pet_the_bird

> And what about dance music? Isn’t that entirely gridified? No. Not by the definition being used by the Musical Form Institute. > > What!? > > The next essay will cover dance music. > > To recap: gridification is not playing to a timing reference. It is correcting human performances to a fixed grid after the fact. I felt that the essay acknowledged differences in production between genres. Tempo is used as the main example, but they also mention vocal tuning as an example of gridification.

Yes, we should reject these words as analogies for AI, because they heavily influence the understanding we form of these systems. (The Language Labyrinth: Constructive Critique on the Terminology Used in the AI Discourse by Rainer Rehak is a great read, although quite academic).

My critique in this case was more on how a essay focussing on language uses language that heavily incorporates the values at the core of these companies (efficiency, measurable improvement, transhumanism, individualism). We should investigate the language that have created the circumstances of these companies and how these companies have affected our language.

I am glad it has, and I think that recognizing the affordable access to welfare, goods, and food has been a valuable gain.

Call me a problem person, but I would say the tech companies you are referring too are perpetrating significant harm, but maybe you do not directly experience this. As an exercise I can use terms that we normally restrict to governments.

Google controls arguably the worlds most advancement propaganda system^1 ever created, has monopolistic control over information access worldwide^2, has farreaching influence on academic research^3456 and can perform censorship on major communication and education sites without overshight or juridicial control^7. They are a dictatorship with liberal capitalism as their ideology.

A world controlled by these businesses may have been beneficial for you so far, but they systematically disempower us. By yielding power and influence to them, we lose the ability to decide how we wish live and how we shape the systems around us.

[1] https://www.justice.gov/archives/opa/pr/justice-department-s... [2] https://en.wikipedia.org/wiki/United_States_v._Google_LLC_(2... [3] https://focs.computer.org/2025/ [4] https://neurips.cc/sponsors/sponsors-list?year=2026 [5] https://www.usenix.org/supporters [6] https://find-and-update.company-information.service.gov.uk/c..., 02 Oct 2025 Full accounts report Google providing £1.325 billion for Deepmind activities [7] https://en.wikipedia.org/wiki/Censorship_by_Google#Advertise...

I like the approach that the writer takes, emphasizing the importance of humanistic experience into technology. Language will be the deciding factor for what technology becomes, and on this point I have some critique of the language that the author uses.

> Become the most flexible, underscore-shaped mind you can cultivate, the best-prepared proto-diplomat and translator of cultures, the family philosopher (because you no longer need someone to fix your printer).

In previous sentences, but here explicitely, the language used reflects that of the AI labs themselves. We do not need to find a metric to reduce our human experience to. We will not address systematic issues by pushing ourselves to become perfect. We need to start talking to our friends and families and question why we use the words we use in technology.

I am not sure I see a fundamental distinction between HN censoring your speech on their platform or the government preventing you from speaking in public. Both are censoring you right? Is the difference the fact that HN is free to regulate its own platform as it wants to?

For a large part I see people using digital platforms as a primary method of expressing their thoughts and opinions. Governments in both the US and the EU have imposed restrictions on what platforms may host. I think we have not reconciled yet how to reinterpret free speech online.

Im no expert on free speech, of free speech absolutism, so please let inform me :)

I am personally concerned by what defensive can mean. Alignment of these models is inherently a non-neutral proces, and currently what values are reinforced is decided by a few OpenAI engineers. I feel that any 'defensive model' will further ingrain current values and actively resist the natural progression of our society. This is especially the case for any use of these models for policing or military.

Maybe people are rightfully concerned about the capabilities of the models of other (non/less democratic) states. But if we are concentrating power in the hands of few and at the same time allowing the creation of a weapon that thwarts any offense, how do we ensure the health of our democratic societies?

The Dutch University of Delft systematically 'maximized' grants and shuffled the money between projects, according to investigative journalists of the NRC[1].

1 https://www.nrc.nl/nieuws/2024/02/15/het-subsidiepotje-moet-...

Highy anecdotal: I have tried various self-hosted models using both vllm and llama.cpp. I am in a situation where I have access to large amount of memory (~320 GB).

While experimenting with quantization I found that there is a non-trivial tradeoff between quality and memory footprint. Overall my experience follows the reported pattern of "2-bit is mwah, 4-bit half decent and 6-bit required for programming. Still, although MiniMax-m2.7 is useable with the 6-bit quantizations that unsloth provides, it felt like such a breath of fresh air when I used the reference full-size model.

I find it difficult to say why. I had mostly the same setup as before (parsing had to be slightly adjusted in Zed). Aside from not experiencing the thinking loops (where minimax would get stuck generating the same sentences over and over) there is little evidence of any real improvement (although the average thinking time felt shorter).

I would recommend against very low quantizations of GLM 5.0/5.1/5.2 or Kimi 2.5/2.6. Smaller models were more reliable, and therefore more useful.

"Cryptography Made Simple" By Nigel Smart and "A Graduate Course in Applied Cryptography" by Dan Boneh and Victor Shoup are excellent resources for people that have affinity with Math and CS. The second resource can be a tough read, and I would strongly recommend not skipping the first few chapters.

I think the article tried to refer to this link https://arxiv.org/abs/2504.10663 As I understand from scanning the paper, the authors attempt to determine differences between the Russian wikipedia articles and the articles on the Russian fork. They show that articles on the fork that were that differ from RU wikipedia have a significantly higher number of edits on RU wikipedia. The authors suggest that these may be signs of manipulations, however, it may not have affected the quality negatively (as stated in the discussion).

I do not find state sponsored activity on Wikipedia unlikely, but I am not convinced there is clear evidence that Russia poisoned wikipedia succesfully.