username223

As a cyclist, I’m with you 100%. Unfortunately we’re probably going to do it again with self-driving cars, with segregated lanes, special markers, etc.

Very strong "reverse centaur" vibes here, in the sense of humans becoming servants to machines, instead of vice versa. Not that I think making things more machine-readable is a waste of time, but you have to keep in mind the amount of human time sacrificed.

That's some next-level incompetence:

1. Allow poorly-vetted third-party tools to run in CodeRabbit's privileged environment. The exploit used a Ruby code analysis tool that was probably written 15 years ago and meant to be run locally by trusted developers, who already had access to /bin/sh.

2. Ask for coarse-grained permission to access and modify others' code without any checks.

Either of those by itself would be bad enough. The future looks bright for black or white hats who understand computers.

Think of all the money spent combatting SQL injection attacks. Now we apparently have people deploying systems that fundamentally can't distinguish between code and data, and using pants-on-head 1990s tactics to fight "bad" input on a case-by-case basis. The amount of technical debt this will create is mind-boggling.

> Grew up reading Dave's columns,

So did I, except I'm probably from an earlier generation. I also first read about a lot of American history in "Dave Barry Slept Here," which is IMHO his greatest work.

> Pretty fascinating how foreseeable these last many years and decades have been.

It's a twisted kind of foresight, a lot like prophesies in ancient stories, that both predict the future and steer their subjects toward it. Neal Stephenson wrote books that appealed to (many, but not only) nerdy boys during their formative years. Lo and behold, 30 years later some of those boys have achieved wealth and power, and are trying to make random aspects of those books real.

Keep that in mind if you ever write a YA novel...

Yes, Apple screws you on SSDs, so that 2 TB adds a lot. If you need it, and don't want an external SSD (2 TB costs $150), the Framework is cheaper with a slower CPU. Maxing internal storage on a Mac laptop is a bad idea if cost matters to you.

EDIT: I haven't felt the need to spec a programming laptop like that. 16/512 feels fast enough, and 32/512 would have room to bloat... er, I mean grow. But I don't use a local LLM, and I don't know whether the difference between a heavily-quantized thing that fits in 16 GB and whatever you can fit in 48 is significant versus the ones running on absurd data center CPUs.

Maybe that's what I enjoyed about the old-school ultra crowd -- they were a bunch of unpretentious geeks with day jobs, doing outdoor stuff as a serious hobby rather than a calculated bid for attention and sponsorship. You can still find them, but you have to look harder.

I wish them the best, but if they can't compete with a MacBook Air on price despite Apple's huge profit margins, then maybe it's just not meant to be. People used to talk about paying the "Apple tax," but how many people are willing to pay the "Linux tax?" Mac OS is a similar Unix with the usual tools, and you can rent a VPS if you need Linux on an x86 sometimes. An MBA with an M4 will last 5+ years with a battery swap, and still probably perform better than whatever Framework releases in 2030.

I guess I'm not the target customer for this. I can see myself tinkering with a desktop, but I'd rather just have a laptop that runs fast and long enough, and stands up to abuse for 3-5 years.

Yes, Hummels was going for speed. See the Mahoney link for people doing various DV crossings in peak heat.

I love it! This was what the ultra marathon community used to be like before it became commercialized and professionalized (https://mattmahoney.net/ultra/, especially the story of Gravel Man https://mattmahoney.net/ultra/death600.txt). Now you can't run 50k without a crew.

EDIT: And this is what a serious amateur can do on that route: https://www.latimes.com/california/story/2022-03-10/astrophy...

I thought this would be a typical odyssey through America's haphazard public transit system. I've experienced my share, and our long distance buses are mostly a disgrace (pretty much anywhere in Latin America is much better), but then I saw this: "no Amtrak or Greyhound."

Okay, this guy plays on hard mode. I'm too old for that, but I respect his style.

That sounds grim, but not implausible. Domain-specific languages seemed like a significant improvement over general-purpose languages plus libraries. But now that we have a tool that lets you make Jazz Hands at your computer and have it spit out something that does most of what you want, do they really help?

Maybe some boring, kind-of-consistent language like C, Python, or Go is good enough. An LLM spits out a pile of code in one or more of them that does most of what you want, and you can fix it because it's less opaque than assembly. It doesn't sound like a job I'd want, but maybe that's just the way things will go.

They're welcome to try. RSS is still going strong, and I use it daily. Spotify and ClearChannel are trying to kill it for podcasts, and that seems to be failing.

ActivityPub in the form of Mastodon will probably never be as big as Twitter was, or as Threads could become if Facebook pushes it hard enough, but I'm fine with that. That's not really a place I would like to hang out, and I like my Masto instance.

This stood out to me:

> the automaker posted a new job listing days ago for engineers to help build a low-latency teleoperation system to operate its “self-driving” cars and robots.

It's June 17; they're supposedly launching in five days. Even if it's 95% off-the-shelf software, that timeframe for getting it up and debugged, then hiring enough humans to operate it, makes absolutely no sense. After nearly a decade of broken promises about FSD, has Elon finally trapped himself? Will there be a handful of "robotaxis" circling around a few blocks of Austin with Mexicans hiding in the frunk to drive them?

I'm glad I don't live there.

This bit from the WebKit response hits hard for me:

> Security Complexity: Proposed security measures (styling constraints, timing, and position mitigation) add substantial complexity, indicating possible fundamental issues with the approach.

"Security complexity" is never something you want to see, because it will become yet another game of whack-a-mole between browser makers and hostile websites (i.e. between Chrome and Google Ads). Does anyone believe that a standard will hold up against the AdTech industry armed with the full power of HTML+CSS+JavaScript? These are the people who brought you the "Enable push notifications? Yes/Pester-me-later" pre-prompt.

That was a weird turn to AI at the end, but otherwise an interesting reflection. I'm a little too young to have grown up in the era of the Cray-1, but even in the early 90s, processors ran at 90 MHz and hard drives cost $1 per megabyte. Back when personal computers ran at single-digit megahertz and had kilobytes of RAM, a Cray was mind-blowing.

The exciting part back then was that, while computers were never "good enough," they were getting noticeably better every few months. If you were in the market for a computer, you knew you could get a noticeably better one for the same price if you just waited a little while. The next model was exciting, because it was tangibly better. At some point personal computers became "good enough" for most people. Other than compensating for creeping software bloat, there hasn't been much reason for most people to be excited about new computers in a decade or more.

> What sort of crazy experiments would that have spawned?

Scientists then (at least a lot of them) knew what they wanted to do, and it required faster computers rather than more of them. A lot of that Cray power at the national labs was doing fluid simulation (i.e. nuclear explosions), and with the computers they had in the 80s, it was done in one or two dimensions, relying on symmetry. Going from n^2 to n^3 grid cells was the obvious next step, but took a lot more memory and CPU speed.

Yes and no. If you think Lima is a "rural village," you need to get out more, or at least check Wikipedia. It's a city of 10+ million people that doesn't follow American traffic laws, and there are plenty more like Huaraz (100,000+) where I can catch a human-driven taxi right now.

I'm saying a Jaguar I-Pace with lidar that knows how to follow lines and a high-res map isn't suitable for a lot of the world's roads. And how it will "scale" to what a taxi can handle right now isn't at all obvious.

> Waymo is operating in a handful of carefully chosen US cities. Uber can probably open in any city in the world (within reason) with probably a few weeks' effort.

This. Uber can operate anywhere that has human drivers and cell service. Waymo needs (I think) high-precision maps that are frequently updated, and simple traffic behavior.

Traffic in Lima looks like absolute chaos to an American, with endless honking and lane markers treated as vague suggestions, but there are not constant crashes, because the (mostly professional) drivers know the local conventions and communicate with each other by horn, eye contact, hand signal, etc. Huaraz is full of blind 4-way intersections with no stop signs, so drivers honk as they get close to one, and there is a remarkable lack of fiery death.

Waymo can't work in most places until it either changes human driving, or achieves AGI. Uber works as soon as it can pay local drivers.

That's grim. But the Eliza effect[1] makes @sama richer, so it's all good. Be naughty[2]!

[1] https://en.wikipedia.org/wiki/ELIZA_effect [2] https://www.paulgraham.com/conformism.html

> Or - are you saying the problem is that you need to install Bazel?

That. Then there's Facebook's build system, a few Python and JavaScript build systems, and pretty soon I have installed and have to deal with a half-dozen things that slightly improve upon Make. It's a maintenance burden if I ever have to touch any of these build systems.

This. When step one is "install our weird build system," I'll immediately look for something else that meets my needs. All build systems suck, so everyone thinks they can write a better one, and too many people try. Pretty soon you end up having to learn a majority of this (https://en.wikipedia.org/wiki/List_of_build_automation_softw...) to get your code to compile.

> What’s Meta’s strategy here? What’s their vision?

I don't get it, either. Facebook/Instagram/WhatsApp are ways to communicate with people you know, and they have a monopoly on that. (Well, Instagram is also softcore porn and product placement...)

TikTok beat them as mindless entertainment, showing people videos they're likely to watch until they end, and Zuck freaked out. Sometimes people would rather just watch TV than hang out with their friends! OMG! TikTok's bottleneck is that humans have to create the videos, so if Zuck can generate videos to maximize watch time, he wins.

Paying billions of dollars for a data-labeling company, though... Well, I guess it's not easy to put together a bunch of digital sweatshops in Kenya and the Philippines, but is it worth that much?

"Requested upon being needed" might work if it weren't possible for sites to get around it by probing and popping up their own "yes / ask me again later" dialogs. Have the APIs ask on the first call, with a "yes/no + make answer permanent" dialog, and return fake data if the answer is "no." If people were sufficiently annoyed by constant requests for stuff a basic webpage wouldn't seem to need, the web might become a better place.

But yeah, web browsers basically run arbitrary code written by hostile companies, with layers of indirection to confuse accountability. In that environment, you have to weigh "nice to have" against "could be abused," and err on the side of caution.

Who wanted their web browser to let hostile programs send notifications and access battery levels, unused fonts, etc.? Ad companies run the web standards bodies, so "people" (i.e. you and me) have to deal with this.

> TacoLoco is a traffic monetization network that uses deceptive tactics to trick Internet users into enabling “push notifications,”

Why is it even possible for hostile code (i.e. JavaScript) to send OS-level notifications? If clicking a link runs untrusted code with layers of legal insulation, that code should run in a very limited sandbox. It's crazy that we're turning the "Open Web" into an ever-expanding attack surface.

Just... sigh

"Spaceballs 2: The Search for More Money" was a joke, until it wasn't. I'm happy to see Mel Brooks being his mercenary self only a step from the grave, but he could probably do better than retreads of old IP. Maybe make fun of the Marvel movies instead?

> I also once had a driver in Chile who, somehow, micro-slept in stop and go traffic every time the car was stopped

Imagine how desperate you would have to be to drive a cab when you're that sleep-deprived (probably haven't slept in 36 hours). Now imagine someone took that income away from you to give it to Sundar Pichai.

Yeah, sometimes it's unpleasant talking to a cabby, and sometimes he won't take a hint and stop talking. But you might learn something if you try to engage, instead of vibe-coding inside a surveillance robot.

> If it didn't have the capability to do anything other than produce text in the client window for the user to do with as they please, it would be fine. Of course that isn't a very useful application of AI as an "Agent".

That's a good attitude to have when implementing an "agent:" give your LLM the capabilities you would give the person or thing prompting it. If it's a toy you're using on your local system, go nuts -- you probably won't get it to "rm -rf /" by accident. If it's exposed to the internet, assume that a sociopathic teenager with too much free time can do everything you let your agent do.

(Also, "produce text in the client window" could be a denial of service attack.)