TikTok[’s] parent launched a [web] scraper [that’s] gobbling up [the] world’s [online] data 25x[-times] faster than OpenAI
265 karma · joined January 29, 2018
TikTok[’s] parent launched a [web] scraper [that’s] gobbling up [the] world’s [online] data 25x[-times] faster than OpenAI
If someone's violating internal auditing procedures, those same procedures won't catch them. It's dangerous because it's a violation of the procedure itself. Proving such violations without tools like no-knock warrants or the NSA moving in is nearly impossible.
So you end up with a misappropriated circus of Big Compliance issuing fines over no wrongdoing and internal audits finding no wrongdoing when you rarely hear about this type of internal abuse unless someone is careless enough to brag about it to their Tinder date.
Anyways I think it's fine for them to define breach as the loss/destruction of data i.e. making a password known, which destroys it's value.
Here's the real debate: SimCity 4 vs Cities: Skylines. SC4 uses a population-based statistical simulation. Cities: Skylines, stemming from Paradox's Cities In Motion (built with Unity engine [1]) focuses on traffic modeling using an agent-based simulation. CS excels at traffic flows because it’s designed for that purpose.
It's the micro-level accuracy of Cities: Skylines "agents" and SC4’s macro-level realism of its "Sims."
Instead of creating individual agents for traffic simulation, SC4 runs a pathfinding heuristic to find the fastest route between Sims' homes and jobs. Maxis designed this simulation to run on a 2004 Pentium III 500MHz with 128 MB RAM. Surprisingly, it still holds up today.
In CS, agents return to designated homes. In SC4, Sims returns to the nearest vacant home after work. SC4's simplified simulation reveals the optimal path for Sims. This difference might seem immersion-breaking, but perfect pathfinding uncovers the most efficient urban designs that might initially be hidden. Isn't that the whole point of the simulation?
SC4 isn’t perfect. Casual players ask: Why no diagonal roads? Why are my buses unused? Why aren't Sims going to work? The Network Addon Mod (NAM) [2], around 2004, has addressed some of these issues. It’s not perfect, but it makes the simulation smarter, so Sims can make decisions closer to agent-based solutions. NAM is the band aid covering the small cut that Cities Skylines makes in its otherwise superior simulation.
SimCity 4 still holds its own as a city simulator.
[1] https://web.archive.org/web/20220716021755/https://unity.com...
[2] https://www.sc4nam.com/docs/feature-guides/the-nam-traffic-s...
Second: I don't get what's wrong with letting person A to post in the wrong/old place if they want, and person B can link the URL of the post to the right place. Instead you have person A locking person B's post which might actually hurt their feelings, like being arbitrarily moderated on any other site does, so they never come back.
Third: When Stack overflow moderation approves something, it actually means it's the end of discussion. IIRC the famous SO post about parsing HTML with regex ended with moderation endorsing a meme (the wrong answer) and locking it.
But I'm giving too much attention to SO anyways they could fix everything and I still don't see how it would win back talent.
I appreciate your perspective. But let's be real here: I just drag the DXVK d3d9.dll for Linux into my Windows game folder. Suddenly, my DirectX 9 games are in the Vulkan future. Exclusive fullscreen taking 30 seconds to alt-tab? No more. Scaling issues? It's better. At least it's an alternative. Truly an exciting time to be alive.
Linux isn't all roses. It's a shitshow of its own:
- Bluetooth drivers (thanks, Microsoft)
- Graphics (Wayland vs. eglstream, thanks Nvidia)
- WiFi drivers (clusterfuck)
- The never-ending "free as in free beer I stole from my employer who doesn't exist anymore" featuring the ZFS red herring or trojan horse depending on whose side you're on.
- epoll < kqueue < iocp, but I hear io_uring has caught up decades later
- Wayland in 2030: Finally running GLX gears over SSH!
I try to be kind to BSD. If you're going into that space, you should already expect nothing to work. Then, it will surpass expectations if anything runs. Wayland should just hang with the BSD crowd because then you'd get kicked in the balls (joking) or told to stay in school for even asking a question like, "why doesn't my Nvidia card drivers work with this". In 2030, Netflix might contribute network optimization code back to BSD. Nvidia at best might throw out a few steaks into the crowd by then (same timeline). It's hard to say whether they'll throw them closer to Linux or BSD because the target has been changing.
2030: Year of the Linux Desktop™. Also the year support begins to expire on my edition of Windows 10 and I'll have to re-evaluate the time I spend inside vs the time it will take to unfuck Windows 12. I probably won't need to upgrade my router hardware before then either, but it oscillates between a BSD and Linux install every year. I choose the right tool for the job, you could say. I've been planning to dig into iPXE and friends to make reformatting a breeze because I don't like absolute choices when there are advantages on both sides.
If you want tragic, have the neckbeards/graybeards gone extinct? I wonder if a lot of them gave up and ended up working at Microsoft.
If you want hell, go check out the world of printer drivers. You'll end up back in WordPerfect for DOS, which lets you import the fonts and print them to Postscript. How hard can it be to print a social media webpage to a piece of paper and have it look good? Between needing Chrome to render JavaScript to a canvas and then trying to get that information into a structured PDF that looks good when printed is hell. Generally, when dealing with PDFs, everyone works backward by trying to unfuck the PDF that was dumped on them by converting it to another PDF and sending it on its way because nobody likes to be in hell for longer than mandatory. But for me, who knows what fonts Chrome shoved into that PDF? How do I select a font? Well, my Chrome browser is still fucked from trying to switch the fonts in the one place you can choose. But that still left out the header (page title, page number) and footer fonts (URL). They require a font too. The closest I could get to choosing fonts is using Puppeteer providing custom CSS to import the font file for the header and footer that's passed to PDFium somehow. I would not trust my life that the CSS imported the font. Even then, the files were an unstructured nightmare, Chrome would override the fonts anyway for several Kanji dialects and their own ugly-ass emojis, and Adobe was still mangling my fonts during linking. How hard can it be to use ABC font with XYZ emoji style? I don't want Google's ugly-ass emojis.
Side note: Acrobat shares a trait with stock exchange trading applications: it charges enterprise rates if you don't use the GUI.
Anyways, after using port monitors to spit the printer job out to a file, I could not get Chrome to send the PDF directly to my own program without using a for-loop to monitor when the file is closed, which, at which point you must assume the file is done printing. Not happy with that loose of a guarantee, I ended up back trying to intercept the fucking SPOOL file in SYSTEM32 to generate a PDF. Or pretending that my Python script is a network printer. Because I was going that far without success, I started reading the documentation to develop a printer driver, which had gotten more hostile since the "PrintNightmare" vulnerability. I relate to this news article because Windows wants me to be a Windows developer by adding security in the area I need to access. Being Printer Driver for Windows Developer seemed like a lot of work, so at that point, I wondered if Chrome's PDFium was giving me garbage in the first place. I tried compiling Chrome to get to the root of PDFium. However, apparently, compiling Chrome requires a fucking mainframe war computer, and my feeble destop computer stood no chance and BSODed. Also if you suggest Ghostscript (Ubuntu for PDFs), I will scream because I need Adobe's demon branding and the fonts I want on my documents. I ended up printing Chromium's PDF clusterfuck to Adobe's PDF clusterfuck using Acrobat to get it to import the right fonts, excluding the emojis which could either go black-and-white or Google's Ugly-ass Color Emoji. That didn't solve the structuring issue either, and I had to do the print process manually opening up Acrobat and hitting control-P enter every time, or deal with undocumented command line features. I concluded that nobody uses physical printers anymore. Until now. 2030 is the year of the DOS printer.
On the upside, .NET Core compiling to a Windows or Linux executable web server written in C# is neat. I'm curious if many Linux guys ended up at Microsoft to survive during the Greybeard extinction. How much pull does Microsoft have with the Linux board? It's hard to truly escape.
But hey, at least we can agree on one thing: Electron is awful and makes me think about how much battery % will be drained per line of text I write. The end result: the Windows Phone. I have to exit this thread quickly now.
This thread is about the gamepad. The lawsuit mentions it. It's relevant. Period.
Showing Rush tossing the controller while saying "it's bluetooth so I can hand it to anybody... it's meant for a 16 year old to throw it around and super durable" is powerful imagery for a jury.
The complaint specifically mentions Bluetooth, which I can't tell if it's an error. They switched from the DualShock controller used in Cyclops I, which does use Bluetooth. There is probably a reason they switched wireless technologies (Bluetooth is even worse than Logitech's 2.4 GHz wireless), which if not documented, could also show negligence.
Your carbon fiber hull fatigue theory is interesting.
I picture it sitting out in the hot sun and cold nights for months. It reminds me of the CNA building in Chicago, where windows fractured due to uneven sun exposure. The owners didn't replace them until one fell and killed a pedestrian, resulting in an $18 million settlement in 1975.
In the prior dive, a passenger hit the sub into a rock. How does that not contribute?
Microsoft: Oh, yeah, to use it, you'll need to deploy it with the Microsoft Store.
Edit: To top this off, here are some references and further reading.
[1] Android apps jamming up 911 functions, "Android phones have a bug that stops them calling the police, Google warns (2021)" https://www.independent.co.uk/tech/google-android-microsoft-...
[2] Blackberry lasts 3 days per charge of normal usage, "BlackBerry Bold 9780: Not So Bold (2011)" https://www.anandtech.com/show/4228/blackberry-bold-9780-not...
[3] The surgical room to replace batteries, "Apple shipped me a 79-pound iPhone repair kit to fix a 1.1-ounce battery (2022)" https://www.theverge.com/2022/5/21/23079058/apple-self-servi...
[4] The government suing itself to stop Blackberry's injunction because they all used Blackberry phones, "Fed Agencies Side With BlackBerry (2006)" http://news.bbc.co.uk/2/hi/business/4672844.stm
[5] Related: Today's HN discussion on bugs in QNX. Comments mention Blackberry's acquisition and why they use Android now, "The weirdest QNX bug I've encountered (2021)" https://news.ycombinator.com/item?id=40837411
The multiple suits issue is generally downplayed because it doesn't promote Speedo's new technology and casts swimmers as cheaters. Imagine showing up to a race with your new carbon fiber running shoes, only to see your opponent wearing "Moon Shoes" strapped over their high-tech shoes. It looks ridiculous, but there was no rule against it at the time.
Perhaps it's making the same mistake as SSNs in that people use public or easily generated information to verify identity in the first place.
> "Personal information" means an individual's first name or first initial and last name in combination with any one or more of the following data elements...
> 2. Driver's license number or California Identification Card number.
https://en.wikipedia.org/wiki/California_Senate_Bill_1386_(2...
What's funny is that I can't remember when I scratched a CD enough for it to have any noticeable effect. Even the ones my neanderthal friends handled and returned covered in fine scratches worked fine. But it might be just because I kept everything in the original holders and made so many copies of everything; it's possible I just never noticed. I've had the actual disc readers burn out, though (looking at you, Fat PS2).
Man, I loved burning CDs. It felt like magic every time. My whole family was into it. I'd sit there the family's stack of slim jewelcases and spindle of Maxwell CD-RWs. Listening to the satisfying BRRRR and minutes later, out pops your warm copy fresh from the oven. I don't miss the random compatibility and disc protection errors. But when it worked, it was so sick.
Anyway, I don't buy Github's excuse that the new search 'takes more CPU power'; this must be to prevent scraping data for LLMs. Have you hit the new search rate limits?