The original research was regarding supercapacitors, not Lithium-ion. Based on the PNAS preview it seems it has to do with changing the design of a supercapacitor to increase charge rate.
Increasing capacitance (how much charge is stored to reach a certain voltage) or the voltage it is charged to would indeed increase the capacity. Putting several in parallel would work, as would making a bigger capacitor. The main problem as far as I can tell is that the energy density of even supercapacitors is low, so you’d need a much larger volume to have the same capacity (and thus a much thicker phone).
Alexandria and Stract use their own open source crawlers. Brave is also independent, if I recall correctly.
DDG’s AI chat isn’t a search engine. It’s just a chat interface for GPT-3.5 Turbo and Claude, without any search access. It’s also not very up to date.
GPT-3.5 Turbo:
Who is the PM of the UK?
As of my last update, the Prime Minister of the United Kingdom is Boris Johnson. Please note that political positions can change, so I recommend checking the latest news sources for the most up-to-date information.
Is Queen Elizabeth still alive?
Yes, as of my last update, Queen Elizabeth II is still alive.
The Claude version doesn’t fare any better:
Is Queen Elizabeth still alive?
Yes, Queen Elizabeth II is still alive. She is the current and longest-reigning monarch of the United Kingdom, having ascended to the throne in 1952. As of 2023, she is 96 years old.
Are there any search engines other than perplexity that tie the sources to sections rather than just popping them all at the bottom? That always felt like the most practical layout for being able to easily cross-check information against their supposed sources.
I guess it kind of depends. Not really sure what most people actually use, but for those who use MS’s services, Office web isn’t great, and Skype for Linux is rather temperamental. A lot of games work under Proton, but not all.
My perception of “average user” is probably skewed towards being not technical enough to troubleshoot on their own, but skilled enough to run through a tutorial of what keys to press. For someone used to Windows, patching things up is simpler than learning all the ins and outs of a new OS.
I don’t disagree that most people would be fine using Linux, but there needs to be a compelling reason why Linux would be significantly better, or else the switching cost makes it not worthwhile.
When going from Windows to Linux, all of the tradeoffs are involved. For me what I don’t like about Windows outweighs the pain points of my choice of Linux distro, but for some they’d weigh the sides and Windows still comes out on top.
Anyway my take is that Linux is better ideologically, but for the average consumer who justs want to use their favorite apps, Windows works fine and they’re not really going to care until Windows piles on enough garbage to make switching worthwhile.
(Not the person you replied to)
Windows has issues, but so does Linux. My personal experience with Fedora (Silverblue) has been fairly good with minimal hassle (Gnome Software breaks sometimes with auto updates, but is leaps and bounds ahead of the Synaptic days). However, someone using other hardware, another distro, or using other software might have a lot more problems to contend with.
There’s a lot of case-by-case nuance that in my opinion makes broad switch from A to B recommendations less meaningful than discussing the pros and cons and letting people decide on their own whether Linux could be useful for them.
The main downside is that there is a lot less customization of filters short of using a different DNS. There is also the potential for logging DNS (present with normal DNS servers as well). LibreOps claims they don’t log requests, and personally I don’t think they have much reason to lie, but there is still that element of trust. Many of the more well known DNS servers don’t offer ad blocking DNS, so you’ll most likely be switching to a different provider.
Another way is through DNS (eg. noads.libredns.gr).
Some useful services:
There’s a similar “feature” in Android that replaced the old Android ad ID. Android rolling out new Ad privacy settings
Good data (and program) structures are definitely quite important. Well chosen structures make implementation much easier (and likewise bad structure makes things needlessly difficult).
Also, the film editing example is also an example of a piece table, which makes cutting very simple. Cutting out a section is just a node insertion + update the end of the original node ({0-3} -> {0-1}, {2-3}).
That would depend largely on the use-case and specific software. I’m fairly confident that Lyx isn’t going to become bloated any time soon, but I can see that happening especially with proprietary alternatives like Word (ignoring for a moment Word isn’t on Linux). It all really depends on whether or not a less bloated alternative exists.
It depends on the use case, but for what it’s worth on a 4GB Android tablet, I can run VSCode + Chromium/Firefox via Termux without too much trouble. ~2GB of memory is taken by Android, so 8GB on a proper Linux system is more like 3x more memory available. It would take a massive amount of bloat to make an impact. My main concern would like with websites being wasteful with both memory and CPU usage via JS, rather than the browser itself becoming bloated.
The study is from 2018, and I wasn’t able to locate the original source from searching. Also, from the author’s bio:
Ph.D. Rocket Surgeon & Aspiring Troglodyte
The Hacker News discussion also does not inspire confidence…
By that I meant from the perspective that the initial allegations still felt like it could all just be a misunderstanding. Now that it has been donated, it seems to be more a matter of who at Open Hand was actually in the know (since it is possible that Jirard geniunely was being misled himself), and why the money wasn’t being donated. The golf tournament stuff definitely feels much more circumstantial since it is based on extrapolation. Overall it does seem like the IRS getting involved is going be the only way definitive evidence of what was actually going on will come out.
VMs have their own drawbacks. There are some projects to integrate a Windows VM with Linux (WinApps), but it won’t quite integrate fully. Graphical performance is bad without a GPU to pass through (Intel GVT-g kind of works, but is a massive pain to get working).
As someone who hopped over to the Linux side of the fence… same. Dual-booting somewhat eased the transition though, since I could do it more gradually and fall back to Windows whenever I needed it. Now that I primarily use Linux, I love how swapping to a new computer is 99% done by just copying homefolders. Even apps copy over, using user installed Flatpaks.
In the EEA, much more is on the way:
Bing’s web search from the Start menu and the Edge browser can be uninstalled Third parties can add to the Windows Widgets Board feeds Third parties, like Google or DuckDuckGo, can provide the built-in web search results that Bing once had exclusively Windows users who choose to sync their Microsoft accounts will have their pinned apps and preferences synced, seemingly keeping their EEA-enabled choices Windows will now “always use customers’ configured app default settings for link and file types”
Good to see Microsoft just blatantly confirming that these are anti-competitive measures rather than any sort of technical limitation.
The number of bytes per image doesn’t necessarily mean there’s no copying of the original data. There are examples of some images being “compressed” (lossily) by Stable Diffusion; in that case the images were specifically sought out, but I think it does show that overfitting is an issue, even if the model is small enough to ensure it doesn’t overfit for every image.
Strings work fine, the problem is the (single) quotes:
~ $ foo="echo 'hello world'"
~ $ for x in $foo; do echo $x; done
echo
'hello
world'
~ $ $foo
'hello world'
~ $ eval "$foo"
hello world
The splitting is by whitespace, so the single quotes remain in the arguments. Using eval (and double quotes to preven splitting), it gets processed correctly. That said, don’t use eval; use functions or aliases instead.
Until Microsoft takes that option away as well…