Some (not all) answers I was able to find in Maryanne Wolf's book "Reader, Come Home". Main concept there is what she calls deep-reading, a complex back and forth between different brain areas and the hemispheres, that needs time and is replaced by specific forms of skimming on the mentioned devices. In particular, the shift between hemispheres allows for integration into the reader's personal store of knowledge and, more broadly, into their own worldview.

The "wasted" time is essential for memory building and consolidation. Add enforced linear reading without immediate availibility to break the flow by googling/notifications/jumping to whatever, also consider haptics and more. Similar effects can also be found in handwriting vs typing, manual sketching etc.

At _this_ moment, AI is in the state of producing things - if you like with factor 10 or more. But what will come afterwards, when all this mush of code shall create _reliable_ results. This means not man month then, rather man years or decades to fix this billion and maybe trillions lines of opaque probabilistic LOC. You have to take the mean of these two stages, if nothing qualitative happens to the models.

Most of this I didn't for many years because it is not my core competence (in particular the security aspects). Properly fleshed-out explanations from any decent AI will catapult you to this point in no time. Maintenance? Almost zero.

p.s. Admittedly, it's not a true self-hosting solution, but the approach is similar and ultimately leads to that as well.

http://bactra.org/notebooks/nn-attention-and-transformers.ht...

https://bugs.kde.org/show_bug.cgi?id=423497

Given the changing voting behavior in Germany, these personal complaints seem to be quite widespread.

https://www.google.com/search?udm=2&q=ensikat+illustration+h...

There is no way back. Relaxed posture, no UI elements stealing my focus unnoticed, parallelism (partially): continually "big-picturing" text; speaking with people while typing. The rhythm of this motoric skill and his quite specific form of memory alone, strangely decoupled from and coupled to the other mental processes at the same time, the interplay is simply marvelous.

After that, try Principal Component Analysis and look, what remains from these dimensions and the labels describing them. Think up names for the Eigenvectors / new axes. Investigate further. For example, look where people are concentrating in this high-dimensional space.

https://docs.astral.sh/uv/guides/integration/pytorch/#instal...

- uv tool replaces pipx etc.

- uv pip --tree replaces pipdeptree (including 'inverse' mode)

- ...

That's intentional, and I like it. It limits the context dynamically to what is necessary (of course it makes mistakes). You can also add files with placeholders and in a number of other ways. but most of the time I let Aider decide. It has a repomap (https://aider.chat/docs/repomap.html), gradually building up knowledge and makes proposals based on this and other information it gathered also with token costs and out-of-context-window in mind.

As for manual changes: aider is opinionated regarding the role of Git in your workflow. At first glance, this repels some people and some stick to this opinion. For others, it is exactly one of the advantages, especially in combination with the shell-like nature of the tool. But the standard Git handling can still be overridden. For me personally, the default behavior becomes more and more smooth and second nature. And the whole thing is scriptable, I only begin to use the possibilities.

In general: Tools have to be learned, impatient one-shot attempts are simply not enough anymore.