable expertise can work against them, because it’s based on a previous generation of technology.

Many people assume that excelling at a role automatically qualifies them to lead, believing firsthand experience is enough. Yet as the gap between how things are actually done and how they think they’re done widens, their decisions can become increasingly detached and counterproductive.

afiodorov··on Can LLMs write better code if you keep asking them to “write better code”?
All I need is the proportion of the qualifying numbers to the input array to run the algorithm and the number of samples. Then we can sample min, max index of the qualifying array and return their difference without having to sample many times if we can derive the joined min max distribution conditional on the Bernoulli.

In other words the procedure can take any input array and qualifying criteria.

The joint distribution is relatively simple to derive. (This is related to the fact that min, max of continuous uniform on 0, 1 are Beta distributions.)

afiodorov··on Can LLMs write better code if you keep asking them to “write better code”?
I claim we can do O(1) complexity (minus precompute) in all cases, see another comment of mine. Curious if O1 will figure it out.
afiodorov··on Can LLMs write better code if you keep asking them to “write better code”?

  a = [int(x) for x in str(n)][::-1]
  assert n == sum(d*(10**i) for i, d in enumerate(a))

Now when you're operating mod 9, 10 == 1 % 9, thus

  10**i == 1 % 9
Comes from the fact that

  (a*b) % 9 == (a % 9) * (b % 9)
Now using

  (a+b) % 9 == (a % 9) + (b % 9)
We get that that sum(a) and n are same mod 9.
afiodorov··on Can LLMs write better code if you keep asking them to “write better code”?
This gave me an idea that we can skip the whole pass over the million draws by noting that the count of draws landing in my precomputed set M (digits-sum=30) is Binomial(n=1mln, p=|M|/100k). Then we sample that count X. If X=0, the difference is not defined. Otherwise, we can directly draw (min,max) from the correct joint distribution of indices (like you’d get if you actually did X draws in M). Finally we return M[max] - M[min]. It’s O(1) at runtime (ignoring the offline step of listing all numbers whose digits sum to 30).
afiodorov··on Can LLMs write better code if you keep asking them to “write better code”?
Another speed-up is to skip the sum of digits check if n % 9 != 30 % 9. Sum of digits have the same remainder divided by 9 as the number. This rules out 8/9 = 88% candidates.
afiodorov··on Where can you go in Europe by train in 8h?
Vilnius Riga connection missing
afiodorov··on Engineers do not get to make startup mistakes when they build ledgers
Blockchains are examples worth learning from about how to build ledgers. Append-only transactions which start as pending and then eventually confirmed. Balances are cumsums of confirmed transactions.
afiodorov··on Lies we tell ourselves to keep using Golang (2022)
Not sure why Go is compared to Rust all the time, whilst most appropriate comparison is Java.
afiodorov··on Show HN: IMDb SQL Best Movie Finder
This query appears to work as well

https://www.imdb-sql.com/?query=SELECT+*+EXCLUDE+%28titleTyp...

afiodorov··on IronCalc – Open-Source Spreadsheet Engine
I am quite excited about the new era of WASM webapps - those are cross-platform & fast. Unlike traditional desktop programs those are sand-boxed & unlike mobile apps the distribution channels are open ("side-loading" by default).
afiodorov··on Show HN: IMDb SQL Best Movie Finder
Yes
afiodorov··on Show HN: IMDb SQL Best Movie Finder
done :)
afiodorov··on Show HN: IMDb SQL Best Movie Finder
deduped all rows with a simple .uniq() call in polars before saving
← PreviousPage 2 of 5Next →