If the box where I can’t set up a third party player to do the ‘replay last 5 seconds with subtitles on’ because it’s all locked down is the best then I don’t want to know what the worst is, I’ll just keep using LibreElec. At least if LibreElec does something I don’t like Claude can fix it.
The Apple TV (hardware) can do what you're asking using a voice command "What did he/she say". It's possible it no longer works in every app because services insist on writing their own players that don't work as well as the player provided by Apple TV.
I use LibreElec on a Raspberry Pi 4, though it doubtless also works well on a Pi 5 (and the cheapest one with 1 GB of RAM is probably plenty).
It does great with local [networked] media and that's all I use it for. There's a million plugins for other things, some dating back to the XBMC days on the OG Xbox 20+ years ago, that others may find utility in... but for regular commercialized streaming I just use a $25 Chromecast and that does the business well-enough.
(LibreElec has a huge advantage to me that most people probably don't care about: My AV receiver sounds brilliant and was once very expensive [~$4,500], but it's rather old and it chokes on HDMI signals higher than 1080i. It is also unable to decode things like Dolby Digital Plus, TrueHD, and DTS-HD.
This would make the receiver a non-starter for post-Blu-Ray film soundtracks, except the Pi4 has two HDMI outputs. It is a no-effort built-in for LibreElec to use one of these outputs for video, while the other one is dedicated to high-res multichannel PCM audio.
In this way: The video system gets whatever video signals it wants, while the audio system is just fed PCM audio and doesn't have to struggle with any of that more-modern business at all. Everything is very happy with this arrangement.)
A cheap Android TV box. Between lack of proper hardware acceleration for 4K/Dolby Vision decoding and lack of sleep(!) support, I would never recommend that hardware, the ones that are actually recommended by the community are RP5 ($$$) and especially Intel Nx00 (N100 etc.) which are actually the gold standard of mainline(-ish in case of RPi) codec and hardware support.
I can empathize with the commenter you're replying to. I still read for pleasure and have to read constantly for work of course. On some days, I feel like I can't. It is like exercising the same muscle every day which will lead to injury in a short amount of time. Switching to audiobooks or just taking a break does resolve that.
From what I am seeing around me though, reading for pleasure or even to gain knowledge has decreased a lot. I noticed it in adults first and this is being reflected in their children.
Except you don't get injured. You know what happens when I read a book at the end of the day after working all day? I fall asleep about 20 pages in, and sleep like a rock. You know what happens when I read HN at the end of the day after working all day? I'm up till 3am and getting 5 hours of rest. Same for any screen time really. People think screentime is some big rest but it isn't really.
Just rip the bandaid off. Get over the inertia of doing the task. You won't injure yourself reading a damn book.
> The Marquis de Lafayette and James Manor were just 18 when the revolutionary war started.
And Thomas “all men are created equal” Jefferson raped his slaves[0], and then enslaved the resulting children, but we don’t do that nowadays either, because I guess some things changed since 200+years ago?
I assume they’re encrypted/DRM’ed when deployed on inference hardware, so only core researchers/sec admins would potentially have some access to unprotected weights, and they are far too well paid to risk it leaking the model
Incentives matter on the average, but people are too unpredictable for categorical statements like that. They can always have other reasons beyond personal gain to leak secrets.
There was no shortage of spies and defectors leaking American nuclear secrets to the USSR during the Cold War.
Newer NVidia cards (H100 and up) support both in-memory model encryption and ‘trusted’ execution environment/remote attestation, not sure how widely used in frontier model deployments, but at least vendor claimed perf overhead is ‘3%’ [0]
There are also lots of cheaper SoMs if you're not allergic to Chinese chipsets, and the cheaper ones tend to have PoP on-package DDR so you can spin your own 4-layer PCB without having to pull your hair out impedance matching DDR3+ traces. That, of course, if you can fit into ~64MB RAM.
> The kernel contributions are also a bit on the shakier side which requires spending expensive developer time to deal with kernel
NXP/i.MX are way better at mainline kernel than Broadcom that RPi is based on, come on.. and they have cheaper options like i.MX 9 series. Other vendors, yes, mainline support could be pretty spotty.
images/screenshots could have watermarks, recreations don't. Sources = not the people who created memes, sources here just mean people who sent the screenshots to the journalist.
The other side of it is forced obsolescence where new OS makes your existing hardware slower. So I wouldn't upgrade my phone beyond iOS18.x purely for performance reasons but if there's a killer feature in a new iPhone I would still consider buying it because its hardware was built to handle the new effects and extra ram it needs.
reply