I'll throw in another "+1, quite satisfied with immich" comment, because I'm honestly that impressed.
The project as a whole feels competent.
Stuff that should be fast is fast. E.g. upload a few tens of thousands of photos (saturates my wifi just fine), wait for indexing and thumbnailing to finish, and then jump a few years in the scroll bar - odds are very good that it'll have the thumbnails fully rendered in like a quarter of a second, and fuzzy ones practically instantly. It's transparently fast.
And the image folder structure is very nearly your full data, with metadata files along side the images, so 99% backups and "immich is gone, now what" failure modes are quite easy. And if you change the organization, it'll restructure the whole folder for you to match the new setup, quietly and correctly.
Image content searching is not perfect (is it ever?), but I can turn it on in a couple clicks, search for the breed of my dog, and get hundreds of correct matches before the first mistake. That's more than good enough to be useful, and dramatically better than anything self-hosted that I've tried before, and didn't take an hour of reading to enable.
It's "this is like actually decent" levels that I haven't seen much in self-hosted stuff. Usually it's kinda janky but still technically functional in some core areas, or abysmally slow and weird like nextcloud, but nope. Just solid all around. Highly recommended.
I largely agree, I think that probably is all that it is. And it looks like shit.
Though there is a LOT of room to subtly train many kinds of lossy compression systems, which COULD still imply they're doing this intentionally. And it looks like shit.
No, the comment correctly points out that the "Soup" button (and all of its siblings... the food categories) is inoperable when JavaScript is disabled. You're stuck with "All" instead of nice filtering. There are ways to achieve this without JavaScript.
ah, gotcha. that would be nice, but tbh it seems rather minor? anyone who knows how to disable javascript (an extremely-clear "power user" signal) can probably be expected to know how to search a page too.
It's just peculiar especially since the repo readme [0] specifically says "Minimal JavaScript - Only when truly needed" and yet it's used for this food category filtering, a feature for which JS is certainly not truly needed.
from what I've seen in a several-thousand-eng company: LLMs generally produce vastly more code than is necessary, so they quickly out-pace human coders. they could easily be producing half or more of all of the code even if only 10% of the teams use it. particularly because huge changes often get approved with just a "lgtm", and LLM-coding teams also often use/trust LLMs for reviews.
but they do that while making the codebase substantially worse for the next person or LLM. large code size, inconsistent behavior, duplicates of duplicates of duplicates strewn everywhere with little to no pattern so you might have to fix something a dozen times in a dozen ways for a dozen reasons before it actually works, nothing handles it efficiently.
the only thing that matters in a business is value produced, and I'm far from convinced that they're even break-even if they were free in most cases. they're burning the future with tech debt, on the hopes that it will be able to handle it where humans cannot, which does not seem true at all to me.
Measuring the value is very difficult. However there are proxies (of varying quality) which are measured, and they are showing that AI code is clearly better than copy-pasted code (which used to be the #1 source of lines of code) and at least as "good" (again, I can't get into the metrics) as human code.
Hopefully one of the major companies will release a comprehensive report to the public, but they seem to guard these metrics.
essentially every YouTuber I've watched who discussed their financials said that their sponsorships brought in several times more money than all forms of YouTube money.
which is a very niche slice, and I have no idea how representative it is in aggregate. but sponsorships happen because they pay well enough to annoy every viewer, not just ones that aren't using the better-paying Premium - they generally are not cheap, to say the least.
If you look at Premium, it's about 100x more lucrative than regular views. So I'm pretty sure I'm providing more money to creators than the skipped ads.
To be clear: I completely believe that Premium is a major source for many people. 100%. I just haven't seen many examples of it, outside tubers that have zero sponsorships (because they're small and/or not doing the low-value slightly-shady ones that get spammed everywhere). I'm thrilled that Premium seems to pay relatively well, it's better for everyone to move away from ads where possible.
LTT though is a rather significant outlier in terms of subscribers (16.6 million right now). For truly large channels it's reasonable for the equation to be different.
And the equation for them really is different. They're a company with ~100 employees¹ and YouTube and video sponsorships came out to just 11.6% (ads AND premium) and 9.2% respectively of their multi-person company income. People claiming "SponsorBlock steals from creators" aren't talking about LTT, they're talking about smaller creators for whom YouTube stuff is a majority of their income.
Plus, like. Ads+premium lumped into one. It wouldn't surprise me if premium was lower than sponsorships.
I suspect we're in different niches then or something. If ya don't mind sharing / have links handy, do you have any examples? I'm curious what kind of channel it works well for.
I can try to hunt mine down, but most of the examples I've had were from a couple years ago, and YouTube's history is rather hard to search for stuff like that :| Not high odds of success.
These are niche Russian-language channels (@Varlamov, @Max_Katz). They disclosed their finances to drive up Patreon/Youtube subscriptions because Youtube stopped monetization from Russia.
I've been trying to find public numbers for English-language channels, but wow. So much slop.
yeah... and it's (partly) based on the claim that it has network effects like how Facebook has? I don't see that at all, there's basically no social or cross-account stuff in any of them and if anything LLMs are the best non-lock-in system we've ever had: none of them are totally stable or reliable, and they all work by simply telling it to do the thing you want. your prompts today will need tweaking tomorrow, regardless of if it's in ChatGPT or Gemini, especially for individuals who are using the websites (which also keep changing).
sure, there are APIs and that takes effort to switch... but many of them are nearly identical, and the ecosystem effect of ~all tools supporting multiple models seems far stronger than the network effect of your parents using ChatGPT specifically.
>To avoid commingling, sellers have long had the option to apply a unique, seller-specific Amazon barcode — known as an FNSKU (Fulfillment Network Stock Keeping Unit) — to every product. This ensures their inventory is tracked and shipped separately.
... is that really all that was necessary all along? I can see that being a problem for, like, 10 cents worth of stuff, but a lot of the commingling complaints have been around expensive items. It's not zero cost of course, but for your average $30+ thing it doesn't seem very difficult to justify.
The project as a whole feels competent.
Stuff that should be fast is fast. E.g. upload a few tens of thousands of photos (saturates my wifi just fine), wait for indexing and thumbnailing to finish, and then jump a few years in the scroll bar - odds are very good that it'll have the thumbnails fully rendered in like a quarter of a second, and fuzzy ones practically instantly. It's transparently fast.
And the image folder structure is very nearly your full data, with metadata files along side the images, so 99% backups and "immich is gone, now what" failure modes are quite easy. And if you change the organization, it'll restructure the whole folder for you to match the new setup, quietly and correctly.
Image content searching is not perfect (is it ever?), but I can turn it on in a couple clicks, search for the breed of my dog, and get hundreds of correct matches before the first mistake. That's more than good enough to be useful, and dramatically better than anything self-hosted that I've tried before, and didn't take an hour of reading to enable.
It's "this is like actually decent" levels that I haven't seen much in self-hosted stuff. Usually it's kinda janky but still technically functional in some core areas, or abysmally slow and weird like nextcloud, but nope. Just solid all around. Highly recommended.
reply