HNHacker News
TopNewBestAskShowJobs

unholiness

1,306 karma · joined August 20, 2013

submissionscomments
unholiness··on Midjourney Medical
Another problem is that there haven't been natural experiments in low dose exposures the way there unfortunately have been for high dose exposures.
unholiness··on Midjourney Medical
LNT is the null hypothesis. No one disagrees a linear model fits the data very well in high doses. If you want to argue that model doesn't work in low doses, you need a model with more parameters and sufficient data to fit it. The issue is that, at these low doses we want to differentiate, we're also looking at effect sizes that are hard to separate from noise, and sampling biases that are hard to erase. There's still lively and ongoing debate.
unholiness··on Midjourney Medical
So, on the one hand, this is interesting! Reducing radiation from CT scans is a noble cause on its own. If on top of that it could make tomography cheaper and easier, you could imagine getting earlier detection of aneurisms, fibrosis, cirrhosis, thrombosis, stenosis, even plausibly cancerous masses (along with plenty of over-detection).

On the other hand, nothing here substantiates this promise. We've got a video render of what a hypothetical device could look like. It's probably more than nothing (they got exclusive license on these butterfly chips in 2025, and it's at least plausible that the best solution to the data bottleneck in an absurdly noisy system like this is real-time AI image processing)... But it's certainly less than something. It's a hype video that doesn't prove feasibility of anything, yet.

EDIT: This is all in reaction to the second video on the announcement post[0], which is much more informative than anything on the page currently linked.

[0]https://www.midjourney.com/medical/blogpost

unholiness··on Claude Fable 5: mid-tier results on coding tasks
> I will never pay the "normal" API token price for it.

Not until June 22 you won't!

unholiness··on I'm Eric Ries, author of "The Lean Startup" and new book "Incorruptible" – AMA
TBF I think it's just a remark on the upvotes. It's a perfectly cromulent comment with no business being at the top of an AMA.
unholiness··on I'm Eric Ries, author of "The Lean Startup" and new book "Incorruptible" – AMA
> Your first 2 to 3ish paragraph (explanation and rephrasing) is very characteristic of an AI

What you're seeing in those paragraphs is patience and empathy. Let's not let LLMs have a monopoly on these.

unholiness··on Raspberry Pi 5 – 16GB RAM
Zeros are just sold out everywhere through, no?
unholiness··on What it feels like to work with Mythos
Yeah, this made it basically clickbait for me, in terms of time I wasted with the wrong expectation.

The lack of downvotes on posts on HN has always felt like more of a bug than a feature to me.

unholiness··on Claude Fable 5
Subagents are a helluva drug.
unholiness··on Ask HN: What are tools you have made for yourself since the advent of AI?
As the 36 millionth person building something similar, I wish[0] there were better info out there on what works well. I can understand why, but it's still frustrating seeing on the one hand how deeply helpful the flywheel of this type of setup can be, and on the other hand how every blog post stops at some incredibly superficial setup to help them write more blog posts.

[0] Yes this is a plea, if anyone has the good stuff

unholiness··on A Farmer Donated Land to Turn into a Park. The City Is Building a Data Center
The Trustees of Reservations, in the Boston area, are a great example of this working well.
unholiness··on Context Sculpting
If it were, I can in theory see situations where improving content cleanliness is worth blowing away the KV cache.

But I absolutely can't see how feeding the entire context into a more expensive model multiple times per task, just to propose context edits that might indirectly help, could ever be worthwhile.

unholiness··on Debug Project
Gene drives are such amazing and such frightening technology. No one puts them in the same conversation as nukes or engineered pandemics, but they share the same pattern of "technology improvements giving smaller and smaller actors globally reaching powers", and have even more potential for consequences not intended by those actors. It's pretty scary to imagine a world where one lab or one rich farmer has the power to (after a few dozen generations) globally make arbitrary edits the DNA of entire species. Even smart and goodhearted people can screw up that world.

So from this armchair, I'm glad to see that at least for aedes aegypti (which seems like the clearest case for deploying a gene drive), there's an alternative like debug.

unholiness··on Elon Musk has lost his lawsuit against Sam Altman and OpenAI
I'm having a hard time squaring this view with the prediction markets. 500k+ of volume on Polymarket[0] had Elon losing at only even odds until March, 2:1 until May 15th, and 3:1 before the verdict dropped.

It seems to have been widely reported from the start[1] and throughout[2] that the statute of limitations was a key thing Musk's team had to prove. If it was so clear, why did people think this case had legs?

[0] https://polymarket.com/event/will-elon-musk-win-his-case-aga... [1] https://www.reuters.com/legal/litigation/musk-lawsuit-over-o... [2] https://www.nytimes.com/live/2026/05/14/technology/openai-tr...

unholiness··on Googlebook
I just open photos.google.com and grab them. No need to fiddle on my phone.

When on wifi, the photo backup upload starts immediately. If it doesn't (possibly due to your settings, this used to be my issue) you can manually open the photos app and tap the backup now button.

unholiness··on Gmail registration now requires scanning a QR code and sending a text message
https://xkcd.com/277/
unholiness··on Talking to strangers at the gym
I think the answer is practice, for a few reasons. One is obvious: conversation is a skill. Just like a novice chess player can spend 5 seconds figuring out which squares the knight can move to while an intermediate player spots a fork to force trading a strong bishop or exposing an overworked queen, exposure to similar situations rewires your brain to work faster in those situations.

Another reason, though, is to me one of the main benefits of social interaction in the first place: The brain rewiring also makes you think about what other people would think, want to hear, say to you, etc, even when they're not around. That sure can give you better answers in conversations, but more importantly, I think this is just genuinely a nice way for the brain to be. In the same way that dogs are happy playing fetch, humans are happy living with other people in mind. Maybe because it feels like not everything is your responsibility, or that you worry less about what you should be doing, or that you look forward to laughing about disasters later... I'm not entirely sure. Whatever it is, it's nicer than the alternative.

unholiness··on Text-to-CAD
> That's getting solved already in china leading labs, and bottlenecked by the lack of good training data, which china is solving with mass labor.

What work are you referring to here?

unholiness··on You can beat the binary search
More accurately, binary search is optimal only if you cannot determine distance between the data points (you can compute `<` but not `-`). It's inaccurate to say it's optimal if you know "nothing" about the distribution. If you encounter a point that's much closer to your high pivot vs much closer to your low pivot, there is no possible prior knowledge state uninformed enough to conclude that the best place to search in both cases is in the middle.
unholiness··on Who owns the code Claude Code wrote?
> Here is the legal baseline, in plain terms:

This particular AI-ism really encapsulates what annoys me about some AI-isms. I don't mind the delves and the em-dashes that just give away the AI source of what otherwise might be good text. But these structural pieces just feel fundamentally not for the reader. Part of it is blatant pick-me language for the human feedback ("hey look you wanted plain language I did that") and part of it feels like it's just helping the future token stream (thinking-like tokens polluting the actual text).

The not-this-but-that, the sycophancy, the symbolizing-vague-significance, they all have this flavor of serving a process that's no longer there as I now need to read it. It gives a similar sickening feeling to the one I get seeing something designed by committee.

unholiness··on Three constraints before I build anything
It applies more to design software, where a user is creating durable things and needs to understand those things themselves. Google Maps is more of an agent: It's responsible for understanding its own complexity and answering your queries.
unholiness··on It's OK to compare floating-points for equality
I cringed very hard in the slerp example seeing `acos(dot(a,b))`. Clamping to [-1,1] to avoid NaNs still gives you bad answers and numerical sensitivity around small angles. acos and asin in general lose ~half your sig figs around their singularities[0]. Working around this by introducing a threshold seems like exactly same flavor of issue he's complaining about to begin with.

There are perfectly good solutions for finding the angle using atan and the cross product. Calculating A x B will yield a vector which lies is along their normal with length tan(θ)(A · B) so we can straightforwardly say e.g.:

`θ = atan2(norm(A ⅹ B), (A · B))`

No thresholds, no branching, only ~1 bit of significance lost. As the original author says himself, when you're introducing arbitrary-feeling thresholds, it's likely you're missing a solution which would improve more than just the weirdness at those thresholds.

[0] A good analysis of this is on Page 46 here, including a more optimized (but less obviously true) alternative to the formula above: https://people.eecs.berkeley.edu/~wkahan/Mindless.pdf

unholiness··on The Accursèd Alphabetical Clock
I love the fractal nature of this, where the big shape of one two three four... is then roughly repeated both on a slower scale (twenty thirty forty...) and on a narrower scale (twenty-one twenty-two twenty-three twenty-four...).

I'm now wondering the hausdorf dimension of the graph of alphabetical numbers <n, and how other languages might compare.

unholiness··on We're pausing Asimov Press
I discovered them last year on Substack and they quickly became a priority read. A sort of Quanta for biology, taking time to explain enough for a popular audience but keeping technical rigor deep into some fascinating topics.

Some highlights:

https://open.substack.com/pub/cell/p/dna-sequencing?utm_sour...

https://open.substack.com/pub/cell/p/phi80?utm_source=share&...

https://open.substack.com/pub/cell/p/antibody-design?utm_sou...

https://open.substack.com/pub/cell/p/viral-capsids?utm_sourc...

https://open.substack.com/pub/cell/p/legibility-problem?utm_...

...All in the last month! At least they went out with a bang.

unholiness··on Build123d: A Python CAD programming library
Solidworks has VBA macros, which (on top of being poorly documented and unstable) subvert to the whole benefit of parametric CAD. Once you're creating features with a macro, you naturally want to edit them, but you also naturally want to rerun the macro itself to create them differently. It's like editing generated code and it's not a viable long-term setup.

FeatureScript is a different beast. It actually runs as part of regeneration in Onshape. Standard features (extrude, loft...) are also defined in FeatureScript, so your custom features are the same first-class citizens with a interactive GUIs and stable updates to upstream changes. You can freely mix interactive CAD and custom code by adding standard features and custom features.

unholiness··on Claude Code Cheat Sheet
The relevance is that Claude made this cheat sheet.
unholiness··on LLMs predict my coffee
This is true if the milk is in the fridge the whole time. With the milk out the whole time, it's nearly the same, exact answer depends on the geometry of both containers.
unholiness··on The Brand Age
He's definitely conflating two things: watches changed from utilities to fashion items, and watchmakers changed from engineering to branding. In this example, the one change caused the other, but it's just one example. It's no excuse to paint all of branding with the same brush and ignore how, say, Milwaukee's branding provides a valuable signal of real utility.

I think it's tempting to be equally dismissive of branding and fashion because they are both forms of virtue signaling and therefore have all its perversions. However, they're operating on different actors. Branding is virtue signaling by companies to its customers. Fashion is virtue signaling by customers to other people.

unholiness··on Show HN: Linex – A daily challenge: placing pieces on a board that fights back
I like it!

One big annoyance with the power-ups is that the failure condition is checked before you can use them. It's particularly painful since they all replace the current piece, so they seem tailor-made to get rid of a nasty piece that would cause you to lose... but then they have to be used before you get to that piece that would cause you to lose!

Anyway, both times I played I got in the 40s without any power-ups used, then saw that the next piece would cause me to lose but none of the powers could save me. Probably the ideal fix is just to not trigger a loss while you still have powers?

unholiness··on Anthropic drops flagship safety pledge
No, their Responsible Scaling Policy and their government contract are not related. The RSP governs how Anthropic itself behaves w/r/t developing, testing, and releasing new models. The contract was signed with stipulations around how the government can use existing models (No mass surveillance, no military targeting without a human in the loop) which Hegseth wants removed in a standoff that hasn't yet resolved.
← PreviousPage 2 of 11Next →