bpavuk

I wonder if it's somehow related to me having to re-sign into my account after clearing Google Play Services data just a few hours ago.

"dots" are something that should work over long horizons of time, no? they are powered by Astra and still not good enough. Mercury, though, is all about making the same unit of intelligence a bit cheaper and a lot faster, which is not exactly what an always-on unpredictable non-deterministic text autocomplete needs.

I'd quit. and I did, even before getting my first job that was not an acquaintance's startup, not counting this summer. in fact, I got fired this summer for having called Claude Code "a decent search engine" afther the first two weeks. I fell in love with making code run fast on poor hardware back when I was 8, and now there is a high likelihood that I'll end up a truck driver.

my only response to those who remain and try to control the slop is

  Praise the Omnissiah!

and really, you should have an Imperium/Mechanicus-level devotion (whether it's to money or to AI itself) to survive this and not burn out. in my region, though, you are not getting money - salaries were slashed in half across the board over last two years.

https://t.me/djinni_official/2163 (go in with translator; infographics are in English but explanations are not)

at this point, for a junior, it's easier to find a well-paying blue-collar job than to get into tech, and the ceiling is falling at the speed of sound.

love these guys. instead of spending time on something akin to those creepy "dots" or obnoxious @Claude in Slack, they ship actual improvements to speed. you can't imagine what kinds of UX a fast LLM can enable! and for coding I have always been a proponent of strong compile-time type systems and handing off only the actual boilerplate typing, all to maximize actual thinking and speed of high-precision prototyping.

I actually built a small prototype that integrates with JetBrains IDEs - it tracks your in-IDE activity (kinda like JetBrains' own "file-scale tab-completion" feature) but you can invoke Codex with all that context at any moment to let back-then GPT-5.4-nano finish the miniature task. say, you wrote a new use case and your goal is to expose it through a new endpoint, but it's kinda boilerplate-y and DRY is not your style either. then you open a needed file, press a shortcut, tell what you want, and you get it. JetBrains Air kinda absorbed that functionality, Codex tokens are hella expensive (don't get me started on API costs), and JetBrains is ready to front that cost through "Junie Lite" (which is currently Qwen 3.6 27B), so I decided to abandon that prototype. but I do miss the combination of speed and control over resulting code.

I may revisit this idea with Mercury 3, or ask JB to integrate it.

putin has objectively caused a lot of harm, and desire for carnage is understandable, yet if we are to break this cycle, we should make the format of Nuremberg trials the new norm and make it so that even if we want to, we can't start a new war. unity around hatred is fleeting, temporary, and often leads to overcorrection, such as me being harassed for talking to my mother in Russian (I also fluently speak in Ukrainian and tell them go fuck themselves using all the regional swears I collected all over Ukraine) or Israeli having to apologize themselves for Netanyahu's decisions.

and everyday matters? they are boring and important, and if someone dares to touch the supply chain, they should face all the economic repercussions imaginable. ideally, this will make punishment for war immediate, inherent, and structural. and if we take it further into pony land, we may deprecate the military because its purpose will be carried by Brenda sitting behind a desk with Excel (or LibreOffice Calc) before her. :)

idk if there are fellow Ukrainians reading it, but I remember getting an account somewhere in exchange for an AFU donation. just a screenshot from your bank app confirming that the payment is within a few days of sending a registration application, and you are in. that tracker also had decent English originals somewhere, and a friend who's there tells that experience is great, but for foreigners he'd advise to turn on Google Translate or whatever.

the point is, some trackers don't even take these one-time payments for themselves and still provide service on the level as if you pay tenfold of that, monthly, to them.

tried https://archive.org, no dice. I should develop a habit of snapshotting every reply I make.

btw, there is a Donate button. just saying :)

I lived close to Donetsk until 2022 and I totally support the "ghoulish" label. we GOTTA find the way to stop this cycle of violence and unify around something. so far, practical everyday matters such as coal and steel, or silicon, have been pretty successful at keeping another world war at bay.

> surrender to the other side (Russia I believe?) and spend 1-2 years [...] citizenship proceedings [...]

I almost read it as an insult, but I'll assume that you either forgot or didn't know one peculiar fact: Russian army almost never keeps their POWs alive, and even if it does, it absolutely never lets them out of the army into Russia - only back into Ukraine, through the battlefield.

this would be a valid option if I was a Russian who does not want to fight Ukraine, sure. Ukraine is ± alright with Russian POWs, partly because we have Europe watching over our shoulders. the reasoning behind your idea is sound, assuming that both sides wage an honest and Geneva-accurate war, except that Russia prefers to utilize the Genevan convention as a piece of toilet paper.

so, your idea effectively becomes suicidal, but you also get a great deal of suffering. why not cut the suffering while I can?

agree. yet, most of the time, most people have enforcement infra, and having an intent to kill, and not resisting it, is ghoulish. that way, one is perpetuating the very cycle that Europe attempted to stop within itself through coal and steel

on a serious note, it really is. death/execution as a punishment makes no sense. there are fates worse than death for those whose offenses we'd love to spill the blood for. Alcatraz is one such place where those fates are served.

and for crimes that are not enough to execute for, bog-standard prisons are enough. they are imperfect, but those willing to change prove themselves there, and a good one should give enough opportunities for a criminal to re-educate, reconsider some of their life choices, and rehabilitate.

punishment by death is never the right fit.

18. same ass of a country (to outsiders, have a look at all the countless debacles concerning corruption to understand why "ass"), except that I have 7 mostly worry-free years left to live and the market for juniors (and thus, me) is closed. if I get drafted, I'm waiting out for an opportunity to shoot my head.

honestly, having a decently paying blue-collar job made me think whether tech is worth it at all, given that I'll lose all the extra benefits (from canteen to proper bomb shelter to gym to commute) and cut my already tight paycheck by 20-25%. for what? for "tech"? where I have to be able to wake up at 3AM when the client's LLM took down their prod, commute to a vulnerable office in Kyiv (and pay for expensive AND equally vulnerable housing there), live under a threat of getting fired because I was "not AI enough" or whatever management wants, or because I don't have gigabit Wi-Fi in the shelter? fuck it, I'll rather be a night-shift loader and idk, learn embedded at home.

the world devolving into proompting has ruined tech for me. prolly I'll get a Cat. C driver's license and spend little what I have remaining of life on the road. if I get lucky, I may eventually reach CE and Euro 95 certs, and that in turn gives me some sort of safety against mobilization.

if the war ends before I become 25, I'll likely end up a stereotypical himbo truck driver. I'm tired of having to perform the smartness.

oh you can if you are younger than 23 and not in education. this exception initially started as "academic mobility" program - you had to somehow prove your intent to study before going abroad, I don't remember how - but then both govt and citizens realized that it's too much paperwork and impossible to organize properly, so the exception was generalized.

yeah, hating papers is our national trait at this point, given Diia and stuff XD

I guess it's France with its Qwant engine. decent for coding and French (culture, language, or both) queries. didn't use it for much else tho

there is no difference because all four will train on your data anyway :D

I miss o3 in that regard. if GPT-4o led people into virtual romance and over-validation, o3 gave me the same sort of "madness" but with work.

when I tried GPT-5, I was sad mostly because GPT-5 had some added wordiness. fluff, if you will. o3, though, is a black hole you can talk to, and it only gave information back if there was something to give back. that kind of vibe is my dream coworker

I tend to agree, but I also should highlight how expensive this shit really is.

in one month, Google actually went cash-negative. [0] even still, they are subsidizing their stuff a lot less, have the most opaque and variable limits, and increase adoption through bundling and shuffling features. I can't even share my Google One storage without subscribing to a Google AI plan anymore, but previously any plan except Google One Lite was shareable.

if you tell me that's not enough to go after frontier, then how much money are Anthropic and OpenAI burning?

[0]: https://www.techspot.com/news/113214-google-records-first-ne...

how the hell do we even track that? and before someone says...

—"Benchmarks!"

...I'll tell that they can be gamed so easily, and they are on a consistent basis.

where did they mention Bitcoin? much like everything "first generation", Bitcoin is objectively garbage nowadays, when we have continuously evolving Ethereum for modern all-around-good experience and Monero for actually private payments. oh by the way, cryptographically significant quantum processors are eventually coming, and Bitcoin still didn't make up its mind on PQC, to my knowledge, whereas Ethereum did.

no one writes a fresh C99 project in a 2000s Visual Studio with whatever formatting was the most trendy back then anymore.

you have more chances to shine only because everyone around you is not shining, but instead competing on volume

this.

crypto has a lot of theoretical value, but in practice people only really think of crypto when they want to tap into shadow economy.

I do not deny that this technology will be instrumental to network communities, provided we are ever allowed to form such. I do not deny that offline POSes and fully offline tap-to-pay may become a possibility if an existing fiat currency adopts crypto as one "interface" among traditional banking, Apple/Google Pay, and cash. (this, by the way, would be incredible in Ukraine - right now it's power off, and you can't use your bank for offline purchases.)

in practice it's only really shadow economy. at this point it's not even a speculative asset of some notable value

> some underpaid military man [...] abuses AI [...] hallucinates that nukes are incoming

here it is: https://arstechnica.com/ai/2026/09/report-us-almost-boarded-...

> To be fair this is a fairly low risk, but it's _a_ risk.

it's much higher than you think, at least during Trump admin. have you seen how LLM agents are being shoved down our throats at tech companies? the government is no better in that regard, there are all incentives to skip output verification

a year or two ago, I'd enjoy this comment as some usual HN satire. nowadays, especially here, I do see a lot of people that actually believe this, enough to make me think whether to laugh

same, switched from Scratch to C at age 8, but I only really wanted my programs go fast back then :)

> The question is, how do we prevent China and India from just taking our research and copying it for free

centralize all chemistry research behind the closed doors, weapon-level export controls on U.S.-made fever drugs simply because they are U.S.-made, severely control international collaboration, and so much more.

the problem is that the scientific community will rightfully revolt and the rest of the world will pull ahead of the U.S. in a relatively short amount of time. plus, since open science will have been long dead in the U.S. under such conditions, random breakthroughs at a basement or a non-prestigious university will cease to exist as a concept.

China and India piggybacking is the least bad thing to happen. rather, it's an indicator of how open the open science in the U.S. is

and Europe. Europe should also decouple itself from the U.S. as much as it is possible, and not repeat the mistake of overreliance with China. not because China is somehow worse, but because you always should be able to say "no", no matter how good your partners are

TIL that "Muphry's law" is a deliberate misspelling: https://en.wikipedia.org/wiki/Muphry%27s_law

well, yeah, now I understand that. whoever has the money dictates the kind of 1984 you are going to have

this was literally the reason pointed out in chat. devs liked how I worked, had zero questions on review, and that's despite me seeing their tech stack second time in my entire life

I consider myself a high-quality junior, rivaling my senior industry peers on flexibility, systems thinking, low-level engineering, and understanding of user experience. yet all I found this year was a place I got fired from 2 weeks after getting hired. the reason? I called Claude Code "a decent search engine" on a meeting. perhaps I would have been fired anyway because the heatwave was building in Ukraine and I had absolutely nothing but a basement (where I can't move my desktop to) to cool off.