HNHacker News
TopNewBestAskShowJobs

ziofill

2,071 karma · joined August 8, 2017

Principal Quantum Architecture Scientist at Xanadu

meet.hn/city/ca-Kelowna

Socials: - github.com/ziofil

Interests: Science, Startups, Technology, Research, Programming, Philosophy, Jazz, Chess, AI

---

submissionscomments
ziofill··on Universal health coverage could save $1T and 114k lives a year: study
It should never be shopping in the first place. I live in Canada, my family has needed the hospital and the ER a few times and we never ever had to even reach for our wallet.
ziofill··on Universal health coverage could save $1T and 114k lives a year: study
No, overwhelmingly more people support a single payer healthcare system. The issue is that the people in Washington won’t make it a “votable” decision.
ziofill··on Qwen 3.8 27B
How much time did you invest in creating this benchmark? Any recommendations/resources you could give on how to do it?
ziofill··on Spaghettifying DRAM
But it supercharges what can be done once you get root, no?
ziofill··on DeepSeek Harness developer preview
Exactly. “Tell me more about X” is an open ended question, not feedback.
ziofill··on DeepSeek Harness developer preview
Sounds very cool! What do you mean by “revert side effects it created”?
ziofill··on Worms: The Future of Yesterday's Worms Today
Haaalleluja … BOOOOM!!
ziofill··on Compression is prediction
Compression and error correction also go hand in hand: in compressed data every bit carries more information and therefore errors are more detrimental. This is one of the results that Shannon phrased exactly in terms of entropy.

My PhD supervisor had a beautiful example. Take the English message "errors can make messages unreadable" and ‘compress it’ by removing vowels. It’s still intelligible because English has redundancy:

rrrs cn mk mssgs nrdbl

Similarly, an uncompressed message with errors (swapped characters) is also intelligible because of the redundancy of English:

erwurs lan nake wesaagis unfeatable

But now we do both: we compress the message AND add errors. The result should be much harder (if not impossible) to read:

rwrs ln nk wssgs nftbl

ziofill··on How Blackwing Pencils are Made [video]
Yes something like that (around 50€)
ziofill··on Stealing Reasoning Traces from Proprietary LLM APIs
I understand it’s cool to have an artistic website, but it’s very noisy and non-accessible.

But very interesting result.

ziofill··on Muse Glimmer: 30B-parameter model optimized for always-on local agent workflows
> And I think the data center buildout will end in carnage.

Isn’t it more likely that they will still all be used to the max? I don’t see how at any rate we will be going “alright, that’s enough intelligence for now”

ziofill··on How Blackwing Pencils are Made [video]
About a decade ago I bought a metal mechanical pencil (rotring 800, 0.5mm). Best pencil I’ve ever owned. It will last so long I’ll leave it to my children.
ziofill··on Everything you do is being recorded
It’s not that wild: governments are made of self-serving people, and money from the industry finds its way into their pockets.
ziofill··on How I use LLMs to learn complex topics
It depends what you mean by “learn”. To have a superficial idea of the high level concepts of a step-by-step process, sure, a 3d visualization like the ones this author makes can work. But it stops there.
ziofill··on Seedance 2.5
I acknowledge (and agree) with the sentiments against slop. But should we also entertain the possibility that AI content might become more engaging and entertaining than human content? To me this is far scarier and concerning than a flood of mid-level crap.
ziofill··on Truth is not a direction: a Tarski attack on LLM probes
You are correct, but this is a made up example, not a practical matter. You’re falling into the same trap as the author.
ziofill··on Truth is not a direction: a Tarski attack on LLM probes
Ironically, some users took the “all” in my first post too literally. It should have read “for most practical purposes”
ziofill··on Truth is not a direction: a Tarski attack on LLM probes
Sure, but that’s not what we are talking about
ziofill··on Truth is not a direction: a Tarski attack on LLM probes
A direction that is 99.99% accurate survives this argument completely. For all practical purposes one does not need totality.
ziofill··on US citizen charged after GrapheneOS phone wipes during airport search
When talking about costs we should remember who is paying. Maybe overall the cost of the government getting that data is higher than the cost of them getting mad at you, but when a single individual is paying for all of it the equation might change.
ziofill··on The new rules of context engineering for Claude 5 generation models
I pointed Claude to this blog post and asked it to update the CLAUDE.md file accordingly. Worked a charm!
ziofill··on Ten Steps Towards Happiness (2015)
These are issues of civility, not politics.
ziofill··on New US homeownership measure puts people first
The problem will keep persisting as long as the politicians who have the power to change the rules of the game have the opposite incentive to do so.
ziofill··on Ring-Zero: Scaling Zero RL to a Trillion Parameters for Emergent Reasoning
I think it’s more accurate to say that the mechanism that grows a brain had millions of years to develop.
ziofill··on Decoy Font
Nice! A few years ago during my PhD I had made a Mathematica notebook that would take two images, crop them to the same size, apply a high-pass filter to one (which keeps the small sharp details) and a low-pass filter to the other (which keeps the large blurry blobs) and then superpose them back together. It was a bit hit and miss because e.g. if the eyes of two people were not in the same location the illusion would kind of break, but for text with outlined fonts it was amazing. I made a large one that would read "SCIENCE" from afar and "WORKS" from up close and stuck it on my office door.
ziofill··on MacKenzie Scott's giving, in quality-adjusted life years
I don’t disagree, but there is a vast difference between a wanted and an unwanted child. This is well studied and well understood. See freakonomics episode 394.
ziofill··on Speculations Concerning the First Ultraintelligent Machine (1965) [pdf]
I’ll take that bet
ziofill··on Snails' teeth beats spider silk as nature's strongest material (2015)
> Thats’s comparable to a single strand of spaghetti holding up about 3,300 one-pound bags of sugar

What an odd example. A mid-sized car would have been much clearer.

ziofill··on Pure-Python symbolic regression that rediscovered Kepler's law from 8 data point
Your answer makes no sense either
ziofill··on Pure-Python symbolic regression that rediscovered Kepler's law from 8 data point
The battery example makes no sense:

capacity_SOH ≈ 0.913 − 0.352 · tanh( cycle^((temperature/cycle)^0.485) )

I understand this fits the data, but exponents should be dimensionless, what is temperature/cycle?

← PreviousPage 2 of 19Next →