OGWhales

I came across this article today, by pure chance, and thought you might enjoy it: https://cs-syd.eu/posts/2022-08-22-how-to-deal-with-money-in...

Indeed, though this is platform dependent. Mainframes have strong hardware support for decimal fixed-point arithmetic, making it the preferred representation for fixed-scale business data, though conceptually it's very similar to binary fixed-point.

Decimal floating-point is also hardware-supported, so it doesn’t carry the same computational overhead it does on platforms where it must be implemented in software.

> What LLMs make possible is for me to say: find out all the ways this thing works.

They are not good at that. The space of possibilities can be massive and LLMs are terrible at exploring such space because they predict from the prior tokens they made. They are inherently bad at exploring new space because it is antithetical to how they work.

To me it's deeply concerning that so many people are getting fooled into thinking that LLMs are actually good at covering their bases like you're describing here. It's one of their weakest qualities.

> Analyze the different ways we can run this software, build a fuzzer, build property tests, and run this software in every scenario possible.

It usually does quite a bad job at this too, often the tests it wrote feel like that of a lazy student that didn't really want to do the task and just sort of cheats at it or does a really shallow job. It certainly cannot run the software in every scenario possible.

> Anyway all of this reads like someone who is not actually using LLMs to build software or hasn't tried them in a while

This is something that is said by everyone who contests anyone pointing out the risks in being overly trusting of AI or otherwise points out their flaws. I use the latest and greatest all the time and all the time I'll point out something that it totally overlooked and get hit with the classic "you're absolutely right". This occurs because I actually read the code and can see the myriad of blatant issues that still occur when using LLMs and know better than to trust them. You will find so many issue by delving into the details.

I think I am! I use it in the browser instead of teams/the desktop app, but my understanding is that it should be the same thing. I do know most people don’t have the premium license for it at my company, not 100% sure what difference that makes but they say it’s better somehow and has access to better models, or at least did at one point.

But yeah, comparing to codex/claude code was a bit silly since those can do way more. It’s closer to the chatGPT. I’m pretty sure 5.6 Luna, not sure exactly since it doesn’t specify Luna/Terra/Sol

Huh... not my experience with the "premium" license. The interface isn't as nice and definitely less powerful than something like codex/claude code/whatever, but I don't seem to have any issues with truncation or things like that. I'm usually using 5.6 thinking (not clear which 5.6 model exactly) when I'm using copilot at work. It seems... fine? I always see people here always complaining that it's nerfed, truncates things, etc etc. The only difference I see is that interface isn't as nice nor as powerful, but I don't notice the answers being noticeably worse or anything like that. Occasionally it messes up code block formatting, but that's the only thing I can think of that annoys me.

Now, the "copilots" integrated in certain Microsoft products can be really horrible. I assume those are some sort of custom models that are not very capable, most of them I consider effectively worthless. But the web chat interface where I can pick chatGPT 5.6 seems fine...

Indeed, I found the oil-painting effect really rough on certain year iphones in low light/when using one of the zoom lenses.

Even if they don't care about bokeh per se, they can still find pictures with really bad bokeh to be ugly.

> I received several very emotionally charged responses centered in the personal credentials of the author

I took a look and honestly the replies were a lot more reasonable than I was expecting them to be. I don't think it's that out of place for people to inform you that the author has been in this space for some time and has prior work to look at (pre-vibe coding era). Really there was one reply citing his prior work, you responded to it calling it a very strange reply and were "very confused" why they would point to his history, even though you had just said things like:

>I'm glad you're having fun vibecoding, and I like that you're interested in this area of research/engineering

To me, it makes sense why someone would say the author has been interested in this area for a long time and pointed you to some prior work that was not vibe-coded, given your comment strongly implies they are just messing around and don't know much about the field.

To be clear, I share much of your feelings in your original comment. I just felt the replies weren't so unreasonable either. At least, I was expecting them to be a lot worse.

> what's best, cheap, no wires, no headaches and happy or expensive, wires

You can get great sounding wired iems for very little money, while wireless earbuds are usually pretty pricey for anything not crummy. Cables are indeed annoying, but so is your earbuds being dead because you forgot to charge them. The battery problem gets worse too as they age. If you want reliable, cheap, good sounding music it is pretty hard to beat chi-fi iems.

I have 15 years worth of songs, account is less than 10 years old. Arguably I would be okay dropping some of those songs if I wanted to buy the songs. The biggest difference for me though is the issue of discoverability. It is very easy for me to find and save new songs on Spotify, it would be much harder for me to find the songs I like if I had to pay for them, especially since a lot of the songs I like are not just songs from popular artists but are often just one of two random songs from random artists. Is there a good middle ground here? I keep a copy of my song list in case I want to ditch Spotify but this is the main hurdle for me.

Flock goes far beyond that, they allow the government to retroactively string together a person's history of movement. Prior cases establish that continuous and retrospective tracking like that violates a reasonable expectation of privacy and requires a warrant.

You’re correct, would be equivalent to 150Hz. I was curious too since my monitor is 360Hz :)

I don't find surveillance cameras better in that regard. I think physical security can look decent too, like wrought iron window security bars can actually look quite nice imo.

“Too quickly” is still quite slow, you can do a lot more comfortably at higher fps and it looks just fine. Also, I’m assuming you only mean movies and similar videos, not the other kinds of videos I mentioned?

Movies are made by professional videographers that take care to limit the downsides of low fps, such as stuttering (still, sometimes movies will pan too fast and the stuttering makes it painful for me to watch). YouTube is very different, there are lots of amateurs and many of the videos don't need to make people feel like it's a movie. It doesn't seem the soap opera effect really applies when it's just a guy showing actual real life (and there's no interlacing like with old soap operas). For this reason, I've always found 60fps better on YouTube. Lower fps doesn't look more sophisticated imo, it usually looks worse unless it happens to be done by a good videographer that wants it to look like a movie. It's especially terrible for things like action content (mountain biking, karting) or things like fps games, where the higher frame rate helps a ton with all the fast motion. This happens to be a lot of what I watch on YouTube.

The way I see it, unless you are specifically going for the "movie" look and take care to work around low fps like videographers do for movies, then higher fps is probably better.

Maybe dogs see little dogs instead of little people...

I'm not sure how well proven it is, but the theory I saw for the common "sacred geometry" that people see while on psychedelics is that the brain normally is quite good at filtering out "noise" and the psychedelics limit this filtering. So, sacred geometry could be a type of noise produced by our biological hardware that the brain normally filters out as unnecessary. Other things besides psychedelics can cause these same patterns to be observed too, from what I understand.

And yet, it remains possible that a human could write the same sequence of characters.

Super cool! What a great page.

> This wasn't a thing 10 years ago and we weren't in a police state

Curious then that you are arguing in favor of a police state with mass surveillance to solve these issues...

The issues you are complaining about are not issues due to a lack of mass surveillance. You're arguing in favor of mass surveillance with a diatribe of complaints over petty crimes that you describe as terrorism, while ignoring any of the potential downsides of what you are arguing for. Consider that you also break laws yourself, even if you think you don't, and what you are arguing for can and will be used against you too (not that one needs to break any laws for mass surveillance to be utilized against them).

Not that shufflecake solves the issue you highlighted, but I found the shufflecake FAQ to be a good intro to the topic for anyone curious. It does a good job explaining the threat vectors and the relevant trade offs, in particular the TRIM and ORAM sections. It’s also just a cool project: https://shufflecake.net/

Yeah, in preferences there is an option by beta options that says: "Use new Reddit as my default experience".

Once it a blue moon it doesn't seem to work and going to old.reddit then back to reddit seems to fix it.

They used to offer a more reasonable mobile website but they got rid of it. Using the old design now is just the desktop version which doesn't work well for the reasons you said.

Is this comparing old.reddit to reddit.com? I am using the normal url but still use the old design. I just toggled a setting in preferences to use the old design without needing to go to old.reddit, only reason I ever use old.reddit is if I'm logged out. That might be even more niche though, it seems like most people actually use the other url which always confused me when there is a toggle.

Gell-Mann Amnesia effect

I can see that happening, though that sounds pretty bleak

>The divergence in business models is directly reflected in financial data. SemiAnalysis estimates that Anthropic's overall gross margin has rebounded from negative 94% in 2024 to the mid-60% range, with the gross margin of its API business exceeding 80%.

Their link seems to claim semi analysis thinks it is 80%. It looks like it might be referencing this newer article from them, as the same picture is in both articles, but I didn't feel like paying to find out: https://newsletter.semianalysis.com/p/anthropic-3q26-profit-...

> even if we assume their margins are 90% (they are not)

How do you know they are not?

It will be curious to see the cost of inference for these newly released open weight models and will help give an idea of the actual cost of inference. But for now, I think saying the $200 plans allows for "tens of thousands of dollars worth of inference" provides very little insight when you are measuring the inference cost in API pricing with an unknown margin.

> You are missing the big picture here The AI literally was able to do a year's work of a top notch developer team in just 11 days (mainly because of a lot of human intervention, otherwise it would have been faster) Why do you think its cant or wont be able to replace the lead developer and/or designer's 11 day job !

It would help if you actually engaged at all with any of the points they just made. They were good points!