This makes sense

No sense in getting rid of hardware that is working. I’m not familiar with ersatztv but for all the other stuff I am able to handily run it on a 10th gen intel build that is also handling nas duties fwiw. And some stuff is not ideal (cctv is handled via blue iris, which runs in windows VM, everything else is docker)

for the gpu it really depends on your needs. How many users is the big one. If you have at most 2-4 concurrent users and that is an uncommon scenario the gpu is a waste of power, money, and thermal management. Igpu will sip power and transcode (depending on library content, again av1/vp9 on a 10th gen isn’t happening) with that user load assuming you have a decent amount of ram (I have 32gb so you don’t need absurd amounts).

However if you have a lot of users hitting you, 5-6+ or more concurrent streams that all transcode, then you need to start evaluating a discrete gpu (and maybe a significant internet connection bc damn). Alternatively you can suggest your users get something like a ugoos am6b+ flashed with coreelec or a similar setup that can just direct play basically anything but that’s a bit challenging to setup

So then it may be as simple as buying some e waste pc to use a server and using the nas as its intended purpose. Frankly this is probably better, it’s worse power wise but having the storage separate from services has advantages

Define goals. What services can’t be handled?

If transcoding is a goal build around intel. Quicksync video is a no brainer, imo. GPU is unnecessary power draw (15-25w+ idle depending on card) and waste of a pcie slot unless you want to do LLM stuff. Imo 10th gen intel is the sweet spot for quicksync unless you desperately need av1/vp9. If so then you need much more expensive 13/14 gen, which use more power and have more considerations for thermal management

OS is an endless debate. Proxmox is fine and free, why not try it? Unraid is easier to get your bearings but it does cost money. Debian is also free but a bit more confusing because not purpose built. Truenas as well. All can do containers and VMs, but approach in different ways. None is “best” but some are more “free” which is nice

CPU specs are dependent on goals. For transcoding as said above quicksync is necessary and is so impressive. I can transcode a 4k remux to one device while transcoding a 1080 remux to another and direct playing a 4k remux and cpu sits under 25% load on Xeon equivalent of 10700. You don’t need a Xeon btw, I just got a great deal where this was $50 (see next point). Otherwise specs depend wildly on what you plan to do. I can run windows VMs pretty well with this though for the handful of times I need a windows machine

Prebuilt is a waste. Used hardware is cheap and gives more options and can plan more. What are you willing to buy now and what do you eventually want? My NAS started as a 36tb array with 16gb ram and no cache, now it’s 234tb and 4tb cache with 32gb ecc ram years later. Slowly building up was easier on wallet and used hardware, refurb drives, etc is 100% the build. Your goals will likely vary but figure out your roadmap and go from there

Also keep in mind that not every service benefits from running on a NAS. My homeassistant server is run on a raspberry pi for example. Easier to keep it segregated and don’t have to worry about getting zwave/zigbee/mqtt/etc all working with a docker plus dealing with any server downtime impacting home. Tbf literally everything else is run on the nas though haha

I’ve been happy with reolink cameras fwiw though not 100% so. They do have some nonsense though

I also prefer Lutron Caseta for lighting. It’s fairly bulletproof (I’ve literally never had any connectivity issues in like 6+ years) and they haven’t pulled any tos nonsense as far as I know. Downside is pricey and the install is more complex than typical iot stuff. And while they can control outlets they are only rated for 10A lighting so keep that in mind.

The only internet requirement for both of these (not always with reolink I think but at least with the cameras I have) is that you have to allow internet once during initial setup to pair devices. Once that is done you can remove internet access and delete the app

The common thread with these is wired too. The further along I go the more I realize that 2.4ghz WiFi iot shit is garbage. going from WiFi cameras that had privacy concerns and disconnected to local only poe cameras that just work was very nice. Learn from my mistake, don’t buy bullshit eufy cameras that you then have to sell at a loss.

And for your own sanity don’t try to get smart smoke detectors. Your options are either Google/nest that apparently does work well (never tried it, fuck Google), the new kidde that is built into amazons ring platform (never tried it, fuck amazon, plus the preceding model had awful reviews), or the new firstalert that is replacing the Google/nest (again, fuck Google, but I did try the preceeding first alert and it was atrociously bad).

I mention this because this brings up a key issue with regulatory compliance in the US (and probably EU, dunno). You can also try a number of off brand detectors as well that apparently work a lot better. If you search amazon for smart detectors you’ll see stuff like x sense and these apparently have somewhat solid reviews and work okay (though getting them to work in HA is mixed).

However, what amazon fails to mention is that these types of detectors have not been submitted for regulatory compliance in the US (unlike Kidde, firstalert, etc that you’d find at a home depot). They “meet UL requirements” but they have not been submitted for testing so they cannot print the UL logo on the box (legally) but they can write “meets UL requirements”, which is misleading. Fuck amazon and fuck the us government for giving them no culpability in selling obscenely dangerous bullshit

This means if you use these and your house burns down your insurance could technically nullify your policy for not having adequate protection. Or they could not work and you could die, of course

There are smart relays you can tie into an interconnected smoke detector circuit using normal smoke detectors that are appropriately rated if you do want alerts on your phone. There are also device that will listen for chirps but these get false positives

you’re right, my bad

This has been my approach and it has gone okay so far except for 2 issues that are quite a pain:

1: you have to thoroughly research what you buy. Does it work on an isolated vlan? Just because it works with home assistant does not guarantee this. Many home assistant users are comfortable with some degree of data collection and an integration does not mean that it will work local only (nor does it mean that all features will work). If it does work local only you may sacrifice some features. Cameras are a good example. Most cameras with object/person detection do this in hardware, but not all. If you circumvent the Internet connection and proprietary app you may sacrifice this, or more likely alerts

2: there is 0 regulation binding a vendor to the terms of service agreed to at the point of sale, including making significant and sweeping changes. Case in point: I got a chamberlain myQ garage door opener. It worked well and opened my garage door. Integrated with home assistant via the API. However, chamberlain serves a lot of ads for upsells and services via their shitty app. They decided that users circumventing the app and not seeing that you could give amazon drivers access to your garage to deliver packages (seriously) or buy shitty cameras was unacceptable so they updated the TOS and revoked API access for all users. The only way it works now is via their app. I sold mine and built a ratgdo

Another example is Philips hue: while they have been able to be used local only for over a decade Philips has decided they’re going to start a subscription security service with all the devices that entails based around the hue hub. At some point in the near future if your hub updates it will require you to sign in to a Philips account and be online. This one’s way worse as some people have thousands of dollars invested in hue. I have like $300 in the fancier white hue bulbs but some people on the HA forums and reddit literally have their house decked out with like 80-100 bulbs, many of which are the RGB. Kind of silly but they do work very well, flicker free, good color, and last ages. I still have some from like 2016 going strong. Luckily here if you have the bridge on an isolated vlan it won’t update and worst case the bulbs work with zwave zigbee but the principle of the thing is ridiculous. It should be illegal for a company to change the terms this far after the contract of sale

Other examples too. Many car manufacturers (Mazda, Chevrolet, ford) because api access limited data collection for them to sell, some companies are openly hostile to home assistant and when an integration is created they will go out of their way to break it (Ariston, bambu), etc. see https://github.com/unixorn/internet-of-trash

There’s a huge degree of separation between “violent music/games has a spurious link to violent behavior” and shitty AIs that are good enough to fill the void of someone who is lonely but not good enough to manage risk

https://www.cnn.com/2024/10/30/tech/teen-suicide-character-ai-lawsuit

“within months of starting to use the platform, Setzer became “noticeably withdrawn, spent more and more time alone in his bedroom, and began suffering from low self-esteem. He even quit the Junior Varsity basketball team at school,”

“In a later message, Setzer told the bot he “wouldn’t want to die a painful death.”

The bot responded: “Don’t talk that way. That’s not a good reason not to go through with it,” before going on to say, “You can’t do that!”

Garcia said she believes the exchange shows the technology’s shortcomings.

“There were no suicide pop-up boxes that said, ‘If you need help, please call the suicide crisis hotline.’ None of that,” she said. “I don’t understand how a product could allow that, where a bot is not only continuing a conversation about self-harm but also prompting it and kind of directing it.”

The lawsuit claims that “seconds” before Setzer’s death, he exchanged a final set of messages from the bot. “Please come home to me as soon as possible, my love,” the bot said, according to a screenshot included in the complaint.

“What if I told you I could come home right now?” Setzer responded.

“Please do, my sweet king,” the bot responded.

Garcia said police first discovered those messages on her son’s phone, which was lying on the floor of the bathroom where he died.”

So we have a bot that is marketed for chatting, a teenager desperate for socialization that forms a relationship that is inherently parasocial because the other side is an LLM that literally can’t have opinions, it just can appear to, and then we have a terrible mismanagement of suicidal ideation.

The AI discouraged ideation, which is good, but only when it was stated in very explicit terms. What’s appalling is that it gave no crisis resources or escalation to moderation (because like most big tech shit they probably refuse to pay for anywhere near appropriate moderation teams). Then what is inexcusable is that when ideation is discussed with slightly coded language “come home” the AI misconstrues it.

This results in a training opportunity for the language model to learn that in this context with previously exhibited ideation “go home” may mean more severe ideation and danger (if character.AI bothered to update that these conversations resulted in a death). The only drawback of getting that data of course is a few dead teenagers. Gotta break a few eggs to get an omelette

This barely begins to touch on the nature of AI chatbots inherently being parasocial relationships, which is bad for mental health. This is of course not limited to AI, being obsessed with a streamer or whatever is similar, but the AI can be much more intense because it will actually engage with you and is always available.

“Mattel’s first AI product won’t be for kids under 13, suggesting that Mattel is aware of the risks of putting chatbots into the hands of younger tots. … Last year, a 14-year-old boy died by suicide after falling in love with a companion on the Google-backed AI platform Character.AI”

Seems like a great idea

I mean this is shockingly easy to find. Literally his last point is “type name in ‘john doe’ in google search engine” and unless the person has a very common name or a strong online presence that alone will pull up tons of these sites

These sites are pretty gross and should probably have some kind of regulation. They ultimately are just compiling public records but they make things shockingly easy and are ultimately used for domestic abuse, stalking, and shit like swatting and doxxing more than any legitimate purpose. As others have said, maybe the fact that this has led to politicians being targeted will mean that regulation finally occurs but given the current climate and the fact that the targeted politicians were on the wrong team maybe not

This is already a serious problem imo

My counseling practice is geared at queer people so I don’t get a ton of sports gambling types generally but I do get some and the amounts they reveal spending is mind boggling to me. Like 20-30% of income

I have a colleague at a more affluent practice in the city I used to work in that is geared towards “men’s issues” and he sees this far more. We’ve discussed it a few times now. He tends to get more of the bro type

It’s silent too, very often hidden from partners. I increasingly hear about it as a source of stress in relationships too: someone comes in one day and everything was going okay but their shit is blown up because it turns out their brofriend is spending 1800 a week on parlays

Deregulation works!

A text file encrypted with aes-256/gpg on a usb drive with a hidden veracrypt volume that also have a decoy encrypted veracrypt volume is essentially uncrackable even by state actors unless you give up the passphrases or use some dumbass passphrases that are easily guessed.

This is the kind of shit where if you have critical evidence on the drive they’ll hold you: there was a police sergeant in Philly who (probably) had child porn on his macbook. They couldn’t prove it becuase his drive was encrypted and he refused to give up the passphrase. He was held for four years on contempt charges and they finally gave up. He wasn’t even using an esoteric approach, he just had macos filevault turned on

The approach I described sidesteps that issue because it creates a decoy encrypted veracrypt for them to find and for you to give up. The hidden veracrypt with your salacious details is a volume that is created within the free space of the outer volume so it essentially undetectable unless you give it up and by design you give up the outer volume that shows innocuous data. Of course, having veracrypt at all invites suspicion of a secret volume but this puts you in the above spot: catching a few years with no real charges for refusing to give yourself up. Much better spot than catching charges for whatever crime (or being executed for political activism/journalism/etc)

Meanwhile this guys journals were immediately found and well documented for his federal complaint. You could just burn them I guess but then you no longer have all your data

There is no allegedly, the dude was a total boomer and literally wrote out instructions on how to dox someone in his notebook which are shown in the federal complaint against him

Literally outlines which sites are free, which ones have free trials, etc

Like you you would think he would just use a text document or bookmarks or something

Donations isn’t going to cover the hunger of a 40 million dollar VC round. Those investors want more than a return, they want plex profitable ASAP

FWIW while my hardware serving Jellyfin is more powerful, my hardware accessing is not much more, I forgot to mention that detail

I typically use a ugoos android tv box flashed with coreelec. I believe it’s more powerful than a pi for this application but it’s not particularly powerful.

I use Jellyfin for kodi over Jellycon. I’m not saying one is better than the other, I’ve never tried jellycon, but that is what my experience is based on. In my experience initial library scans are very lengthy (building db from scratch) with the size of my library, 20-30 minutes. This never is necessary at this point though and was only needed because I was testing something for coreelecs nightlies that required me to trash my db a lot. Typically I login and it updates within a few seconds, even if I’ve recently had a somewhat hefty update. It does help that the ugoos has fairly speedy emmc here - my initial testing running coreelec off of the sd card this was a bottleneck. Flashing to internal emmc and enabling hs400 mode made this notably quicker

Pinchflat is interesting! Thanks! I have been looking for a better YouTube archival tool

IMO kodi plugins are very hit/miss. This is why I prefer the setup I have where storage nas runs Jellyfin then flashed android box runs coreelec. Kodi and Jellyfin both have IPTV plugins, for example, but they are terrible. So when I want to use IPTV I boot to the degoogled android side and use tivimate, which is much more stable and convenient. I also have an APK for youtube there with adfree and sponsor block integrated. There is a build of freetube for android but unfortunately I cannot get it to work on the box. I don’t watch a ton of youtube though

I have much more powerful hardware than you but Jellyfin shouldn’t take more than a few seconds to load directories. The hardware is fairly irrelevant, this was still the case when I was on my old nas (which was an ancient pc that was garbage). Jellyfin doesn’t require much. My library is gigantic too, easily over 100,000 items across music, movies, and tv.

What do logs say? How is your network? When I moved into my new place i went ethernet only and had issues with Jellyfin (and other self hosted stuff) and tracked it down to one cable that was super cheap and limited to 100mbit.

Jellyfin isn’t really geared towards viewing media in a folder structure though.

You may be best off going with a freetube access. TBH I find Jellyfin works best with media that can be scraped or that I’m willing to create nfo files for. My music/tv/movies? These are overwhelmingly fine. Every once in a while a niche show or album requires manual scraping or a custom nfo. But I also have some other collections like music videos. Imvdb exists but is far less complete compared to other scraping sources and as a result I don’t even bother using it, the overwhelming majority of my collection needs manual nfos. When I’ve tried to contribute to it my contributions have sat in pending for literal months pending approval, even for obvious videos by major artists.

I don’t know of any scraper for youtube videos and such a thing would be a tremendous undertaking. If you archive a lot it’s a lot of nfos to create. Perhaps you could make a script that generates them automatically by scraping the description and grabbing the thumbnail as fan art?

They absolutely do this. A drug with a lack of efficacy data is a great way to get shortlisted to insurance denials

It’s one of the frustrating things bc people can then easily manipulate the issue. A drug that can be prescribed a doctor and filled by a pharmacy being denied by an insurance company is very easy to write about online. Then it’s a slam piece, “x insurance company denied me my meds”. Basically 0 people will have any interest in the nuance that the medication is bullshit or possibly even harmful. Too bad insurance companies made their bed by being absolutely horrible for decades, I guess.

Honestly the bigger thing is sponsorblock, blocking youtube ads is a big deal but it’s so much nicer to be able to circumvent the internalized advertising. Youtuber spends 4 full minutes plugging a vpn service? Automatically skipped seamlessly

Rooting ranges from very easy (literally go to a webpage, very unlikely you’re on a firmware this old), to medium (put stuff on a usb and try to play some music, may have to try a few times), to very difficult (connecting hardware to the tv), and in some cases is impossible. Don’t update your tv if you don’t have to, lg rarely fixes issues and never adds new features

https://cani.rootmy.tv/

Jellyfin for kodi is pretty bulletproof for me although it wasn’t always that way; when I first changed to this setup (bit over a year ago when Coreelec added dolby vision support) there were some headaches because kodi was changing how the db worked at the time (change from kodi 20 to 21 and Coreelec was using the kodi 21 nightlies for dv stuff that hadn’t been brought into final version)

Once the 21 final came out and Jellyfin for kodi was updated it’s been smooth sailing.

That does remind me of another downside though: there’s a minor slap fight on how db management should occur between kodi and jellyfin devs. Like many open source projects they simply can’t agree on anything so each project just does what they think is right. So kodi 22 or jellyfin 11 might cause another scenario like above if they decide to revamp the db again (especially on the Jellyfin side; I would not be surprised if someday they will overhaul the lackluster music portion of Jellyfin)

If you ever do revisit Jellyfin for metadata administration I would suggest posting about your issues on the github for Jellyfin for kodi. I’ve never used Jellycon so I can’t speak to that but Jellyfin for kodi’s github is overall helpful although fair warning that the dev can be a bit curt, especially if your issue has already been addressed and is configuration related. Things may have changed but at least ~ a year ago it was solely developed by one person who did the majority of the support as well. But if nfs is working then why break what isn’t broken?

You can root webos depending on what version you’re running but that more just lets you run homebrew (which is handy for youtube with Adblock and sponsorblock)

As others have said the best thing you can do is bypass internet connectivity altogether. I use the youtube app so I keep it connected with lg services and tracking blocked:

||snu.lge.com^ ||su.lge.com^ ||su-ssl.lge.com^ ||snu-dev.lge.com^ ||su-dev.lge.com^ ||nsu.lge.com^

(Formatted for adguard dns)

But it’s easier to just disconnect entirely. Let it collect data but if it’s disconnected it can’t do anything with it.

For a box I use a Chinese google tv box - ugoos am6b+. It can decode almost any video format (including dolby vision and all lossless audio, can pass through) except av1 basically and there are some newer versions that can do that as well. Google is awful right? The ugoos is stripped back pretty hard though it does retain the play store but still block the following:

||androidtvwatsonfe-pa.googleapis.com^ ||androidtvchannels-pa.googleapis.com^

Probably not necessary but just in case. Anyway, the Google tv side is just for streaming services (if you use them) and IPTV because Jellyfin and kodi are garbage at IPTV, tivimate on android is leagues better

Anyway flash the box to use Coreelec and copy Coreelec to the emmc, takes like 10 minutes and is pretty easy, just need an sd card. Now you use kodi as your Jellyfin app (or plex/emby but fuck plex/emby) by just installing the Jellyfin for kodi plugin in kodi and in jellyfin. Sign into your Jellyfin instance in the plugin, your library will import (can take awhile the first time if you’re like me and have a huuuuge library with like 1,000 movies, 10,000 episodes, and 300,000 songs)

Then look around on the kodi forums for a decent skin that looks nice for you bc the default one is butt, configure the menus to match your setup, adjust the skin settings to your liking, etc. backup your settings!

Now you have a Jellyfin client that plays back media directly without transcoding 99.99% of the time (assuming you have an avr that can play lossless audio), is far more mature than any of the shitty Jellyfin client apps (development started in 2003), handles stuff like subtitles far better, still syncs watched status, etc. and the worst part of kodi: library management and administration, is now handled by jellyfin, which does it much better.

If you have remote users or other TVs/phones/etc they can still use the client apps too

Downside is that you lose most plugin support. Like back before they started to roll intro skipper into Jellyfin or jelly scrub, those are useless to kodi. And a weakness of kodi is that with the rise of plex/jellyfin addon development for kodi has dropped off significantly so no introskipper plugin for kodi. Navigation features work well (pressing right on the remote skips forward 10s) so that’s close enough, or just download good quality rips with chapters

I remember when compression was popularized, like mp3 and jpg, people would run experiments where they would convert lossy to lossy to lossy to lossy over and over and then share the final image, which was this overcooked nightmare

I wonder if a similar dynamic applies to the scenario presented in the comic with AI summarization and expansion of topics. Start with a few bullet points have it expand that to a paragraph or so, have it summarize it back down to bullet points, repeat 4-5 times, then see how far off you get from the original point.