I'm making the first effort in many years to finally be rid of windows,because of recall. I'm really uncomfortable with the idea that once that is implemented I really will have no un-monitored/spied source of computer usage. It used to be almost impossible to reach some sort of usability parity linux/windows but I think its possible now. A big part of that is with computer power long surpassing daily usage needs the lack of cutting edge driver support on linux is no longer as big a deal. I started with ubuntu, but I also am pairing it with NIX package manager. I'd like to use this experiment to get familiar with the disposable OS paradigm. I think that could be a killer feature to move away from windows. I don't see them ever getting windows to be rebuildable on demand.
I realize NixOS has some hurdles and learning curve though.
<rant>
Been using Manjaro XFCE for a minute, KDE is great if you're rocking ~32GB of RAM but for lightweight, MX Linux (Debian), Lubuntu (LXDE) Lightweight Ubuntu. Most distros come with drivers that trackpads and wireless just ..work. So many flavors but with KDE you can customize everything with rightclicking, no config edits. It's fun, but lower end machines I'd suggest XFCE, KDE will use 2GB of RAM just idling on desktop.
I ditched Windows completely back in 2017-2018 and haven't looked back. I'm no star-programmer or star sys-admin or anything but the love of tinkering with things, VMs, networking, sockets, SSH, SCP, rsync, having direct access to my kernel (like sysctl stuff), managing my ssh keys in ~/.ssh/, using a drop-down Quake style terminal (Tilda) has been a dream.
Why would somebody want to sign on with a 'cloud account' for their desktop login, have forced stuff like Teams, Edge (Are you SURE YOU WANT TO SWITCH TO CHROME/FIREFOX? notifications), OneDrive, and have things like Recall pushed on them.
Linux isn't rocket science any more there's Add/Remove programs on every distro and people can browse from there, there's software pulled straight from github for compiling/installing as well as the 'app store' for exploring.
It's been my tinkerer's heaven. Most you need is a trustable VPN with a killswitch that runs Wireguard and your linux install is going to be great.
Gaming wise, well gaming needs to get their heads out of their asses and dev for Linux and respect that just because there's not yet manjority market share doesn't mean there won't be.
Disposable OS now adays is just boot a live CD (USB) that loads everything in to RAM with NVidia drivers already loaded.
My suggestion to anybody, check out Ventoy: https://www.ventoy.net/en/index.html
Try a bunch of linux distros .isos in oracle VMs or any VM managers, use Ventoy to put the ISOs on a flash drive, my 128GB stick lets me have dozens of them, and boot any distro you want you can have as many ISOs that will fit, they'll all boot when you select which one you want.
For the nerds: https://www.ghostbsd.org - BSD Desktop.
Say no to Recall.
</rant>
> KDE will use 2GB of RAM just idling on desktop
As much as I dislike KDE, this is just not true. Perhaps you mean your whole system idles at 2GB including all other running services?
To anyone new to Linux, I recommend sticking to Wayland. Gnome if you're familiar with MacOS, KDE if coming from Windows. Use a popular distro with good documentation online (Ubuntu, Fedora, Arch).
Don't worry too much about memory. Just about anything will be lighter than windows. I've had friends choose the absolute worst X11 desktops because it's "light" then run into all sorts of multi monitor issues
I'm talking utilizing it with the desktop widgets, tools, customizations cool taskbar widgets (I'd add the openAI app), different system informations/rss feeds embedded across multiple monitors.. the stuff KDE is awesome for.
KDE is amazing. I love it. My little laptop though I get by with just XFCE on an i3 8GB RAM, 250GB NVMe SSD.
Agreed on everything else you said. People switching from Mac's ultra-HD for fonts get the 'ick' switching to Linux desktop too. We'll get there.
What you are saying about KDE is misinformation. Plasmashell in KDE 6.2 uses between 300 and 500MB RAM depending on the age of the session.
I replied to an above comment, not when utilizing KDE's widgets and other desktop integrations like RSS feeds, weather, embedded system/server informations, etc. KDE is amazing, but I like to utilize the widgets, and full functionality of things with it. With my i3, 8GB RAM I opted for Manjaro XFCE.
I absolutely suggest using KDE for people switching over. Full desktop experience.
When not using anything my systems would idle at around 800MB-1GB running default install Manjaro. Firefox after awhile and I was using the full 8GB and tabs were crashing.
[dead]
> Earlier this week, Microsoft again clarified that Recall will not be mandatory on Copilot Plus PCs, and will be an opt-in experience that can be fully removed. The clarification came after various YouTube videos claimed Recall was being installed on any PC with Windows 11, version 24H2.
Oh, thank god. I definitely heard the "You can't opt out!" line and was worried.
Until Microsoft gives users a legally binding "And we'll never quietly turn it on with a Windows Update" promise I don't trust them, given the way disabled features repeatedly get reenabled by updates in Windows 10 and 11 already
Hope they warn us of the incoming spyware, so we can uninstall on the very next second, right before it grabs all it can and runs away with it.
Seriously. Not sure if the marketer who's preparing the "Get started with Recall!" graphic that's gonna pop up on our computers realizes that they're doing us a different kind of service, lol.
What exactly keeps people on Windows if they feel like theyre in such a relationship with their operating system?
In my household it's MS Office. My partner very much enjoys the UX of KDE, but can't switch because she's highly reliant on features in MS Office that I cannot find transliterated well enough to work on Linux. Otherwise, there is no place for Windows in my house.
Personally? Inertia. I hate switching tools/systems unless absolutely necessary. If I were a frog you could absolutely boil me.
For me? Gaming and lack of alternatives. I hate MacOS' hostility toward anything not Jobsian, and Linux doesn't have parity with Windows particularly around gaming.
I prefer Linux, but I dual-boot Windows for gaming, since I haven't been able to get my Nvidia card to work at its full capacity on Linux.
Yeah that's the best news so far. I would have used hacks to disable it anyway but the problem is that at work I can't do that.
But I'm in the IT department at my work and our privacy teams are also not even considering to implement this as it's a legal minefield in Europe. So big brother won't come to my work yet, at least for now.
Question: why are people having a tizzy about this specifically yet touting Apple Intelligence?
- people are going to be using intelligence with Siri. Siri sends all transcripts to Apple, even if processed locally. Will this include text transcripts too? Thatâs some seriously private information that gets sent to Apple that you canât opt out of.
- have you looked at whatâs in knowledge.db?
People hissy with Microsoftâs product because of OCR of the screen or is that only a factor?
A key and screen logger is what I consider malware. At least with Siri it can be disabled during the setup Wizard and you can choose to use it or not (i never do on Mac)
And Apple intelligence is mostly local on device. I don't think I'll use it either as I have my own AI server already and I only use a Mac for work. But Apple's implementation is privacy-first, where Microsoft's is "get this to market first, worry about everything else later"
> At least with Siri it can be disabled during the setup Wizard and you can choose to use it or not (i never do on Mac)
How is that different than disabling Recall on first boot/setup?
Microsoft is notorious for resetting user permissions randomly. Like after updates
Options turn themselves on for some reason, historically speaking
Knowing the Microsoft of this decade, eventually you will only have the options [Yes, spy me] and [Ask me later].
> How is that different than disabling Recall on first boot/setup?
Microsoft is known for being utterly user-hostile when pushing features on people. I have a gaming PC I occasionally boot up and every single time I do so it harasses me to use their cloud service in a way that very much seems like it will restrict my ability to use my own computer if I don't.
User trust is expensive to buy and cheap to relinquish!
Microsoft has shown time and time again that they will gleefully reactivate things you have disabled and they will use dark patterns to hide their intent.
> people are going to be using intelligence with Siri. Siri sends all transcripts to Apple, even if processed locally. Will this include text transcripts too? Thatâs some seriously private information that gets sent to Apple that you canât opt out of.
I don't use Siri. I have never been interested in talking to my computer. So that's one way to opt out.
On the other hand I kind of need a screen to use my computer.
I donât use Siri because of the transcripts myself. I want to, but I donât.
I think youâve skipped over the knowledgeC.db part. Know what that is and how it works?
I didn't skip over it, I chose not to reply.
My understanding of it is it collects metadata. It doesn't collect screenshots or do OCR, again as far as I know. It'd a completely different system from Recall.
Their security track record does not really help. I will never trust them again after they refused to tell their customer about a workaround because they wanted that juicy DoD contract (https://www.propublica.org/article/microsoft-solarwinds-gold...).
Why would I trust that company with my data?
I have a slight suspicion that the overlap between people upset about Recall and people excited for Apple Intelligence is very small. As for why the gap, let's be honest. Nobody has more loyalty to a giant corporation than many people do to Apple. You may speculate exactly why that is, but the reputation doesn't lie.
From this thread:
> I definitely heard the "You can't opt out!" line and was worried.
> Microsoft is known for being utterly user-hostile when pushing features on people.
In truth ( so far ) .. you can opt out (not easily) and here Microsoft is guilty of poor engineering rather than active hostility (although hard to seperate).
Chris Titus, the source of a two week(?) old HN thread about Recall being "Mandatory" and unremovable from latest Windows w/out breaking file explorer released an eight minute update video on the matter:
https://news.ycombinator.com/item?id=42002418
For those not watching, the problem was a bad Microsoft dependancy manifest - a common DLL not marked as a file explorer dependancy got removed along with Recall causing file explorer to glitch.
This is now 'solved' and included in his cleanup | debloat scripts for those that want the feature gone.
He closes with some other thoughts on Microsoft feature software deployment going forwards into the future.
'Recall' for now is little more than a placeholder stub for most, incapable of activation or more for many w/out specific hardware, and intended to be "removable" provided the engineers get their manifest dependancies sorted out.
> In truth ( so far ) .. you can opt out (not easily) and here Microsoft is guilty of poor engineering rather than active hostility (although hard to seperate).
I find that a bit hard to believe given how aggressive Microsoft is putting copilot in the market. Copilot buttons on keyboards, in edge, in the windows toolbar, all without being asked for it.
I think it's more of a combo of both elements. I'm sure the marketing people don't like it being too easily removable. And the "not easy" opt out process is something I already call user hostile for such an invasive feature.
Sure it's only on copilot+ hardware but soon that'll be the only thing one can buy. And there's more reasons to want local AI without having a big brother spyware integrated.
Chris Titus is skeptical and has a community of windows hackers and former (and likely current) Microsoft engineers commenting on his work and offering tips.
_So far_ to date it's plausible to see what's happened with Recall being hard to "safely" remove as casual incompetence rather than deliberate malicious behaviour.
He expands on this in the latter part of the video linked above - but a very good point is that moving into the future, whether intended or not, care will likely not be taken to ensure ease of removal and seamless operation after scapel applied simply because the Windows stack of cards is getting big, unwieldy, and harder and herder for single engineers to grasp all of.
> And the "not easy" opt out process
In this specific instance it looks more like human error on a dependancy tree .. generally these are correct in Microsoft packages, screwups can happen.
I mentioned this in another post recently https://news.ycombinator.com/item?id=41699265
but I've been using something of a homegrown equivalent of Recall most of the year and I really do think the concept has a lot of potential. definitely understand the privacy concerns (and especially if there's no opt-out), but I just want to encourage y'all to be open minded and consider the potential benefits.
anybody use google photos? I've got almost 20 years of photos and videos on it. But there's so many gaps in the timeline! so why not just record everything. and with an AI layer on top to help surface and search interesting/relevant moments. same thing for facebook and twitter and ig and hn and whatever social media you might use. it's such a privilege to be able to go back and look at our digital artifacts from years and decades past. so why not do it in 4k+audio.
I take screenshots and use pHash to see if I should save them. Then I OCR them etc. It's basically a homegrown little recall. Then a timestamp of when the screenshot was taken goes into a db. I also have a keylogger and some more personal data collectors. Still working it out into a "usable platform" though.
Do you have any sharable info as to how you setup your own private version of recall?
I talk about it a little bit here (high level not technical) https://youtu.be/2zqXkNhaJx0
but it's basically private daily youtube livestreams whenever I'm at my PC (so like 8+ hours a day) + limitless.ai for real time audio transcription (and whisper for local/offline transcription) + every 30 minutes throughout the day I do a quick status update+reflection in my obsidian notes, or ad-hoc annotation of notable moments in the day. WIP is using local multimodal models to add additional more fine-grained annotation from my livestream videos.
Steve Jobs once said something about working back from the desired user experience to the technology. What Microsoft is doing here is the complete opposite. Theyâre making the technology fit the user experience.
near as i can tell, the "ai revolution" consists largely of nontechnical climbers trying to get a promotion package (or scamming VCs, just like the XML craze of 1999)
it's very much along the lines of "i've got a great idea for a video game, all you have to do is implement it". that and two bucks will get you a cup of coffee
also, get off my lawn :-)
In other words, they are struggling to implement it. Software development in an average company is on some sort of a weird asymptotic curve towards getting anything done becoming literally impossible. And Microsoft is so, so painfully average.
I've been waiting for Windows to get the vertical taskbar back, but I'm beginning to think that the difficulty-to-develop is increasing faster than the development itself is progressing.
Get a daily email with the the top stories from Hacker News. No spam, unsubscribe at any time.