HN user

ctoth

13,297 karma

I'm a totally blind software developer.

I can be reached at q.alpha@gmail.com

Posts649
Comments1,245
View on HN
www.ai.mit.edu 5d ago

Black Horse: One Stop to Orbit

ctoth
2pts1
github.com 3mo ago

The 200k Ghost: Instruction Degradation in Long-Context LLM Sessions

ctoth
3pts1
www.wsj.com 4mo ago

A US citizen now runs Mexico's top drug cartel–and targeting him is complicated

ctoth
12pts5
www.eff.org 4mo ago

A Declaration of the Independence of Cyberspace

ctoth
7pts2
www.ribbonfarm.com 4mo ago

You Are Not an Artisan (2013)

ctoth
1pts0
thezvi.substack.com 4mo ago

A Tale of Three Contracts

ctoth
2pts0
www.johndcook.com 4mo ago

Cepstrum, Quefrency, and Pitch

ctoth
1pts0
www.theargumentmag.com 4mo ago

Anthropic is both too dangerous to allow and essential to national security

ctoth
3pts0
www.secondbest.ca 5mo ago

Empiricists vs. Extrapolators

ctoth
4pts0
malmesbury.substack.com 5mo ago

My journey to the microwave alternate timeline

ctoth
6pts0
aribrin.substack.com 5mo ago

A Metal-Detection Objection at the Reception

ctoth
1pts0
www.metropolitanreview.org 5mo ago

The Writers Came at Night

ctoth
32pts21
www.theargumentmag.com 6mo ago

We do know that Waymos are safer than human drivers

ctoth
3pts4
news.ycombinator.com 6mo ago

Ask HN: Tips for getting the ROM for an old speech synthesizer?

ctoth
3pts2
www.theargumentmag.com 6mo ago

I can't stop yelling at Claude Code

ctoth
3pts5
arxiv.org 7mo ago

Information Flow in Logical Environments (2016)

ctoth
1pts0
yaschamounk.substack.com 8mo ago

The Peculiar Persistence of the AI Denialists

ctoth
4pts1
www.dwarkesh.com 8mo ago

Satya Nadella – How Microsoft Is Preparing for AGI

ctoth
3pts1
www.thefp.com 8mo ago

Peter Thiel: Capitalism Isn't Working for Young People

ctoth
15pts16
www.maximumtruth.org 8mo ago

Deep DIVE: AI progress continues, as IQ scores rise linearly

ctoth
5pts0
www.dwarkesh.com 8mo ago

Thoughts on the AI Buildout

ctoth
5pts0
www.bostonherald.com 8mo ago

Navy loses two aircraft from USS Nimitz aircraft carrier within 30 minutes

ctoth
16pts3
news.ycombinator.com 9mo ago

Tell HN: Claude's weekly limits don't reset every 7 days

ctoth
3pts1
thezvi.substack.com 9mo ago

Bubble, Bubble, Toil and Trouble

ctoth
1pts0
www.dwarkesh.com 9mo ago

Andrej Karpathy – It will take a decade to work through the issues with agents

ctoth
1212pts1115
newsletter.pragmaticengineer.com 10mo ago

How Claude Code is built

ctoth
3pts1
www.dwarkesh.com 10mo ago

Autonomous robots are much closer than you think

ctoth
2pts0
aiascendant.com 10mo ago

Why 95% of AI Commentary Fails

ctoth
3pts0
www.denverpost.com 11mo ago

Trump seeks to fire official overseeing jobs data after weak report

ctoth
35pts3
news.ycombinator.com 12mo ago

Ask HN: What's your uv exit strategy?

ctoth
13pts8

The prices are nowhere near each other. K3’s API runs $3 per million input tokens and $15 per million output. Claude’s top model costs $10 and $50 for the same units.

And this is the point where your internal compiler should have started shouting 'Type Error'

Notice the trick here?

Then there’s the fine print. Claude couldn’t sustain Fable access on the twenty dollar plan, so they turned it off, and the plan quietly falls back to Opus.

Where is the Fable-class Kimi model at all?

When Edward Snowden revealed that Somebody Important had been taking an interest after all, the tech bros, who had grown up free of adult supervision, felt betrayed and started a campaign to encrypt everything and anything so that prying eyes could never again look them in the cards.

TIL: Phil Zimmermann was a "tech bro" and had a time machine.

Watching a company I respect turn to the classic "first hit's free" dealer technique does make me feel a certain way.

On one hand the thing they are offering is legit great, and how do you get people to use/understand a whole new type of thing without offering them some for free?

On the other hand ... you really gonna get people used to an extra 50% then take it away? When this has been most of your new signups experience with the system?

The predictable pushback when people realize the workflows they developed over these two months are no longer viable will be ... quite something.

AI 2040: Plan A 11 days ago

When I learned that this website is by the same people who gave us AI 2027, I immediately thought about the Wikipedia page on doomsday predictions.

Then, you opened the page and read it and realized that this prediction was contingent? You know what a conditional is (I would assume) if this, than that?

Then you realized that the only reason you were posting this comment was as a sort of silly gotcha "Oh look at the guys who keep increasing the number" instead of talking about the differences between the scenarios?

Then what happened?

Fable 5 Is Back 21 days ago

Cybersecurity? It won't even help me work on my speech synthesizer[0]!

I guess? If you squint? DSP code could look a little like AI training code? ... Er. No. Not really I'm pretty lost on this one.

The task was literally just to compare against the "make a beautiful voice" plan, see what we've implemented, see what's left to do, and to make recommendations for low-hanging fruit, anything we've done wrong so far? (aaaaand ... downgrade. At least it wasn't silent.

[0] https://github.com/ctoth/qlatt

Actually, it might be closer to say this is similar to a child playing pretend alone with their toys, except perhaps a bit less challenging in that creative roadblocks or narrative building is instantly abdicatable to the machine.

Wouldn't the obvious analogue be a video game? Especially one where you can edit the asset files (making your weapons super-strong, for instance?)

If it's every summer, how is it unnatural? If it's poor planning how is it every summer? There's poor planning and then there's ... what, forgetting that summer happens? It sounds to me like somebody sat down and penciled in some numbers and decided that "it makes less money to let it break?" which seems pretty weird when you consider second-order stuff (but it's not like people tend to do that anyway)

In general though, if you want Fortune 500s to utilize your service/company, don't utilize a novelty TLD.

New potential technique to not have your open source project yoinked/resold by cloud providers? :)

increase stock repurchases

I just can't wait to get back to when innovation meant financialization, can you?

Building things? Real Jobs for real electricians? Real buildings? Real compute? Why would anybody want that!

Yeah pretty sure we had a whole bruhaha ~250 years ago about this question of where precisely power belonged. I for one think we mostly got it right then and would be reluctant to shift the power back to the individual sovereign and away from the people.

Came here for a five-gallon bucket hooked to Dulwich (archiving rain?), Slightly disappointed :)

Go Git and Dulwich and friends are indeed fun tech.

Claude Tag 29 days ago

As someone who has Claude and Codex both bridged into a MOO along with other humans I can 100% believe this. It really is a different paradigm that you have to experience. Of course a MOO is already set up for social programming, so it will be interesting to see what platforms evolve (or find themselves coming round again) to facilitate this.

Brother, I'm trying to. Any thoughts?

Also, what's a "worthwhile" job?

Have you read Player Piano?

Are brainjobs more important than musclejobs?

If we can automate musclejobs so people don't have to die in the heat and we can automate brainjobs so people don't have to lose their sanity in offices, wouldn't that be a better world?

Note: it probably won't be, because of what I like to call the "Enterprise High" effect, but here we are. Basically once the only thing people have to compete over is status, shit gets nasty quicklike, think PG's high school essay but for everything and forever. Everything goes into rivalrous goods like who's sleeping with who.

Star Trek as soap opera where all their needs are met and the only concern is purely social.

But this is probably better than a world where people don't actually have material needs met?

I legit don't actually know. Would love to figure it out though, what do you think!

Who do you think will be answering these questions

Yes, this is the conversation we should probably be having! Thoughts?

Given that this is what is happening, given that we aren't going to stop physics, getting a handle on this indeed seems like one of the more important things we can do. If it will have any chance of working at all. If we don't get paperclipped.

But this is at least a much ... (forgive me) less wrong conversation to have compared to the one where everybody assumes we've already lost.

What do you think we should do? What powers do we still have where we is general humans who just want a pretty good life?

What do we do about the inherent centralization that big models seem to require, but how do we trade that off at the same time from everybody being able to synthesize the next Covid by asking their cell phone a question? What does it mean when most if not all white collar work actually can be automated?

Do we all end up playing VCs in our underwear swiping left/right on ideas our agents have to make money? Are we still competing in the market with the AIs?

Is there a class war? Is there some other weird thing? I don't know but man ... I sure would like to have those conversations.

The pizza analogy smuggles in this idea of cheep/mass-produced. I'm talking about blind people who can now prompt their way to an accessibility mod for their favorite game, the sort of thing which literally would have never been written before. How you know it wouldn't've been written is by counting the accessibility mods pre and post LLM.

Now generalize this. Every tiny community, every person with a disability, everybody for whom the default software doesn't work right? Can now change it specifically for them. Not add peperoni, that's far too low-dimensional to capture what is happening. Actually build their own interface, be able to use something they simply didn't have access to before, and critically not depend on another programmer (there are like a dozen of us blind ones!) to build something for them.

The thing is I totally, 100% get this. The other thing I can't help but see though is how excited my non-programmer friends are to finally be able to make software. The sense of pride and accomplishment from non-coders who are finally able to make something work the way they wanted to.

We almost need like ... noncanonical software? Not so much forks, but like ... Maybe software as like a cluster? an ecosystem? On-demand app store where features / forks are shared/upvoted/evolved by the community where the maintainers don't have to get burnt out, and when it inevitably becomes a ball of mud oh well it does the job? I really don't know!

I hope we can think about some answers and not get tribal though because this is really a huge problem and also a huge opportunity and so a minor reminder that there is a baby in that bathwater?

Waymo Premier 1 month ago

Flight?

I want to drive. I want to bring my cat and bring some stuff back from my dad's house. My parents just drove up here to visit me, I would like to do the same. Not take a train. Not take a plane. I want to hop in a robocar and drive to Florida. The same thing that every other person with a car can do whenever they want to. Freedom.

Waymo Premier 1 month ago

I'm blind. I wish to hop in a robocar and drive from Denver to go visit my folks back home in Florida. Is the Denver Access-A-Ride going to take me? Which public transit is available?

Waymo Premier 1 month ago

It's a class marker

autonomous 6000lbs tank

Hmmm. The meta here made me chuckle. Calling cars tanks is certainly a class marker.

The interesting thing is how I was making a very contained point pertaining to cameras, and how cameras, which we were talking about in this thread, seeing a verbal confrontation, could not and should not make a call, because a verbal confrontation is not a legal event. You then took this into a totally different case involving ... what? hypothetical recording of a conversation between two hypothetical terrorists? To prove ... what? My point is that it is not a shortcoming of the camera that it is not making a judgement call on the thing OP was originally talking about. A verbal altercation between two people. I was not talking about a hypothetical bombing. I was not citing a specific law, I was not advocating that there should be a law, I was not advocating anything about whether or not we should ban collection of existing evidence. I was not making any of these moves. I was saying simply: a camera looking at two people in a verbal argument from far enough away that it cannot hear the conversation is not a failure of the technology. Not every negative interaction between two human beings is criminalizable.

This entire thing is extremely historically contingent, and we could easily stop this train tomorrow.

So, the Baruch Plan?

The Manhattan Project was $~2B in 1945 dollars, and a national-scale industrial mobilization. Now North Korea has the bomb. That's with nuclear material, which doesn't get easier and easier and easier to work with every year.

Compare to the price to train GPT-2 in 2019 ($43,000), and in 2026 ($73) [0].

[0] https://x.com/karpathy/status/2017703360393318587