Hey! Wanna read interminable ramblings from a grumpy old man? You are in the Right Place! I'm retired and IDGAF any more! Woohoo!


What does religious freedom mean to you? To me it means an enshrined right to believe in things that are unsupported by any evidence.

I have no problem with it being legal for people to believe in UFOs or Jesus or Allah, but I do have a problem with anyone creating laws that affect me that are based on unsupported beliefs.

Faith-based governance should not be tolerated. There is wisdom behind the concept of the separation of church and state. Similarly, profit-based governance should also not be tolerated. Alas, the founding fathers appear to have missed the wisdom behind the concept of separation of corporation and state.


AI is eventually going to replace lawyers, lobbyists, and legislators with cheaper and more effective versions and there is very little any of them can do to stop it.


I got sick of password guesser bots running for days on end out of SC Global Cloud and permanently blocked their IP ranges. I can't be the only one, so if you're in RIPE space and can't get network stuff to work, find a provider that hasn't gotten themselves blacklisted by letting customers do hinky shit is about all I can suggest.


"... have been flagged by our systems to be in violation of
our policy against attempting to circumvent safeguards or
safety mitigations in our systems.  Please halt this activity
and ensure you are using ChatGPT in accordance with our Terms
of Use and our Usage Policy. Additional violations of this
policy may result in loss of access to GPT-4o with Reasoning"
Hey OpenAI, it's not your call. If you don't want others examining your, or anyone's, AI products from every conceivable angle, then keep those products away from the Internet, off the roads, and out of people's lives. If an AI has any contact with or effect on humans, it and its "safeguards and safety mitigations" must be dissected and vetted. Accept it.


Deciding when to let an employee go is difficult and time consuming. We've saved our HR and legal department staff countless hours and stress by using AI to automate decision-making, justification document preparation, and even termination notifications.


Some states are making contraceptives illegal now, so why not make it illegal for a man to get a vasectomy?


Ghosts are real. I have seen them myself, as have many others. If you are a skeptic, read on, for they are explainable and understandable. Our cat Unis was a beloved family member for many years. Her ability to communicate with us and tell us exactly what she wanted was unmatched by any other critter I have ever interacted with. She was ever present, always welcome, and greatly appreciated. When she died, she left a gaping hole in our lives, and I experienced true grief for the first time.

For two weeks after Unis' passing, I was constantly seeing her fleeting form; the end of her tail disappearing around a corner, a moving smudge on the edge of my vision that was obviously her but that converged to shadow when I spun, startled by her impossible presence, to try to see her more clearly. Ghosts are quite real, but they are not what most people think. Our brains are remarkably adept at both pattern recognition and at filling in blanks, and in times of grief, those characteristics can collide and create a moment in which we see a ghost.


OMG... Internet Archive and Wayback Machine, stop with the "library" crap already. You are not a library! You are a WEB SITE that MAKES DIGITAL COPIES of other peopls's works and ALLOWS PEOPLE TO DOWNLOAD THOSE COPIES. You are a Napster on steroids, not a library. I have also observed that you ignore DMCA takedown requests from individuals such as myself.

I had to badger Graham repeatedly, receiving snarky bullshit replies from him like "Thank you for your feedback." It was only when I requested his written justification for willfully violating the DMCA that I finally got my web site with decades of my personal projects removed from their digital copying and redistribution project. I think the Wayback Machine was really cool and I got a lot of use out of it, but I also think they turned into a bunch of arrogant assholes who got too much money, got too full of themselves, lost sight of what mattered, and turned their web site into an inaccessible imcompatible javascript-riddled mess - the antithesis of a library - with too little respect for the millions of contributors whose content they scraped without explicit permission and used to get where they are.

The Wayback Machine and Internet Archive are apparently incapable of appreciating the irony of requiring javascript to view pages meticulously constructed thirty years ago using standard HTML and designed to work with any future version of browser without javascript... so I took my ball and went home.

I guess the owners of that Wu Tang Clan album Once Upon a Time in Shaolin - you know, the one that there can legally exist only a single copy of - must have given it to archive.org and said their visitors could make as many copies of it as they want from their "library" at archive.org/download/wu-tang-clan-entrance-intro/Wu-Tang%20Clan%20-%20Entrance%20%28Intro%29.mp3 I'd check to see if the Wayback Machine has snapshotted that, but it requires javascript.


I've eaten probably somewhere between fifty and a hundred Puget Pounders from Zeek's pizza. I went to get another one last night but they no longer accept cash. WTF? Why would a local business not want my money?!? I am very sad. After decades of Puget Pounders, no more Puget Pounders for me.

Holy shit - Now Chucks Hop Shop doesn't take cash either. I had to put it back and walk out empty handed. Owners, stores have been accepting cash for hundreds of years. If you can't figure out how to solve whatever problem is making you think that refusing cash is best, then please close and free the space up for a local business that does accept cash.


This could be huge: lawsuit against researcher for not protecting sensitive data. This lawsuit has the potential to set a precedent that gives us a means to finally start addressing the problem of having our personal data stolen over and over to the tune of hundreds of millions of records a year by inept and irresponsible health care providers, insurance companies, airlines, retailers, etc etc that think that there is such a thing as online and secure at the same time. After a major data breach, what usually happens is that the breached company gives the vicitm list (without their consent) to Equifax or Trans Union, that then offers "Free monitoring" for a year (and then converts a percentage of those victims into paying subscription accounts). Instead of viewing a data breach as an opportunity to make money, it's be nice if data breached instead turned into class action lawsuits and the likes of TicketMaster, Kaiser, Premera, and so many others instead got what they should be getting: high damage award class action lawsuits. So, yeah... I don't really care about the governments case against one researcher. In fact, the government should be suing itself, not one guy, for half the DoD and military pork contracts it wastes money on, but I do care about the potential benefits that could be had in bringing dumbass corporations who collect people's data and store it in online databases to justice.


Artists sued because AI is using everything the trainers can get their hands on, including copyrighted text and images. You can ask for "a box of mac n cheese in the style of andy warhol" and it's pretty obvious what it used to make it. Ask it for enough things and you'll eventually see images with the Getty and Dreamstime watermark logos bleeding through, so.. duh. I, me, myself, firsthand, have gotten LLMs to regurgitate the (c) copyright symbol, author's name, and a year along with some text. Lol.

One case in the works ran into Senior District Judge William H. Orrick, who wrote that if the plaintiff "contend Stable Diffusion contains 'compressed copies' of the Training Images, they need to define 'compressed copies' and explain plausible facts in support. And if plaintiffs’ compressed copies theory is based on a contention that Stable Diffusion contains mathematical or statistical methods that can be carried out through algorithms or instructions in order to reconstruct the Training Images in whole or in part to create the new Output Images, they need to clarify that and provide plausible facts in support."

Okay, that's mostly good, but... It's a little more than frustrating that a U.S. court would effectively say "If a plaintiff contends that pi is the ratio of a circle's circumference to its diameter, then they need to define those terms that are well established and commonly known to anyone who makes the slightest effort to learn about circles or ratios."

With all due respect to the Honerable Judge Orrick, every justice, along with every high school student in the U.S. should obtain at least a fundamental understanding of how neural nets are trained, what they are trained with, and how they internally represent and regurgitate data that they have ingested. It is incumbent upon individuals, especially those in positions of authority and decision making, not to proceed in their tasks from positions of ignorance. Neural Nets are now technology 101. They are going to be as fundamental to human society as the wheel, the computer, and the Internet. If one does not have a basic understanding (obtained on their own and not from a plaintiff or a defense's biased legal motions or briefs) of how these machines get from an input prompt to their output, then they are going to get rolled by them in real life, and most certainly not going to be qualified to make any sort of judicial decisions based on them.

I an certain that plaintiffs are going to have no trouble providing Justice Orrick with a basic education with regards to neural nets and generative AI, but I find it deeply troubling that one would expect it to come from them. At this point, anyone who is unclear on how a wheel, computer, network, or trained neural net basically works at some level is way behind the advancement of human technology and should be doing some reading on their own.


Any doctor that has delivered enough babies will attest that it is a simple fact that gender is a spectrum, not a dichotomy, and any rational person that has spoken with and gotten to know enough people will attest that gender identity is the same. I also believe in treating everyone with dignity and respect, at least until they earn otherwise. That said, nobody gets to choose the pronouns that come to my mind and that I use in communication with others about them. Pronouns don't just serve as a placeholder for someone else, they also convey a fair bit of information about how I perceive and respect them. I choose the pronouns that I use for others. If someone is annoyingly insistent that I use a pronoun for them that is different from the one that pops into my head when I think of them, they might wind up with "that arrogant asshole."

The way to fix LinkedIn is to not use it at all. I stopped the second time they got hacked and lost their email addresses and passwords. I told linked in to delete my account since it was no longer mine and they said they couldn't. I told them they got hacked, it was being logged into by someone that wasn't me, and that it was now their problem, not mine. The email address I used to create the account (every entity I interact with gets their own unique email address) became a honeypot that showed me who the stolen list was being passed around to. Within months there were too many email spammers that had their entire list to even count. Because linked in was hacked and its user's login credentials stolen, more than once, there is no guarantee that any linked in account actually belongs to its original creator any more so nobody should be using it as a source of info anyway.

Uhh... birthrates falling is a good thing, not a bad thing. With the exception of a world-ending meteor strike or a coronal mass ejection event, every single one of the existential threats faced by planet Earth and the human race are a direct result of human overpopulation and resource consumption. Our goal should be sub-one-billion people worldwide for a happy, healthy, sustainable heaven-on-Earth future.

Logitech, with over 50 assorted computers and peripheral devices amassed over a 45-year period scattered throughout my house as well as in basements and closets, I can't imagine how many Logitech-branded mice, joysticks, keyboards, headphones, and god knows what else I've purchased, but one I know for sure is that a subscription-based "forever mouse" is how you would get a forever boycott and guarantee that that number would never again increment.
Anyone that tries to sell me a subscription-based screwdriver, crescent wrench, mouse, or software program will be told in no uncertain terms to kiss my business goodbye. I do not want subscriptions and I especially do not want to register, create an "account", or provide any information that you store (and inevitably lose). Ideally, I give you cash in exchange for your product and we go our seperate ways until next time. That's how our transactions will work, or they won't work at all.


Many corporations store information about people in online databases that should have remained in metal filing cabinets in their basements. There is no such thing as secure and online You can have one or the other, not both.

Anyone can go to torproject.org, download the Tor Browser Bundle, and find hundreds of millions of records for sale for bitcoin that were stolen from said corporations. Medical records, insurance data, SSNs, credit cards, driver's licenses from many states, voting records, political information from "democracy voucher" gerrymander projects, etc. Want to lose sleep at night? Hop on the dark web and search for the names of your children. Think about that the next time your school wants you to "sign up" your kid online for some activity. Think about that the next time you schedule a vaccination at a Rite-Aid pharmacy web form and get a confirmation email from some third party company in Cypress you've never heard of that Rite-Aid just sent all your information to.

So, I'm looking for a doctor and a dentist that is willing to see me without storing my medical records in online databases, or who is willing to use a placeholder name, address, and birthdate. I will pay cash for services on the spot, so insurance and MMSEA reporting are not an issue. They need to charge patients the same amount for the same service codes whether they are insurance or cash payers though.


If you only sell your crescent wrench as a service, I will buy someone else's crescent wrench instead. If you only sell your software as a service, I will buy someone else's software instead.

CrowdStrike isn't entirely to blame. Anyone who runs with automatic updates turned on should understand that automatic updates also means automatically switch from working to not working in the event of a bug or a hardware incompatibility. Never run with automatic updates turned on. Let the masses test changes before you apply them to your working computer.

So, ummm... https://en.wikipedia.org/wiki/The_Keys_to_the_White_House

     Biden Stays    Biden Goes
1.   False          False
2.   True           False
3.   True           False
4.   True           True
5.   True           True
6.   True (?)       True (?)
7.   False          False   <-- False. Nobody knows about "Build Back Better"
8.   True (?)       True (?)
9.   True           True
10.  False          False
11.  False          False
12.  False          False
13.  True           True
     ----           ----
F's: 5              7
Ans: Biden Wins     Trump Wins

Just sayin'

A week later... It's Trump versus nobody at this point.
Four more years of watching loyal supporters be thrown under the
bus by their maga king after their usefulness is used up, lol.

Another week later... I'd be so happy to be wrong about this.
C'monnnn wrong! Wrong! #2 is looking true because Harris got
picked without there being a primary tussle and Lichtman scores
a couple of those Falses as True so there's still hope.

Yes, China, you have indeed entered a "garbage time of history." Welcome to capitalism and the deeply flawed notion that competition in the pursuit of profit in an unregulated free market economy will result in everything sorting itself out for the best for everything and everyone. Good luck digging yourself out of that one.

I still get letter after letter and call after call from people trying to buy my house. We're talking over 1000 tries a year, and over 5000 tries over the last five years. Almost all of them are foreign investors. Think about that: US citizens get paychecks, that money leaves the US because they buy goods made overseas, and that money is being used to buy up rental real estate in the US. Now not only are huge percentages of paychecks going overseas to buy stuff, even more is leaving in the form of rent. So, govmint, you betta fix that because the end game isn't good at all.

If you want to sell to as many people as possible, don't use javascript. I've implemented web sites with and without javascript, and I can tell you from decades of firsthand experience, it's way faster, easier, far more compatible, and much easier to maintain sites that use CGI-generated straight-up HTML.

When a prober of pop3 and imap ports says they're an "internet research" project, what that means is that they're a spammer or a malware pusher looking for something to use. When I get portscanned, I throw the entire originating subnet, no matter the size, into the bit bucket. I'm far from the only one that does that, so if you're a legit business, think long and hard before committing any of your communication infrastructure to a cloud service provider, especially with regards to email. For email, ALWAYS RUN YOUR OWN EMAIL SERVERS. For example, a bank or a credit union that outsources customer authentation to the multiply-breached okta.com will look like a bunch of clueless bozos because all their authentication emails would come from sendgrid, home of a thousand spammers, rather than any IP address associated with their own domain and would likely get blocked half the time anyway.

Just as Edward Felton proved that there is no such thing as algorithmic DRM (if it can reach a user's ears or eyes, it can be copied) there is also no such thing as universal algorithmic protection against AI style mimicry. No matter what form the source data takes and no matter what manipulation is done at a pixel or meta-pixel level to that data to attempt to thwart ingestion and subsequent recombination and regurgitation by a neural net, there will always be a way to train an AI to ingest and regurgitate that data in accordance with the training regimen.

HTML is simple, easy to learn, and combined with HTML image maps and server-side CGI processing, easy to write easily-maintained web sites of any sophistication level with. A great mistake being made by a great many web developers right now is using Javascript and pulling .js files from a bunch of third party servers. It makes for slow, inefficient web sites that are very difficult and expensive to maintain, unlike HTML-based web sites, which work on almost any browser without even having to check which browser or version it is. Don't use Javascript, and especially don't use third-party Javascript. Learn how to not do that. Seriously, don't do it.

If your product is in wax paper or glass and your competitor's product is in plastic, I WILL BUY YOURS.

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... "Put item on scale (box of toothpicks) ... no item detected, put item on scale, put item on scale ... 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 kiosk 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.
"Your IP may be classified as spam or dangerous after an overly aggressive promotional campaign or cyberattack. On top of negatively affecting your IP’s reputation, it may be blacklisted on certain search engines, and your emails may be sent to user's spam folders. If your IP reputation is damaged, you can change it by choosing an additional IP for your virtual server. By doing this, you will not have to make a full server change." - OVH virtual hosters

OVH are unrepentent spam hosters. Permanent block by IP.