Okay, this is knee-jerk stupid from some very software-ignorant bureaucrats. Banning software because you're afraid of what might happen without any code-based evidence that what you're afraid of exists in that software is multiple kinds of stupid. Listen up, Dept. of Commerce boneheads: You don't actually know that there's a problem in Kaspersky's products and you obviously haven't examined it and found anything or you would've said so. Even if we accept that there's no need to bother with things like evidence or even common sense to make decisions regarding useful commercial products with, the fearmongering is still stupid because the instant that a Kaspersky update attempts to install a rootkit or modifies a file it shouldn't or attempts to make an outbound connection it shouldn't or that fails to detect a virus that ten other AVs do, we'll know because one or many of us tech nerds is as likely to spot it in Kaspersky's AV software as in any of the other thousands of software products that could just as easily be used for nefarious purposes. If you're concerned about one-off targetted attacks, the least sensible approach is to get rid of things that *might* be exploitable. You won't have any software or hardware left if you go down that road. Instead of worrying about Kaspersky having KGB programmer plants because of vague concerns or because the company name ends in "sky" (I bet the actual ban reason is literally that stupid), how about worrying about the thousands of barely-vetted third party private contractor corporations that you give billions of U.S. dollars to for plenty of other backdoor-able and compromisable tech? How about worrying about all of the US government agencies that have recoded their web sites to require javascript and refuse connections from the very secure noscript Tor? How about worrying about the entire US healthcare industry handing over the names and medical records of millions of US citizens, including government and military personnel, to the Russian government by storing names and medical records in online electronic databases instead of secure metal filing cabinets like they used to? The current levels of stupidity are indistiguishable from the KGB having already had their way with all sorts of stuff. Want to ban things that Russia might be able to slip something into? Don't stop at Kaspersky! Ban Linux and Windows and all the other software products you use on a daily basis that have Russian nationals working on their codebases! Ask Microsoft how many Russian nationals have worked on and are currently working on Windows and other products they produce! I double-dog dare you! I'll type it again in caps because it's fun to yell it: ASK MICROSOFT HOW MANY RUSSIAN NATIONALS THEY HAVE CODING ON PRODUCTS YOU USE! Man, let's all hope nobody mentions to anyone at the U.S. Dept. of Commerce how AI, github, or python work or that'll be the end of AI in the U.S.! Hmmm... maybe that wouldn't be such a bad thing.

A conversation with AI LLM Zephyr-7B.

If your web site uses the javascript "lazy loader" I won't see any of your pictures.

Businesses: The email address that we give you is unique, created just for you. Don't share it with anyone.

Another software product falls to subscription-based licensing. Bye VMWare! We had some good times together but you've gotten a bit clingy and it's time I moved on.

Using an LLM as a writing assistant is like using a nailgun instead of a hammer to install a roof. Rejecting scientific papers solely because you *think* they were written by an AI is like assuming that roofers used only nails and no plywood, tarpaper, or shingles.


"Many of us like to ask ourselves, 'What would I do if I was alive during slavery? Or the Jim Crow South? Or apartheid? What would I do if my country was committing genocide?' The answer is, you're doing it. Right now.'" ... "I will no longer be complicit in genocide... I am about to engage in an extreme act of protest." - Aaron Bushnell Feb 25, 2024
Mr. Bushnell lighting himself on fire in front of an Israeli Embassy in protest of the ongoing killing of civilians in Gaza (if it's not disturbing, you're a sociopath)


This is why you should not store patient data or medical records in electronic form: Medicare, CVS, MetLife, etc etc. Data cannot be both online and secure. You can have only one of those for a given collection of data. Patient data transmitted to and stored in electronic databases is stolen to the tune of hundreds of millions of records a year. Paper charts in filing cabinets are not, and cannot be. This problem would vanish if patients stopped consenting to the storage of their info in electronic form. Paper medical charts still work great.

While I'm ranting about this stuff, I'd like to remind the world that Medicare violated USC 42 1395b when they mandated reporting by SSNs and told private health insurers to not insure anyone who wouldn't provide an SSN. Still waiting for the Department of Justice to actually enforce the Social Security Act USC 42 1395b and indict the Medicare administrators responsible for the design of the MMSEA reporting API and telling insurance execs not to insure people.


GREAT NEWS!!! The House of Representatives passed a bill requiring TikTok's owner to sell the company or be banned! Now that precedent has been established, we can move on with getting rid of Facebook, Twitter, Youtube, and the rest of the harmful social media cesspool!
Again, there is no such thing as both secure and online. Stop storing medical records in online databases.

I hate automated checkout kiosks at grocery stores. Every single time I've tried to use them, it's "Put item in bag (40 lbs of cat litter) ... item is not in a bag... beep beep beep Get assistance" over and over until the attendant gets tired of sliding his go-ahead card through the slot and hits something that shuts all the fucking errors off. I've noticed that typical shoppers go through with five or six items, and maybe it works fine for them, but I go twice a month and have $300 in 80 items in a cart and what would take five minutes at a checkstand takes fifteen minutes and a bunch of attendance overrides and a hunmdred times more anger and frustration for me at one of those error riddled bugfest kisosk bullshit machines. I'm not going to buy groceries from a giant vending machine, so Safeway, if I don't see any checkstands open with human checkers at them, I'm walking out.

The OpenAI crawler kept doing nothing but checking robots.txt over and over every 20 minutes here for months. I finally tried going to the URL in the user agent string www.openai.com but the fuckers block tor so I blocked them. Enjoy the 403s, OpenAI.

(This is unrelated; I blocked OpenAI because their crawler sucks, not because they're a major copyright violator
(along with all other large scale neural net trainers to date)):

Suppose an author writes "This is copyrighted."
Now suppose we encode that sentence in a database using tokens and weight
parameters such as "This ", "is ", and "copyrighted." like this:
  "This "        --> 0.9 "is "
  "This "        --> 0.1 "copyrighted."
  "is "          --> 0.9 "copyrighted."
  "is "          --> 0.1  "This "
  "copyrighted." --> 0.1 "This "
  "copyrighted." --> 0.1 "is "
  "copyrighted." --> 0.8 *STOP PROCESSING*

Is creating a database of tokens most likely to follow another token a violation
of copyright? It can be used to regenerate the sentence "This is copyrighted."
How is that different than saving a PNG image as a lossy-compression JPG image?

AI startups and AI hardware manufacturers say it's fair use.  Authors say it's
violating their copyrights.

I'm not an author. I'm 100% pro-AI development. I think the recent advances in machine
intelligence are astounding and the biggest thing since microcomputers and the Internet.
The implausible conversational and reasoning abilities of robots and androids in
many sci-fi movies are not only plausible, but happening. I'm pretty excited about that.

However, I also think the copyright question is a no-brainer. Copyrights were, and are,
being violated on a mass scale to build LLMs and GANS image generators and recognizers.
Pictures of people's faces have been used without consent. Web and social media sites
have been and are being harvested and reencoded without knowledge and consent of their
owners, just as they were prior to the advent of AI.  I've generated text using an LLM
with their copyright notices and author's names intact.  I've generated images with their
Getty Images and Dreamstime watermarks intact.  The copyright question is a no-brainer.

Google Chrome's plan to limit ad blocking extensions guarantees that I will never use Google Chrome.

Digital Ocean is block-on-sight at this point.  Hacking attempt sources this week:
107.170.0.0/16     digital ocean
159.203.0.0/16     digital ocean
161.35.0.0/16      digital ocean
162.243.0.0/16     digital ocean
165.22.0.0/16      digital ocean
167.99.0.0/16      digital ocean
198.199.64.0/18    digital ocean
206.189.0.0/16     digital ocean
209.38.0.0/16      digital ocean
64.227.0.0/17      digital ocean
172.104.0.0/15     linode   soooo much spam
173.255.192.0/18   linode   sooooo much spam
198.58.96.0/19     linode
66.228.32.0/19     linode
5.180.181.0/24     cloudwebmanage - trying to look as bad as digital ocean?
185.139.228.0/24   cloudwebmanage
185.47.172.0/24    cloudwebmanage
45.147.250.0/24    cloudwebmanage
172.200.0.0/13     microsoft   stop using outlook is about all I can suggest
20.118.0.0/16      microsoft
4.240.0.0/12       microsoft
52.160.0.0/11      microsoft
162.216.148.0/22   google cloud  hey google, accept abuse reports by email same
35.192.0.0/12      google cloud  as everyone else or get off the internet
13.40.0.0/14       amazon ec2    hey amazon, accept abuse reports by email same
13.52.0.0/16       amazon ec2    as everyone else or get off the internet
34.192.0.0/10      amazon  34.209.250.210
147.185.132.0/22   palo alto networks - i don't care, don't try to login
147.185.136.0/22   palo alto networks - i don't care, don't try to login
185.180.143.0/24   internet censys - i don't care, don't try to login
45.83.64.0/22      more supposed 'security reasearch' DON'T CARE: login=blocked
185.242.226.0/24   criminal IP - i don't care: try to login, get shitcanned
154.212.141.0      cloud innovation seychelles - uh-huh... seychelles
104.152.52.0/22    rethem hosting
115.231.78.0/25    chinanetzj
122.224.129.234    chinanetzj  200~122.224.129.232 - 122.224.129.239
152.32.252.0/24    ucloud china
165.154.164.0/23   ucloud china
183.129.159.240/29 china
185.219.141.0/24   packethub  panama
94.165.0.0/16      italy
194.165.16.0/24    flyservers
43.157.79.0/24     aceville tencent
49.51.128.0/17     tencent cloud - getting tired of seeing these
8.222.187.0/24     alibaba cloud
91.238.181.0/24    onehost

If you live in any of those, you have my sympathies.