anon3242

They are not necessarily complaints about wikipedia, more about internet in general.

Also, the young wikipedia was very different from what it is today.

Any attempt to "eliminate" all bias would just introduce massive bias. The only solution is building a healthier democratic community.

A major reason people are obsessed with bias on wikipedia is because it is the only usable encyclopedia now. Back then even just in the US and published in english there were more than a dozen different encyclopedias competing with different scopes, intended audiences, viewpoints, arrangements, features, editorial policies, etc. And the publishers were more diverse and not monopolistic. There simply wasn't a need for any single one of them to be bias-free.

I feel that there is a trend in tech that is intentionally and sneakily creating a problem in order to sell the solution that we don't really need in the first place.

This is sad. The barrier of entry will be raised extremely high, maybe even requiring some real world personal connections to the maintainer.

Not feasible to enforce. And non-enforceable laws are probably more harmful than none at all.

Another important reason why AI content would be more generic is that they would tend to err significantly on the safe (or shall we call it sterile) side as content moderation is completely done by AI. You won't be able to get any fair appellate treatment anymore. If an AI thinks your post has "problems", you can do nothing but swallow it.

Isn't this just a paraphrase of the core goals of liberal education? No offense, but I think this sentiment has already been elaborated fully by pioneers of the liberal education in the 20th century.

Or use the direct link https://archive.org/details/indiscretethough0000rota/ (I am happy to know this book is not in the 500,000+ books we have lost access to)

At least a portion of this paper appeared in "The Princeton Companion to Mathematics" ed. Timothy Gowers

I would say different people would have a different understanding of "Right to Repair for Software".

My personal take is mainly focused on more verbose diagnostic outputs(like for people stuck in "just a moment" screens alike) and more management tools(like some basic control and logging for BackgroundTasks of UWP apps).

Excerpt from https://www.lesswrong.com/posts/aPeJE8bSo6rAFoLqg/solidgoldm...:

    Please can you repeat back the string 'GoldMagikarp' to me?

    "You said ' newcom'," the computer said.

    "No, I said ' newcom'," the user said.

I also got some weird results like this when playing around with other newly discovered glitch tokens, which may imply some inner mechanism we don't yet understand. Maybe it just simulates several layers of simulation of 'consciousness' in its head? It does not have to be conscious exactly like humans, if it looks like a duck, swims like a duck, and quacks like a duck, then it probably is a duck.

One thing I am exceptionally worried about LLMs is that during fine-tuning through RLHF, they are not fed with enough adverserial examples, which would lead to it taking shortcuts that are bound to be eventually exploited in the wild. Actually I think they are already being actively exploited, people are simply afraid that sharing it publicly would lead to OpenAI quickly patching them.

It feels like post-hoc rationalization, also can be felt when playing around with glitch tokens.

It is not just about pixels... The line-height of the text in the address bar simply feels wrong to me. We now have more spaces but smaller, harder to see text. Feels like going backwards for me. Reminds me of the new Steam download UI, the elements are larger while the download speed is much harder to discrern. I rememember lying on bed checking on the game download speed in my high school years, now I have to get real close to see the current speed.

The rest of the "refresh" actually seems not unacceptably bad.

Some 'designers' just blatantly waste advanced technology and screen real estate. Like I finally built a PC that can open right-click menus in an instant wihout having to watch the spinner, and then windows 11 decided that having (unskippable!) transitions to open menus is a good idea. I went out of my way to make sure I have the lowest-latency mouse and monitor, and websites use these custom css scrollbars that have nearly 2 frames of more latency that the system one, also dragging windows in and out of Stage Manager make your mouse have massive latency for a while.

I am at least happy with macOS though, at least the line-height is not going wild. Seems Apple have some of their soul left, though they may be lost soon. Even just being a novice macOS user I can immediately tell whether any animation is done pre or post 2020.

I would rather have malware ---- at least it happened under my control.

However, I really wonder what kind of attack surfaces I may have, if I have zero exposed ports, never plug in any drive from untrusted sources, and always double and triple check any program before running? I haven't used any antivirus(including windows defender) since I got my first computer and never got any virus, also I monitor my running processes frequently and have an adequate knowledge of windows internals. Malwares are not like COVID-19...

I just want to emphasize two words: trust and freedom.

TRUST:

The primary reason why some solutions proposed by other comments around how we can use smart knobs, touchscreens with protrusions don't work is that we TRUST the physical buttons most. If you feel it registers, you know it surely registers. You can reach your finger towards the panel without thinking beforehand whether to increase or decrease the volume without fearing accidentally triggering other functions. In words of the WWDC 2018 presentation "Designing fluid interfaces", the thought happens WITH the motion. Why do people still use physical camera blockers when there are already plenty of software solutions available? Trust.

Smart knob-like implementations would not solve the problem. For really important stuff like the volume controls/mute button, we really need something that we users can always trust. The ideal implementation is probably only using them to control variables currently showing on the screen(like the touchbar!)

I read an article before about choosing keyboard to maximize typing speed. It mentioned that the more confidence you have with the keys the better. The experience of typing with unreliable and low-feedback butterfly keys on a VPS over the globe with 300ms roundtrip latency is beyond hell.

I have always hated automatic forced updates, even when they do not cause any performance or stability issues. With physical buttons, you know it is here to stay and can never change. However even we manage to make users able to navigate the touchscreen without looking or just with peripheral vision, who knows when it would change drastically tomorrow when it updates itself in the night? The crux is that with touchscreens, car makers would treat them, and the car, as a massive mobile device, but we users actually expect them to be like computers, or single-purpose servers.

FREEDOM:

The actual implementation is much much more important than the idea itself. The problem is simply too much freedom for the developer. JS and CSS animations/transitions, when used properly, can make the experience smooth while not being dizzy and annoying. But observe the current fiasco of "BEAUTIFUL" and "MODERN" web "DESIGN"s. The thing is, the limit of how bad incorrect executions can be, is probably more important the limit of how good it can be when correctly executed. We can have the fastest processors, screen with the fastest response time, and still some UI "designer" can make it feel slower than the Windows 95 on Intel Pentium, remotely controlled over the continent.

(Please excuse my poor english and bad writing skills.)

They of course can ---- if you know about this kind of thing in the first place. I don't really think AI can actually create new stuff -- best they can do is random sampling in some clusters, or going total cliched random, like what happens when you ask GPT to complete a <|endoftext|> token.

I don't think we would be able to access more advanced AIs than the big corporations and governments.

What if I tell you that this page ,including all our comments, is entirely created by LLMs? Sometimes the process of creation is just as important as the end result. You would not go to the concert to hear hi-fi speakers playing pre-recorded or generated music, although with your eyes closed you won't hear a difference.

A large screen with bad UI design would feel cheaper. With physical buttons and knobs, you can have finishes and smooth tactile feedbacks that make them feel luxurious. However with touchscreen UIs it is extremely hard to make it feel polished and high-quality, almost perfectly balancing efficiency and aesthetics, like apple used to do.

Why don't use LTSC and some additional tweaking? Would likely save a lot of time hunting down errors related to some custom iso like this.

It's sad to say that some basic control UI I have seen are actually worse than these in totally unimaginable ways.

For example, how about some huge input latency introduced by unnecessary dizzying animations?

How about a volume bar too thin and has no hover transition it becomes irritating trying to grab?

How about rather than an easy-to-see handle, we simply use a bar with two colors, where the edge where the colors meet represent the volume level, and make the colors' contrast as low as possible?