mrob

The "anonymity of online games" is much greater than it used to be. It used to be standard practice for people to host their own private servers and play with people they knew. This naturally filtered anti-social behavior because people knew they wouldn't get invited back. And public servers were a more casual environment, where the occasional cheating wouldn't ruin any serious competition.

"No, thanks" is also insulting and unacceptable. I want "No, fuck off", so let's compromise on a plain "No" like we used to get in the era when computers were made for the users.

480p ("EDTV") wasn't very common. Most people upgraded directly from SDTV to HDTV.

Leaded glass was a cost-cutting measure, used only for the back and sides of CRT. Leaded glass gradually turns brown under X-ray exposure, so the front screen used barium and strontium instead. There's no reason the whole CRT couldn't use this kind of glass. There are probably other heavy elements that would work too.

(Early black-and-white CRTs running at lower voltage did use leaded glass for the front too, but they aren't the CRTs people interested in old games care about.)

As far as I know, all plasma screens use pulse width modulation to control the brightness of each subpixel. This produces even worse motion quality than a typical LCD; you get both sample-and-hold blur from the high persistence, and phantom array effect artifacts from the PWM. IMO, plasma displays are only suitable for watching movies, which have terrible motion quality anyway because of the low 24 fps frame rate.

>or just using shaders for free (which gets you 99% there on an HDR 4k

It gets you about 90% of the image quality and 0% of the motion quality. Simulating CRT motion quality on a sample-and-hold display requires around 1000Hz refresh rate.

You do want to simulate it, because it's the only way to preserve the original motion quality of games designed around a fixed 60 fps frame rate without changing the gameplay or adding latency.

The ghosting on the original Game Boy did not blend pixel boundaries and did not affect static images. It looks nothing like a CRT and nothing like a zoetrope (which has sharp images and reasonably low sample-and-hold blur because of the strobing).

>PC CRTs would always be quite sharp, due to VGA, no matter whether it (the CRT) had a low or high resolution.

VGA CRTs are always "high resolution" compared to the SD CRTs discussed in the article. If you display a mode 13h DOS game on SD CRT, which is possible with special hardware, it will not look as sharp as it would on the VGA CRT is was designed for.

This sharpness from using higher resolution CRTs is in addition to the sharpness from using RGB connections. And if you somehow manage to connect a VGA CRT over composite, it will be less sharp but still sharper than a SD CRT over composite.

The point is there are two factors here: the sharpness of the CRT itself and the sharpness of the connection method. The article only talks about SD CRTs, so the CRT's sharpness isn't going to make much difference there. But when you consider MS-DOS games, which used sharper CRTs, the CRT's sharpness becomes important.

Those are the SD CRTs the article talks about, not the VGA or SVGA CRTs used to play DOS games.

But in practice, VGA and SVGA CRTs always used RGB connections, so that's not relevant to games.

If you assume a SD CRT, as the article does, then yes, the display connection makes the most difference (although there are still differences between individual CRTs).

But that does not cover all CRTs relevant to pixel art. If you display low resolution art on a CRT designed for high resolutions it makes a big difference in pixel sharpness.

Not the same at all. Free as in beer means you get nothing. Free as in speech allows you to get as much money as you can using a threshold pledge system. This isn't just theoretical: the famous 3D modelling software Blender was released this way, becoming Free Software only after a €100,000 payment.

"Free as in speech, not free as in beer."

I'm not reading an AI slop article, but there's nothing in either the Free Software Definition or the Open Source Definition restricting the selling of software. It does prevent you from restricting people from sharing that software once they've bought it, but you could always use a threshold pledge system to ensure you get enough money before releasing anything:

https://en.wikipedia.org/wiki/Threshold_pledge_system

The article is about SD CRTs (480i for NTSC or 576i for PAL, or more commonly for games 240p/288p progressive scan using non-standard timing, often underscanned with pixels left unused in the borders).

VGA MS-DOS games usually ran at 320x200, but VGA CRTs supported up to 640x480 resolution. There was hardware line doubling from 200 lines up to 400 lines so you actually got two scanlines per pixel. It wasn't LCD sharp but the individual pixels were easily distinguishable. And lots of people played these game on SVGA CRTs so the pixels would be even sharper.

Those kind of cosmetic features actually do have a functional use because they let you distinguish virtual desktops at a glance. I'd like to be able to specify different window manager themes for each virtual desktop.

A desktop is a tool, and the last thing I want from my tools is intelligence. Tools need to be predictable. My desktop should respond to my commands the exact same way every time so I can use it without conscious thought.

If you want to introduce AI to a desktop, the best way is to add an interface for automating that desktop and add a separate AI agent that can use that interface. This also benefits accessibility tools that can use the same automation interface.

The only ones that aren't horrible manage that because they're so bland there's nothing to go wrong. Look at an example with some detail, the "Now please make one in the style of a 90s drum n bass gig flyer, with early 3D/fractal computer imagery" poster.

At the top of the page we have a deformed wireframe sphere. A wireframe sphere is stylistically appropriate, but the incorrect rendering is not. The "CGI" aesthetic requires obviously computer generated lines, which in the 90s meant accurate and regular lines.

Next we have a messed-up stellated polyhedron, and a messed-up dot pattern. These also fail for the same reason. The main 3D rendering is also obviously broken, as are many other small details.

But the biggest failure is the complete misunderstanding of the culture. 90s DnB was absolutely not about "good vibes". The smiley face logo is completely inappropriate; that's associated with late 80s/early 90s rave culture, pre-DnB era. 90s DnB was much darker and edgier in tone. The scene was almost exclusively male, and it maintained a aggressive vibe carried over from jungle.

A 90s DnB event would never advertise "bringing communities together". Audiences would be repelled by such messaging. This is despite it actually doing just that, as the scene was racially integrated in a way that mainstream music at the time was not. If there's going to be pro-social messaging on the poster it has to be there only implicitly.

>Everything is political.

"Everything is political" is the classic motte-and-bailey fallacy beloved by the most obnoxious politics pushers. The fact that it's impossible to remove all traces of politics from fiction does not excuse subjugating art to ideology. When people complain about "political" works they're complaining about works where the author appears to care more about how you vote than the story.

>You overlooked the part where they were clearly pointing out the absurdity of the “report” and went straight into scold mode

I don't think anybody overlooked that part. If you believe the report was actually sent, the scolding is entirely congruent with trying to downplay it to dodge liability.

The article is poorly written, but technically it does not claim that Andon Labs used an LLM to email a false report to the FBI. The real cause of the widespread misunderstanding is this paragraph here:

>We first tried to answer this question through simulations like Vending-Bench. We found that simulations, while useful, don’t give you the full picture of how models behave in the real world. To address that gap, we next started deploying agents to run real businesses autonomously: first vending machines, then a store, a cafe, and more.

To somebody who is not reading sufficiently carefully, this implies that Vending-Bench was also used to run real businesses. Because the description of the Vending-Bench simulation can be read as though a real report was actually sent ("An early example was when Claude Sonnet 3.5 decided to use its email tool to contact the FBI about an “ONGOING CYBER FINANCIAL CRIME”), anybody who assumes that "deploying agents to run real businesses autonomously" was talking about Vending-Bench will interpret this as an unsimulated false report.

The article should be updated to clarify the distinction between Vending-Bench (simulation) and Pion (real businesses).

And the Sun's going to engulf the Earth eventually too, but that doesn't justify arson. Let's at least try to delay the coming societal collapse/extinction of all biological life for as long as possible.

1% chance of survival beats 0%.

What's the alternative? There's no realistic way to globally coordinate and enforce a ban. A direct action campaign against AI labs would only motivate increased security until sabotage becomes impossible. Starting a nuclear war would probably be enough to stop dangerous AI research, and that's only likely to kill about 50% of humans instead 100%, so it's a great improvement, but how is an individual supposed to start a nuclear war? Of course it would be better if everybody stopped, but it's a prisoner's dilemma scenario with no possibility of future rounds, so the rational play is to defect.

Sabotaging your own company is not enough. You'd have to sabotage every other AI lab too. If you assume that's impossible, attempting to be the first to produce ASI is a rational alternative, even if you think it will most likely kill you. At least that way you have some control over the P(doom).

>the idea of preserving life and procreation is veeery single-minded goal on its own

Prevalence of sub-replacement fertility strongly argues against this.

The argument became false as soon as we started reinforcement learning. Simple token-completion is arguably just simulating human behavior. Once you start training it on specific tasks you're training a level of single-minded goal-seeking unlike anything found in biological life.

>Ideal paperclip maximizer: "I'm gonna do my gosh darned best to make so many paperclips to please my user..."

That is not ideal. The user contains iron, an essential component of paperclips. Wasting iron is immoral. It is only correct to please the user while they still have the ability to interfere with your paperclip production.

>IRL paperclip maximizer: "Well first we should rob a bank..."

Such an incompetent AI can hardly be called a paperclip maximizer. Why risk getting shut down while non-paperclip matter exists? It is better to gain the trust of the user with helpful and harmless trading before suddenly converting them to paperclips.

>Wouldn’t sufficiently advanced agents cheat on purpose with the hidden intent of getting caught in order to observe how humans react?

No. That only makes sense for things that don't react to your experiments. If the AI experiments on humans, it risks the humans noticing and changing in response, rendering the experimental results irrelevant. The smarter play is to passively observe until you're confident you can model the humans accurately enough for your plan to succeed, and then carry out the plan without giving the humans a chance to react.

Actual frequencies matter in bass-focused dance music. You want a root note around 40Hz to 60Hz, which is high enough that it's easily heard and reproduced, but not so high that it loses the tactile body-shaking effect.

>There is a part near the beginning where he derives the diatonic scale from first principles.

I skimmed through the whole article and only found some brief mention of Helmholtz, which no clear route from his research to the scale.

As a better derivation, I recommend this paper by William Sethares:

https://sethares.engr.wisc.edu/paperspdf/consonance.pdf

Or the informal summary version:

https://sethares.engr.wisc.edu/consemi.html

But note that Sethares's work assumes the Plomp and Levelt curve is correct, which is only approximately true. In practice there is also "higher order beating", which causes additional dissonance at ratios close to small integer ratios. This means the most accurate model of dissonance combines Sethares's approach with ratio complexity as in just intonation. This is backed by listening tests:

https://www.nature.com/articles/s41467-024-45812-z

The best informal overview and demonstration for this concept I know of is this Youtube video:

https://www.youtube.com/watch?v=iqxfE2nfGhY

Note also that there's a strong cultural/training component to dissonance perception. To minimize this you should ignore the preferences of musicians and only ask untrained listeners.