hamburglar

To be frank, no. In order to make windows not suck, you must invest a lot of time into knowing what to disable on the various SKUs, and in my opinion, it stopped being worth it around the time of windows vista (2006-7ish). I worked for Microsoft back then and ever since I left in 2007 i have thankfully been able to have a no-windows policy. I did briefly try going back last year for Unity dev but it was a mistake that made me want to quit computers entirely.

What are you trying to achieve with this comment? Does everything need to turn into a sniping contest about whether or not something is ai-produced? This is the world we live in now.

Dude. 80’s kids think of FAT32 as that new filesystem that supports more than 8.3.

I started my kid at 12 with an extremely locked down iPhone. She fights the restrictions at every turn and I have to make sure that she understands that finding loopholes is fun but also if I catch her violating the spirit of the restrictions there will be consequences. So she proudly tells me about clever workarounds she finds but still puts the phone away at the appropriate times. It’s kind of fun that she’s developing an instinct for subversion.

That seems like an odd distinction if it's clearly portraying a Cray. It's not like we have any proof that e.g. the Commodore 64 used in Mr. Robot was the real deal.

No Cray appearances? Surprising.

If you get an open source app from the App Store, is there any assurance it actually reflects the code in the repo? I’d think the signing step happening in isolation opens the door to tomfoolery.

For me, I don’t care one bit whether the text is AI generated or not. What stood out to me on this one is that they talked about something called shadcn, something called base UI, and something called radix as if they were things that everyone should be familiar with, but failed to consider that a reader might never have heard of any of them. I read for a while to see if they give some clues but nothing was forthcoming so the whole thing might as well have been gibberish.

They are slow. They all feel the need to make a “unique” UX that doesn’t actually work very well. They require some sort of subscription. They have limited functionality because they want to gatekeep functionality on some proprietary app. They require some godawful mobile app. They are just universally inferior.

Tesla’s is probably better than most, but they still lose by not letting me just use my phone, which is what I like.

I own a Ford and a BMW, both with CarPlay. The BMW is flawless. The Ford just refuses to connect about 10% of the time and requires the infotainment system to be rebooted. It also occasionally connects but leaves audio coming out of the phone speaker. So yes, implementations vary.

They all think they can do better, and it is sheer arrogance, because even the best of them is utter garbage compared to software that is actually built with fast iteration, UX research, and real user testing. No car manufacturers do a good job of this, and they all bake their atrocious UX into a $50k piece of hardware you keep for decades and which *never* gets a significant software update. The fact that they don’t see how impossible it is for them to win at this game is why they will never win. Sorry Rivian. Your vehicle is great but you’re handicapping yourself with software hubris.

Cow is not a verb

Are you cowed yet?

Its not exactly the same, of course, but its pretty analogous. It’s a minor, technically incorrect change that doesnt change meaning or understandability most of the time. Ive left out a bunch of apostrophes in this paragraph and Im pretty sure nobody will have a hard time understanding any of it. Yet it would bug me to actually write like this.

Specifically, not overestimating their own abilities and thus biting off more than they can chew. “Stepping on their own dicks” is a superset that includes all sorts of buffoonery like constantly making empty threats and backing off, but the basic assumption that the US Military is flawless and omnipotent and we can act accordingly seems to be our most fundamental error.

I think people who care about correctness and also read a lot automatically see the difference and it seems (and is) technically incorrect to use a hyphen where an em dash belongs. That’s really it. Kind of like you wouldn’t just leave out the apostrophes in your writing even though in most contexts they are not strictly necessary for comprehension.

Agreed. On an iPhone that’s the easiest way to type an em dash and consequently the easiest way to fuck up trying to write out a command line example.

Same. And my current daily driver laptop cost me $400 9 years ago. You can still do a lot for incredibly cheap.

I don't want to go back to the Frame's software

It’s not really a viable option if you wanted to. I have two Frames and they both just keep deteriorating to the point they are pretty useless. I have added an appletv to one of them and it’s vastly improved the UX. Now the only thing I need to do with the Samsung software is the unreasonably slow task of switching inputs.

Not having a lot of experience with this, I ask a naive question: is there a world where you can take your local LLM and hook it up to Claude and get more Claude-like results from your local model? Obviously, there are going to be material differences in how these perform, but are we getting close to a place where this is viable? I imagine that the answers are a combination of “not yet” and “yes but it’s a lot slower” and “yes but there is actually little point to doing this because ‘what Claude gets you’ is highly baked into anthropic’s models and that’s part of what you’re paying for.”

I think it’s pretty telling that with the YouTube example, I legitimately couldn’t figure out what he could have a problem with until he slowed it down. The overall effect worked and gave the impression that it was aiming for. The fact that you can get out your calipers and find flaws in a paused animation is not compelling in the least to me. I don’t think looking at your animations in slow motion is a bad exercise — it may reveal unintentional things — but I don’t buy that animations need to “make sense” when paused in the middle any more than a 250ms snippet of audio clipped out of the middle of a word needs to make sense.

Can you elaborate a little on what you asked Claude to do here? This is a pretty impressive one-shot.

I think there’s a palpable difference but many young developers have no concept of why.

Reminds me of “pre-jit”

Sure, but it turns out it isn’t guaranteed to listen. :)

I wrote a skill and included the instruction in Claude.md to always use this skill if it ever feels like grepping around in Unity files. It still took a lot of reminding it before it did it consistently. I’d interrupt it and say “it looks like you’re using grep for something you have an explicit tool for” and it would go “oh, my bad” and do it right. Took a few days to really sink in.

I work in Unity and I got frustrated with Claude constantly doing gross bash/grep/awk/sed/grep nested loops that took forever that I finally described (and had Claude implement and install) a tool that could, in a single pass, gather all this info from a Unity forest of scenes at once and answer all the questions Claude ever wanted to ask about a Unity project in a single pass that takes 50ms instead of 10 30 second iterations. It still took a lot of coaching to get it to actually use this tool, but it seems like I’ve convinced it.

Well one idea I had that changes the operational posture somewhat is to have the “reader” UX be a static page served completely from S3 and referencing other static S3 resources (the encrypted blob), then have the dead man’s switch “fail open” if the “user is not dead yet” signal goes offline. That way the company could almost run out of money and cease active operations but still have the data be readable by end users (with the keys, of course) for a long time. Having such a fallback plan in place gives confidence.

The other thing is being very transparent about the real world incremental cost of a user’s data. It slightly undermines the price point, but being able to say “look, incremental storage cost for your data over your lifetime is 33 cents. We have a bond that covers our current user base in perpetuity ten times over” is pretty compelling. You can say “yes, you’re paying way more than the incremental cost, but you’re paying for the care that went into making sure it’s got longevity.”

I can’t stress enough how happy I am for you that you built this thing that I consider a fabulous idea. I have an idea about how you can market it if you want to hear it. Let me know the best way to email you.

That’s great, and exactly what I was thinking. In fact, you have to wrestle with the fact that if it’s too cheap, people won’t trust it.

One thing I had been thinking about was that it’s important to be extremely open about the algorithms and infrastructure required, with a very transparent whitepaper on how it works (enthusiast level discussion of the technical details) and how you have designed the company to be extremely sustainable. I think your job is to convince people of two things: 1) this actually respects my privacy in a way I can verify myself (or lean on the technical expertise of other outside observers), 2) this company actually has a chance of surviving until I die.

If I can see those two things are true, a $199 lifetime purchase is a no-brainer (and that probably covers a dozen lifetime members’ operational costs). :D

I’d probably do more development in the meta quest virtual desktop environment if my primary occupation these days wasn’t writing software for the meta quest. It makes it difficult to iterate when you constantly need to flip to a mode where your monitors disappear.

I’ve been noodling over this exact business idea for a couple years! Part of my value prop was to make ops scale down to as close to zero as possible by having the encrypted data and most of the front end for accessing it live in S3 objects (or some similar mega-cheap object store). Aside from that, all you need is the dead man’s switch mechanism. And the secret fragments would live on QR codes that bring you to the static SPA with one of the fragments already loaded up, prompting you to scan more to complete the decryption.

Good work doing it. I guess now I don’t have to. :D You may have gained a customer from this HN thread.