robots.txt is 100% honor based. Well known bots like Googlebot, Bingbot, etc. definitely honor them. But there are also plenty of bots that completely ignore them.
I would hope the bots used to collect LLM training data honors them, but there’s no way to know for certain. And all it really takes is one bot ignoring it for the content of your website to end up in a random set of training data…
Try using “curl -A” to specify a User-Agent string that matches Chrome or Firefox.
deleted by creator
I hope you realize that virtually every CDN provider does the exact same thing in similar ways. Sites that use Akamai, AWS, Google cloud, Fastly, etc. all give those companies access to unencrypted content. It’s just how CDNs work…
Getting rear ended on the highway by a drunk driver. Had all of about 1 second of warning when she slammed on her brakes and started skidding.
My favorite when debugging some code for a memory manager, written in the days of DOS extended memory, was shit_cookie_corrupt.
The original author called blocks of memory “cookies”. If too many cookies were corrupted then eventually the function ohShitOhShitOhShit was called, which shut everything down.
Saw Queen perform with Adam Lambert a few months ago. They played one or two pieces written by Brian May that really tied those two professions of his together. It blows my mind that he’s worked with NASA and the like quite a bit over the years.
You died every week (from diabetes)?
I got royally screwed by the federal government personnel data breach that happened a number of years ago. As a result I have free identity theft monitoring with a really good company for the foreseeable future.
A well thought out and implemented backup system, along with a good security setup is how you deal with malware. If backups won’t protect you from malware then you’re doing backups wrong. A proper backup implementation keeps a series of full backups plus incremental backups based on those full ones. So say your data doesn’t change very often, then you might do a full backup once a month and incremental ones twice a week. You keep 6 months of the combinations of full & incrementals, you don’t just overwrite the backups with new ones.
If you’re doing backups like that and you suffer a malware attack then you have the ability to recover data as far as 6 months ago. The chances you don’t discover malware encrypting your data for 6+ months is tiny. If you’re really paranoid then you also test recovering files from random backups on a regular basis.
My employer has detected and blocked multiple malware attacks using a combination of the above practices plus device management software that can detect unusual NAS activity and block suspect devices on our networks. Each time our security team was able to identify the encrypted files and restore over 99% from backups.
Suppose you’re hit by a ransomware attack and all the data on your NAS gets encrypted. Your RAID “backup” is just as inaccessible as everything else. So it’s not a backup. A true backup would let you recover from the ransomware attack once you have identified and removed the malware that allowed the attack.
As far as BitTorrent itself goes, your optimal speed is also going to depend a bit on your client and the number of peers in the swarm.
Suppose you’re seeding a file to 3 peers. It’s not very efficient if your client uploads part 1 of your file to each peer, then uploads part 2 to each peer, etc. A more optimized upload would upload part 1 to peer A, part 2 to peer B, part 3 to peer C, etc. Then the peers can share each of those parts with each other. This way you are effectively only uploading the file one time before other nodes start seeding as well.
The thing is, this sort of seeding only works well in specific situations, including when there’s only one seeder, etc. And not all clients support this. Take a look at qbittorrent’s super seeding option for an example of one client that does.
It was pure c code that was used to print reports, and included the date in a header. Whoever wrote it miscalculated the size of the buffer for the header by one byte. When the date was the longest month & day spelled out plus a two digit day of the month then it would overflow the buffer, resulting in the program crashing.
Decades ago I had to debug a random crash. It only happened on Wednesdays. On Wednesdays in September. On Wednesdays in September after the 10th…
I’m a 50+ year old IT guy who started out as a c/c++ programmer in the 90’s and I’m not that worried.
The thing is, all this talk about AI isn’t very accurate. There is a huge difference in the LLM stuff that ChatGPT etc. are built on and true AI. These LLM’s are only as good as the data fed into them. The adage “garbage in, garbage out” comes to mind. Anybody that blindly relies on them is a fool. Just ask the lawyer that used ChatGPT to write a legal brief. The “AI” made up references to non-existent cases that looked and sounded legitimate, and the lawyer didn’t bother to check for accuracy. He filed the brief and it was the judge that discovered the brief was a work of fiction.
Now I know there’s a huge difference between programming and the law, but there are still a lot of similarities here. An AI generated program is only going to be as good as the samples provided to it, and you’re probably want a human to review that code to ensure it’s truly doing what you want, at the very least.
I also have concern that programming LLMs could be targeted by scammers and the like. Train the LLM to harvest sensitive information and obfuscate the code that does it so that it’s difficult for a human to spot the malicious code without a highly detailed analysis of the generated code. That’s another reason to want to know exactly what the LLM is trained on.
Sue the mine in China that supplied the raw materials that went into the dielectric material in the capacitors in the power supply of the computer that facilitated the downloading of illegal content….
deleted by creator
Since we’re an iPhone family I use iCloud3 in HomeAssistant to track our devices. After setting that up and associating the phones with people in HA it was just a matter of creating triggers based on us entering the home zone:
alias: Somebody Arrives Home
trigger:
- platform: zone
entity_id: person.jack
zone: zone.home
event: enter
- platform: zone
entity_id: person.jill
zone: zone.home
event: enter
We live at the end of a dead-end, so I set up the home zone to extend down the road a bit. That gives iCloud3 enough time to figure out we’re home and trigger the automation in HA while we’re still approaching. I combine the above with a check to see if it’s roughly sunset to sunrise, and if it is then turn on the outdoor lights.
For doing things like turning off the lights when nobody is home, I have a similar trigger for everybody leaving the home zone, followed by a conditional that verifies everybody is away:
condition:
- condition: or
conditions:
- condition: template
value_template: >-
{{ states('person.jack') != 'home' and states('person.jill') != 'home' }}
I came up with these myself over time, they started 10+ years ago using a Mac platform called Indigo, but since moving into our new house a year ago I’ve switched over to Home Assistant as well. I have a background that includes 10 years of software development & another 15 of IT, so these sorts of things come fairly easily to me when I think about it a bit.
My wife and I use iPhones and I found iCloud3, which is a great add-on that tracks locations based on those devices. I started out with the washer/dryer notification with the simple realization that we regularly forgot about laundry in our washer in the basement since we’d forget that we had started a load. So I figured out how to have HA send us alerts when the washer or dryer finished. But I quickly realized that wasn’t all that useful if one or both of us was at work, shopping, etc. I then realized I could set up a trigger based on a time schedule that checks if we’re home or not, and only sends the alert if we are. So I have a HA automation that is triggered every 10 minutes by a timer, but the timer is disabled by default. That automation uses iCloud3 to check if either of us are home. If we are then the automation calls a script to notify us. If we’re not home then the automation resets the timer for 10 minutes. When the washer completes it triggers an automation that just starts that timer running. That gives us a little time to park the car, bring in any groceries, etc. before being reminded that we have laundry waiting.
Speaking of slot machines, every slot machine, electronic poker machine, etc. are just state machines that operate based on a stream of random numbers fed into them by another device.
The random number generators (RNG’s) used for gaming are highly regulated (at least here in the US) and only a small handful of companies make them. They have to be certified for use by organizations like The Nevada Gaming Control Board. RNGs have to be secured so only NGC officials and other key people can access them. If they are opened unexpectedly or otherwise tampered with then they need to go into lockdown and stop generating numbers until an official resets it.
The RNGs also need to be able to replay sequences of numbers on demand. If the same sequence of numbers are fed into a game and the user plays the same way then the result of the game should be 100% identical each time.