HNHacker News
TopNewBestAskShowJobs

irjustin

7,091 karma · joined May 6, 2019

irjustin.com
submissionscomments
irjustin··on If AI writes your code, why use Python?
So, it's really interesting. We've started moving away from python libs because 25% OSS is out of date and another % is custom tweaks to the software help our use cases. In both scenarios it means our own fork.

And honestly it's not burning that many tokens if you've got an existing example lib to point to.

irjustin··on I want to live like Costco people
> Yet, I keep going.

Then I think Costco has done its perfect job and exactly the goal the founder set out to create: Quality goods at affordable prices.

They optimize for those 2 things first. Consequently, everything else becomes a management of chaos (the part that stresses you out and thusly hate).

If they did try to make the experience better, it would cost them someplace. And honestly, you're just at whole foods at that point.

irjustin··on How ChatGPT serves ads
this would be a breach of trust and short term would work great but long term is too detrimental.

same thing could've been said for search results, so at least that part is still "safe".

irjustin··on San Francisco, AI capital of the world, is an economic laggard
OpenAI was originally founded on the parent's concepts, but obviously that changed at some point.

Currently, Elon is suing Altman because Altman turned it for profit. Oddly, in that Elon is taking the moral road?

irjustin··on CRISPR takes important step toward silencing Down syndrome’s extra chromosome
I wondered the same thing and according to Gemini a chromosome is massive vs a few genes. Cutting it out with crispr is possible, but it's too big of a change and would lead to cell death rendering whatever change either useless or kills the host given the possible stage this treatment could be delivered at.
irjustin··on Artemis II safely splashes down
> Artemis acceptable crew mortality rate is 1 in 30.

This seems insane to me. That X decades later we accept, with all our advancements in tech, a weaker system than ever before. That if we send 30 people we _accept_ that one is possible to die.

That's the starting point? That's what we document as acceptable?

irjustin··on We've raised $17M to build what comes after Git
I'm not sure if I should take these comments seriously or as a joke...
irjustin··on Artemis II is not safe to fly
> It's not like the NASA administrator gets a bonus when a rocket launches

It's related to funding. I mean it's always money, right?

But in Challenger's case, there was very heavy pressure to launch because of delays and the rising costs. I remember in a documentary they explicitly mentioned there was a backlog of missions and STS-51 had been delayed multiple times. To rollout/fuel, costs a LOT and challenger had been out on the pad for a while. Rollback was a material risk+cost.

For columbia, yea less about money. They ignored the requests to repoint spy sats and normalized foam strikes.

> I think NASA engineers are well aware of the risks and have done the math to convince themselves that this is safe.

And that's the way it should be. Everything has a risk value regardless if we calculate it or not. It's never 0... (maybe accidentally going faster than light is though?) We just need to agree what it is and is acceptable.

Story time - I was a young engineer at National Instruments and I remember sitting in on a meeting where they were discussing sig figs for their new high precision DMMs. Can we guarantee 6... 7 digits? 7? and they argued that back and forth. No decisions but it really stuck with me. When you're doing bleeding edge work the lines tend to get blurry.

irjustin··on Artemis II is not safe to fly
1/100 is absolutely terrible. Shuttle had 1.5% failure rate. Bonkers.

[edit]

For comparison, commercial aviation has something like 1 in 5.8m or 6x 9's of reliability.

irjustin··on Artemis II is not safe to fly
> In both Challenger and Columbia, nobody bothered to analyze the problem because they didn't think there was a problem.

Being pedantic, NASA management "ignored" engineers - because money.

That said, I 100% agree with you assuming:

> “We have full confidence in the Orion spacecraft and its heat shield, grounded in rigorous analysis and the work of exceptional engineers who followed the data throughout the process,” Isaacman said Thursday.

I only say assuming not that I don't believe Isaacman, but historically NASA managers have said publicly everything's fine when it wasn't and tried to throw the blame onto engineers.

With Challenger, engineers said no-go.

With Columbia, engineers had to explicitly state/sign "this is unsafe", which pushes the incentivisation the wrong direction.

So, I want to believe him, but historically it hasn't been so great to do so.

irjustin··on iPhone 17 Pro Demonstrated Running a 400B LLM
a sandwich bag would work wonders, then you could use ice to counter the plastic's thermal inefficiencies!
irjustin··on US SEC preparing to scrap quarterly reporting requirement
I'm generally with the report often camp. It forces automation all the way down even the auditing.
irjustin··on ATMs didn't kill bank teller jobs, but the iPhone did
> But will it?

No, because if you think about Startrek the endgame is replicators. Well the concept that 100% of basic needs are met.

At some point work becomes unnecessary for a society to function.

irjustin··on Simple screw counter
Sorry i completely missed this. If you don't see it, it's okay - I probalby miss any replys going forward.

Being upfront, I have no idea what I'm talking about. Just some arm chair engineer.

The poster needed 6 parts which is JUST into annoying. My personal thoughts are what they need isn't dispensing but alignment. Thinking deeper I can agree that weight might not the most efficient here.

They're building the aligning and dispensing tool but I argue that's over engineering the problem. If it's aligned it's VERY easy to count 6 via a mark along the track and just push it to the end against your finger and based on the mark you know you have exactly 6.

To me the hardest part to make "just work" is the dispensing, but if you remove that it becomes a much easier problem. There's enough sales volume, you can make a vertical fixture that is a stack of fixed aligning tracks. Your fingers become the dispenser. Sweep and move to the next track.

Just random thoughts.

irjustin··on Simple screw counter
You grab a "rough amount" and by using weight all you need to do is diff 2,3,4? Ideally 5 and under.

it's very easy to count <=5 visually, but if your package requires 12 nuts, repeatedly counting up to 12 is so stressful the poster built an entire counting machine.

irjustin··on Simple screw counter
> Is this done because of the error in weighing the smallest parts, so they have a margin for error by allowing for an extra 1 or 2?

This is a secondary benefit, the primary benefit is if the end user loses/breaks one. That part very well could be show stopper (Ikea 110630 anyone?). Now the end user is stuck - has to call, you have to ship, do you charge? do you give for free? they have to wait. they're annoyed, you're annoyed.

No one is happy.

The supply chain headaches for giving exact number of tiny parts is terribly expensive, relatively speaking. So you give spares because in the long run it's way cheaper.

irjustin··on Simple screw counter
At scale, use weight and supply 1 or 2 extra.

This is how pretty much every IKEA, LEGO, etc works with very small, cheap parts.

End users benefit because it's easy to drop/lose/break one.

irjustin··on Jane Street Hit with Terra $40B Insider Trading Suit
Probably coincidence - general market is up strongly too. Or, too hard to tell anyway.
irjustin··on We installed a single turnstile to feel secure
That is just the part that gets the most press. Having lived here for a while now.

1. At a young age, you're taught to follow the rules.

2. "Someone's always watching". Lots of CCTV. Community reports.

3. Plenty of police who have the ability and time to investigate even the most petty things.

Trust in the system starts with 1 but is really carried day to day by 3.

irjustin··on The Age Verification Trap: Verifying age undermines everyone's data protection
I was more responding to the part about not trusting your own gov cuz how do you build a system where you don't trust a central authority when identity is required.

I don't think it's possible.

irjustin··on The Age Verification Trap: Verifying age undermines everyone's data protection
I'll bite.

> It's not zero knowledge for me then. Also - if there is ANY possibility to track anyone. And/or centrally mark someone "nonverified" then it makes more problems than solves.

> Even if I trust my govt (no way), even if it'd be fully ZK with no way to track anyone… still govt would have a way to just block some individual "because".

Is this even actually possible? If you want any sort of identity verification you HAVE to trust someone, whether age or full ID. Literally impossible.

Zero trust systems in society don't work. If you don't care "who" then yes, zero trust is just fine... but then what's the point of "age verification"?

irjustin··on Binance fired employees who found $1.7B in crypto was sent to Iran
While likely true, that's not how our brains work.
irjustin··on Blood test boosts Alzheimer's diagnosis accuracy to 94.5%, clinical study shows
> why we would want to detect Alzheimer’s

At a personal level, I've been through this with my grandfather.

I want to know. My family wants to know. I want to prepare because there are things I want to do today that I know I won't be able to do in the future.

In many ways, it's just like many terminal cancer diagnoses. You're going to lose that person, but you have some time.

irjustin··on Binance fired employees who found $1.7B in crypto was sent to Iran
Literally - this is my friend =(

For him, it was something like a ~$30m pizza at peak. We're asked not to bring it up.

irjustin··on How AI is affecting productivity and jobs in Europe
FWIW, these studies are too early. Large orgs have very sensitive data privacy considerations and they're only right now going through the evaluation cycles.

Case in point, this past week, I learned Deloitte only recently gave the approval in picking Gemini as their AI platform. Rollout hasn't even begun yet which you can imagine is going to take a while.

To say "AI is failing to deliver" because only 4% efficiency increase is a pre-mature conclusion.

irjustin··on Warcraft III Peon Voice Notifications for Claude Code
If we're being strict to peons, then no, that's the human's voice - https://www.youtube.com/watch?v=5r06heQ5HsI.
irjustin··on xAI joins SpaceX
At these scales, financial and social are very intertwined, it's both.
irjustin··on Please don't say mean things about the AI I just invested a billion dollars in
In broad strokes - disagree.

This is the knife-food vs knife-stab vs gun argument. Just because you can cook with a hammer doesn't make it its purpose.

irjustin··on I let ChatGPT analyze a decade of my Apple Watch data, then I called my doctor
Yeah I guess I just don't see eye-to-eye on this.

I 100% understand those frustrations. That the "detectors" should've been more accurate, or the fears, battery of tests, and costs associated of time and money. But, if you have the means to find out something that could have been extremely concerning is actually "nothing wrong" - isn't that worth it?

My friend is 45, had bloody stool -> colonoscopy -> polyps removed -> benign. Isn't that way better than colon cancer?

Maybe it's a glass half-empty-full thing.

irjustin··on I let ChatGPT analyze a decade of my Apple Watch data, then I called my doctor
How is this apples-apples at all?

But to answer directly... yes? yes, I am.

[edit]

A bit it more real. My blood pressure monitor says my bp is 200/160. Chat says you're dead get yourself to a hospital.

Get to the hospital and says oh your bp monitor is wrong.

I'm happy? I would say that I am. Sure I'm annoyed at my machine, but way happier it's wrong than right.

← PreviousPage 3 of 34Next →