In any case, you've inspired me to post the original reply I had composed for https://news.ycombinator.com/item?id=48575653 (the immediate parent to your comment), below. This is what I wanted to say, before then deciding to just be grateful for the sharing of the parent's perspective:

""" I'd frame AI as a plausible hypothesis engine, not as a working scientist (yet). I think AI can do some things that look like rational analysis (far better than many/most humans much of the time, perhaps), but I reserve that (most rare and prestigious and important) activity of actual science for humans too, when it counts, for sure.

I get the main article is about the very real "chaos/threat" of no funding, not the "chaos/threat" of AI-articles/"research" nor the "chaos/threat" of "real issues in the state of Science (before funding crises)".

IMO, the state of Science (before funding crises) could be, perhaps, inextricably (though not overtly) linked to the later/current chaos/threat of markedly reduced funding. No? Maybe it's not stated anywhere, but it seems oh so likely, reading between the lines.

If funding cuts, in the medium to long term, lead to a good thing (which would be the best we can now wish for -- and, after all, everything comes and goes in and out with the pendulum of time), it will be a much needed "reset" of science onto a more honest (and net knowledge-learning productive) model.

It (Science) was, arguably, already well by the wayside. Not just sort of expensive (though not very, compared with other budget items). But more importantly: inefficient (to put it nicely). And more importantly still: often (perhaps more often than not!) plain wrong. And that means, sadly: fairly/largely ineffective (degree depending on the domain). Which is the opposite of what is wanted. If you're going to do Science, it should at least be valid, or if it's not, it should be possible for those in their own field to tell that it is not. Else, it's kind of broken.

And if it doesn't serve it's purpose, what can you do but reboot. Reset. Just like a computer.

At least Science can be rebuilt. You just start doing it again (with what you have/can). With more rigor.

Maybe this reads like more of the same. But I don't think "being well funded" correlates well with "doing good science". (Only if the science is measured by the paychecks. Which is economics, not science.) """

At least it's a good thing that we're able to a) observe and b) talk about and c) acknowledge openly(ish) that academic, mainstream, practicing "science" (including as visible to microscopists and all that entails) is currently a "mess".

This allows us to, eventually, address those issues (or die trying!).

Science used to move at a pace of one lifetime after another (pearl clutching 'til the end and confirmation bias and careers built on saving face and economic entrenchments all that).

But I hypothesize that with AI, we can point to "a thing that is not a person with all that is bundled up with that" and say "look, maybe this other train of thought is worth entertaining". Not to say the AI is right. Ideas will stand or fall on their own merit. Just that an AI is not a person outside the field. Normally, an outsider says something, nobody listens. But, if an anonymous AI says something (of course, cleaned up for voice and concision and validated by a human as a first pass), the worst you can say is "ok prove it" or "here is where that is wrong". Instead of: deafening silence.

In other words, I hope AI augments our ability to have those hard conversations that need to be had. Without people losing their jobs due to their prior (understandable) errors, and within the spirit of always using the best available information.

I shared this optimistic indirect use-case for AI with (less technical) friends recently, and they literally were speechless and finally one person said "you're the only one who thinks that".

Am I right, though? There's a there there, isn't there?

I don't want to self-host my photos, too much management.

I don't want to use Apple or Google or MS Clouds for various reasons.

I do want to support a pure-play, independent-ish, profitable, consumer friendly platform. To upload hires shots and have them easy to tag/share/access among those to whom they may be of interest.

I expect to pay, but not through the nose. Reliability matters. I would want the company to be around in 20 years and still reasonably priced/useful. Suggestions?

I liked Maemo 5. Having never used the n810 nor Maemo 4, I suppose didn't know what I was missing.

iPhone did not get copy-paste until mid-2009!

Eventually my n900's microSD jack broke off the board, which I've read was common.

As an original N900 user, I got one of the eBay "refurbed" N900s from China I think a few years ago for fun. It was a piece of junk, literally, like arrived with broken keyboard etc. A clear case of false advertising. I got a full refund.

YMMV. I was really thinking I was buying a proper refurbed N900. Maybe they're out there. Buyer beware.

If so, how is the tradeoff justified? (Make money first, then do "ideal" things?)

If not, why not? (Other than that it's unsuccessful strategically/statistically and wasteful, I guess.)

Any elaboration/response on this theme would be appreciated. Thanks!

The simple layout/theme does not get in the way of the reading.

As far as I can tell, the modern alternative is to set up a MS Power Automate flow to watch the folder and send alerts. Which to me seems a bit clunkier in initial setup and perhaps even in maintenance/durability. This means having the Power Automate product (may cost extra?), setting up some kind of graphical workflow (if the built-in AI doesn't build it right the first time, which for me it doesn't seem to), testing the workflow to see if it as designed right / works (having now built a specific solution using a generic engine), and maybe maintaining that workflow if/when the modules used in the flow get deprecated/retired/changed in the future.

A May, 2025 announcement was covered at [1], but with the full "end" is approaching next month, it seems like news again.

[0] https://techcommunity.microsoft.com/blog/spblog/sharepoint-a...

[1] https://news.ycombinator.com/item?id=43967776

Comments URL: https://news.ycombinator.com/item?id=48416353

Points: 2

# Comments: 1

Comments URL: https://news.ycombinator.com/item?id=48391267

Points: 2

# Comments: 1

I really hope Komai start getting built for macOS.

That's for signaling/logging you "really are" someone. That has a place and can be quite useful, e.g. when that is important. I don't really think HN one of those places.

HN thrives on pseuds, for good reasons.

HN doesn't even need/use "sign in by Facebook". HN is quite simple, and it works well enough as designed.

That would be like, for OfficialProfileHackerSocial.com or something. It could work, but it would no longer be HN.

Seems there are other options, too, e.g. Garage [5].

I only used MinIO briefly in the past, but it was good/worked, and I do plan to set up S3 blob storage again soon.

[0] https://github.com/seaweedfs/seaweedfs

[1] https://github.com/versity/versitygw

[2] https://blog.elest.io/rustfs-vs-seaweedfs-vs-garage-which-mi...

[3] https://news.ycombinator.com/item?id=47806348

[4] https://github.com/RustFS/RustFS

[5] https://git.deuxfleurs.fr/Deuxfleurs/garage

1. Go to https://en.wikisource.org/wiki/1911_Encyclop%C3%A6dia_Britan...

2. Click button "Search the 1911 Encyclopædia Britannica". This currently goes to the page at https://en.wikisource.org/wiki/Special:Search?search=&prefix...

3. Enter the search term and click Search. (There is auto-suggest for some topics, but Search button seems to give more complete results.)

4. Get to the text page of interest, such as https://en.wikisource.org/wiki/1911_Encyclop%C3%A6dia_Britan...

5. Notice the left margin contains hyperlinks like [105] whwere 105 is the page number nd links directl;y to the side-by side view of page 105. Click the [105] link on the left (for example), to get to https://en.wikisource.org/wiki/Page%3AEB1911_-_Volume_02.djv... which shows the text-and-image side by side (for that page).

This flow avoids the hunting-for-the-right page step, by using the direct links.

TFA does mention the word "clipboard", but the containing sentence, while perhaps technically correct, seems a bit misleading, as follows: "As implemented in macOS (both as guest and host), there are also extensions to support keyboard and pointing devices, a shared clipboard, and high-performance graphics with Metal and GPU support." As I understand it, even if those extensions "exist", what good are they if they are not adopted?:

- If you virtualize macOS within macOS on Apple Silicon using UTM, you cannot copy paste between systems reliably (bidirectional shared clipboard is very, very fragile; "can" work a little but is essentially fully broken/unreliable).

- If you virtualize macOS within macOS on Apple Silicon using Parallels (often considered a best-choice solution), you cannot copy paste between systems at all (bidirectional shared clipboard is an explicit non-feature at this time).

Thus, if you want bidirectional clipboard, on a macOS host, you'll have to run a *nix (seems to work) or maybe Win (I haven't tried) guest OS.

I would have tried VMWare Fusion (free for personal use), but after jumping through all the signup and download navigation hoops, I couldn't even get the download link to un-gray itself out for me. Is bidirectional macOS to macOS clipboard implemented there? IDK and I cannot tell.

That's where my thoughts went, too. I can make SMB "better" but not "great" usually, but it's annoying to have to look up and apply, and still have things not optimal. Just in case, IIRC I find this the most useful:

  defaults read com.apple.desktopservices DSDontWriteNetworkStores
  defaults write com.apple.desktopservices DSDontWriteNetworkStores -bool TRUE

As an example flow (since it took a minute to figure out): we can start at https://en.wikisource.org/wiki/1911_Encyclopædia_Britannica then click to navigate/browse volume > section > topic to get to a text page, then click Source tab, then click a Page Number (maybe hunt around for the correct page number), and see the parallel view, text + image. With previous and next page buttons available, retaining the parallel text + image view.