Isn’t that the essence of the issue, that those models are loaded with biases, that might or might not overlap with dominant ones in inscrutable ways, hence producing new levels of confusion and indirection?

I’ll bet people said the same thing when Intellisense started suggesting lines completions.

I’m sure many did, but I’m also pretty sure it’s easy to draw a line between code assistance and LLM-infused code generation.

Telegram never was private, group chats never were encrypted (and that’s not an opinion: the feature simply is missing). If anything, they are just removing their false and deceiving claims. That they remained there for so long is something I can’t wrap my head around.

I’d argue XMPP is less ideal than Matrix because groups are located on a single server, which makes them easier to take down than Matrix’ replicated state.

That is true, but it’s never been a problem in my relatively long experience with XMPP: some server software can be used as a cluster and distributed, making it highly available (basically, the whole of WhatsApp runs on a fork of ejabberd), and the comparatively tiny resource usage of XMPP contributes to its stability.

XMPP does have a spec for F-MUC (distributed rooms somewhat like Matrix, many years before Matrix) and my rationale as to why it never picked up despite a whole decade of “competition” from Matrix is that it’s a problem that just doesn’t need solving. The price to pay for it is hefty: Matrix resource usage (bandwidth, CPU, RAM) is insane, its protocol complexity makes it a single-vendor implementation (which is risky on very practical grounds), and it’s not even bulletproof for the niche use-case it set to tackle: in the end, your identity server on Matrix remains centralized.

You can tell that I’m partial to XMPP, but that’s only after having been a service operator for years, with my original expectations largely favouring Matrix.

I think you should give Trilium(Next) Notes a try:

• it has the hierarchical notes structure that you are familiar with in obsidian

• it has better ways of keeping things organized (attributes can be values or references, can be shared and inherited, which provides a flexible framework for having notes “types” as templates that can be extended, e.g. people vs. colleagues, businesses vs. companies, etc)

• it has the concept of note hoisting (which lets you focus on a note and its sub-notes, so other projects/spaces don’t come in the way of autocomplete and placing references), and workspaces that builds further on top of that

• it can be used standalone (local client/offline-only, like obsidian) but coupling it with a remote-server opens more interesting use-cases (synching, sharing notes with others by public URLs, one-user/multi-client editing) which gives the best of both worlds (local-first/online-first) and lets you access your personal notes on devices you don’t necessarily own (which obsidian doesn’t). The mobile app story isn’t great (it’s a PWA with limited offline capabilities at the moment), but isn’t worse than the alternatives either (I can’t really work and think long form on a handheld, no matter the editor experience, but perhaps that’s just me).

You need to list out your requirements. What do you want to do? Where do you need your data? Do you care about open source? Self-hosting? Do you have an idea how your content will be organized? Will you ever need to tap into it as data? Etc

Have you tried trilium notes? Not as hyped and polished, but does extraordinarily well IME.

I didn’t like obsidian’s lacking in attributes structuring/typing and the fact that it cannot serve over a web UI (for wherever you cannot install the heavy client or just to share notes via URL), and found trilium notes to be doing that perfectly, and much much more. Highly recommend.

How it works (to simplify) is them giving up on matrix clients ever becoming performant and well behaving on handheld devices (because of the absurd complexity of the protocol), and, instead of doing something about that, just decided to shift the client logic onto the server and castrating the clients (esp. for offline features). It’s also good short-term business because it makes hosting Matrix even more cumbersome and expensive, giving a compelling reason for the type of midscale/corporate deployments previously on the fence about their self-hosting costs (due to poor design and scalability) to just pay Element for that (while probably contemplating an alternative future).

Matrix has the tendency to require all participants’s servers to replicate all of the room state (who joined when, who said what when, whose avatar changed to what when, …) practically forever, and is sucking a ton of bandwidth and CPU for the privilege. It’s pretty bad, unfixable, and, if you ask me, over hyped.

Speaking about XMPP, compared to centralized services, at least the “who talks to whom” and metadata concerns in general are partially mitigated by not having all the metadata converge towards a single host, being able to selfhost, and being able to host behind tor/i2p/…

If you have the impression that there’s a dominant, homogeneous “mass” sharing the same opinion, you are right there in the middle of an information bubble and a victim of those “algorithms”.

Would that make a difference?

With only one having your interests at heart. An easy choice.

The UI of Prusa slicer is hot garbage though.

I give orca/bambu the edge for “prettier on screenshots”, but in practice, I don’t find their UI paradigm to be more efficient nor convenient.

I’d like to share your optimism, but what you suggest leaving us to “deal with” isn’t “AI” (which has been present in web search for decades as increasingly clever summarization techniques…) but LLMs, a very specific and especially inscrutable class of AI which has been designed for “sounding convincing”, without care for correctness or truthfulness. Effectively, more humans’ time will be wasted reading invented or counterfeit stories (with no easy way to tell); first-hand information will be harder to source and acknowledge by being increasingly diluted into the AI-generated noise.

I also haven’t seen any practical advantage to using LLM prompts vs. traditional search engines in the general case: you end up typing more, for the sake of “babysitting” the LLM, and get more to read as a result (which is, again, aggravated by the fact that you are now given a single source/one-sided view on the matter, without citation, reference nor reproducible step to this conclusion).

Last but not least, LLMs are an environmental disaster in the making, the computational cost is enormous (in new hardware and electricity), and we are at a point where all companies partaking in this new gold rush are selling us a solution in need of a problem, every one of them having to justify the expenditure (so far, none is making a profit out of it, which is the first step towards offsetting the incurred pollution).

You can always give a shot at using a third party client (possibly acting as bridge for other/better protocols, like e.g. slidge.im>xmpp or the buggy matrix equivalent), but you need to keep in mind that they will all require you to authenticate (and remain authenticated) using a smartphone, and that usage of 3rd party clients is forbidden from WA’s terms and conditions (which may lead to your account being blocked/deleted).

How about nextcloud with only the bare minimum amount of plugins? Filles alone is pretty snappy.

Pydio used to be called ajaxplorer and was a pretty solid and lightweight (although featureful) solution, but then they rewrote the UI with lots of misguided choices (touch controls and android inspired interactions on desktop devices) and it became so horrendous, heavy and clunky that I almost forgot about it. I wonder if they reversed the trend (but from the screenshots it doesn’t look so).

Aren’t they not the same thing at all?