Hacker News: vajrabum

New comment by vajrabum in "Squillions: How money laundering won"

vajrabum — Tue, 02 Jun 2026 16:23:52 +0000

I don't understand what is meant by "hardly possible with crypto." I've been hearing that for a long time but I've also heard so many stories about theft of crypto assets and seizures by government. In the US that seems to have slowed to near nothing with the current administration but it doesn't seem to be a capability issue. Or maybe I misunderstand you or the situation?

New comment by vajrabum in "Texas woman arrested for Facebook post about town water quality"

vajrabum — Sat, 23 May 2026 20:59:14 +0000

Make the police officer like the Doctor pay for their own insurance.

New comment by vajrabum in "I don't think AI will make your processes go faster"

vajrabum — Mon, 18 May 2026 02:53:02 +0000

I'd love to hear of any currently available commerical C compiler which has that level of issues. I would bet you'll be hard pressed to find one. C compilation is a quite thoroughly solved problem. In any case please provide an example.

New comment by vajrabum in "I moved my digital stack to Europe"

vajrabum — Wed, 13 May 2026 15:59:12 +0000

I wish I believed that they have to go to the FISA court for much of anything any more. Instead they go to Palantir and the like which simply buy the data and aggregate it. Very similar to the process of money laundering. And for the data that can't be bought there's the five eyes work around.

New comment by vajrabum in "Alzheimer's disease mortality among taxi and ambulance drivers (2024)"

vajrabum — Sun, 29 Mar 2026 13:57:44 +0000

p is not p-hacking. p-hacking is when instead of making a hypothesis and then looking for a correlation instead you look at the data and see if there are any correlations to be found. The problem with the second is if there are lots of possibilities it's much more likely that you will find a spurious correlation. That happens because the sample distribution is not the population distribution. And in fact in complicated data sets it's more likely that if you look at everything that you'll find spurious correlations than not. In social sciences these days people often register their hypothesis before running the study to prevent p-hacking and reduce the possibility of spurious correlation.

New comment by vajrabum in "Why so many control rooms were seafoam green (2025)"

vajrabum — Fri, 27 Mar 2026 17:55:22 +0000

Uh, I think they didn't use WW2 surplus anything when they build Oak Ridge and Hanford before the end of WW2. I also think given that those two plants were key bits of the Manhattan Project that they didn't cheap out on anything. And the color dude was in fact hired by DuPont who built those two plants and adopted his theories because they increased measured factory productivity and safety. Lastly the OG dude in the article was not an engineer. In fact he was an art school dropout who was very interested in color. So no on the no.

New comment by vajrabum in "Claude Code is being dumbed down?"

vajrabum — Wed, 11 Feb 2026 20:40:30 +0000

Or is this PM and executive management aiming for the no and low code users? That would fit the zeitgeist especially in the tech C level and their sales pitch to non-tech C levels.

New comment by vajrabum in "A lot of population numbers are fake"

vajrabum — Thu, 29 Jan 2026 16:45:04 +0000

I suspect they'll just replace the old recommendations with new ones.

New comment by vajrabum in "Reminiscences of a Stock Operator (1923)"

vajrabum — Thu, 01 Jan 2026 04:45:43 +0000

Supposedly this is a roman a clef about Jesse Livermore's career. There's a lot of stuff in this book that makes sense of markets in ways that pretty much no other investing book I've ever read does. Some what I remember are bucket shops, tape sense, marketing campaigns for new stocks, risk of ruin (Livermore went bust over and over), and what amounts to compulsive gambling.

New comment by vajrabum in "Backpropagation is a leaky abstraction (2016)"

vajrabum — Sun, 02 Nov 2025 17:18:02 +0000

None of this is about an end user in the sense of the user of an LLM. This is aimed at the prospective user of a training framework which implements backpropagation at a high level of abstraction. As such it draws attention to training problems which arise inside the black box in order to motivate learning what is inside that box. There aren't any ML engineers who shouldn't know all about single layer perceptrons I think, and that makes for a nice analogy to real life issues in using SGD and backpropagation for ML training.

New comment by vajrabum in "How well does the money laundering control system work?"

vajrabum — Thu, 21 Aug 2025 22:38:23 +0000

C'mon, moving money is not a crime but moving money that has been illegally obtained is a crime (drugs, prostitution, illegal sports book,...) as is concealing the source or target of money sent to terrorist organizations and yes it makes certain things harder than they used to be for the rest of us.

New comment by vajrabum in "GPT-5: Overdue, overhyped and underwhelming. And that's not the worst of it"

vajrabum — Sun, 10 Aug 2025 00:56:41 +0000

And even if it's not locked down hard how do you separate the signal from the noise with all the ai generated blah blah blah.

New comment by vajrabum in "Achieving 10,000x training data reduction with high-fidelity labels"

vajrabum — Fri, 08 Aug 2025 01:18:38 +0000

Not quite the same thing but some non-negligable percentage of ads I see on Facebook are outright scams which purport to be selling musical instruments at a 'markdown'. First guitars supposedly from the Sam Ash bankruptcy sales linking to an obvious fake site and more lately 'free' giveaways of high end Gibson acoustic guitars. When I've reported them I got the feedback that it didn't violate community standards, but my insta account got perma-banned when I posted the original of a song on youtube from 1928 on a thread which started with a cover from 30 years ago. That was considered spam.

New comment by vajrabum in "Huawei's star AI model was built on burnout and plagiarism"

vajrabum — Tue, 15 Jul 2025 23:22:09 +0000

I'm pretty sure I've seen that openai was complaining about one of their customers doing exactly that a couple of years ago. Yes, here https://aibusiness.com/nlp/tiktok-parent-allegedly-used-open...

New comment by vajrabum in "Anthropic, Google, OpenAI and XAI Granted Up to $200M from Defense Department"

vajrabum — Tue, 15 Jul 2025 01:31:15 +0000

I know this is likely in the pipeline anyway and maybey not covered by this news but now we have the prospect of agentic llms hallucinating enemies and a digital finger on the trigger.

New comment by vajrabum in "Amazon CEO says AI agents will soon reduce company's corporate workforce"

vajrabum — Sun, 13 Jul 2025 20:06:27 +0000

American capitalism in the 2020s is no such thing. It's goosing this quarters numbers so the management can get their incentive bonuses and stock and buying buying business advantage from the legislature.

It's all in Adam Smith and economic history.

New comment by vajrabum in "Andrew Ng: Building Faster with AI [video]"

vajrabum — Sat, 12 Jul 2025 00:13:49 +0000

The platforms I've seen live on top of kubernetes so I'm afraid it is possible. nvidia-docker, all the cuda libraries and drivers, nccl, vllm,... Large scale distributed training and inference are complicated beasties and the orchestration for them is too.

New comment by vajrabum in "Surprises in Logic (2016)"

vajrabum — Tue, 22 Apr 2025 17:43:56 +0000

I would have thought that the proof shows this problem is unavoidable. How in your view do we avoid this problem?

New comment by vajrabum in "Three Felonies a Day (2013)"

vajrabum — Sun, 20 Apr 2025 15:34:41 +0000

The feds don’t generally indict unless it’s a slam dunk. That means two things. Lots of people walk and they will go after someone if they have the evidence. As a small investor I’m glad they get at least a few. Insider trading is a kind of stealing where the victims don’t generally even know they’ve been robbed and at the high end the amounts are large.

New comment by vajrabum in "Ask HN: Is Washington Post correct in saying Signal is unsecure?"

vajrabum — Wed, 26 Mar 2025 16:26:23 +0000

There are laws about this sort of thing that have severe penalties attached. When I was in the Navy handling encryption gear I had to sign a paper that stated that I understood that compromise of the secrets I'd been entrusted with could lead to the death penalty. Are you saying that shouldn't be true? Or shouldn't be true for people above a certain level?

Are you claiming that Signal running on consumer iPhone and Android devices where Pegasys and 0-days are for sale is secure?

Are you claiming that it's secure to conduct classified business on a platform where you can add anyone to the conversation without the appropriate documented approvals?