Zombie Posts, Usenet Regurgitates.

Usenet gets positively weird sometimes, even on a technical level. Lately 
I've been seeing a lot of new posts that look familiar. That's because 
they are, they're old posts that are somehow being autoreposted. So the 
posts of the past are being regurgitated into the present and old names 
that have long disappeared from the scene are reappearing in some groups. 
Like zombies rising from the grave, these old posts are being brought to 
life again. I've seen content from 2010 to 2019 autoreposted the last few 
weeks. This also happened last year for a short while but the posts were 
always incomplete. This year they are mostly but not always complete. 
Because it takes some time to load several gigabytes of headers, I check 
for new headers with a build of Pan that has indexes going back to 2016 
and do deep indexing with another that has indexes going back to 2008. 
I'm seeing all the zombie reposts in the build for new headers as being 
posted now but the build with deep indexes doesn't reload them if it has 
them already indexed when they were originally posted. I think that's 
because the posts are carbon copies with identical headers including the 
same message ID in some cases. So from a Newsreader's perspective they 
are the same articles and it doesn't reload a header it already has in 
its index with the same message ID. The only difference in the header is 
the date. Date is somewhat arbitrary in Usenet headers and is easily 
spoofed. All the NNTP protocol cares about is that the date is in the 
right place in the header but not if it's accurate and several different 
time and date formats are accepted. Message ID is much more important. In 
theory, each post should have a unique message-id even in a group with 
billions of posts.

There was some speculation last year about it being due to a server 
issue. There wasn't anything concrete to back that up, it was just an 
assumption. I think that could be possible if some new server is syncing 
files and there's some sort of mismatch in their configurations but I 
also think it's possible someone is coming up with code that does this. 
There's a lot of randomness in what groups and how much it's happening 
and small but noticeable variations in the zombie posts in different 
groups that make me think that someone is experimenting and tweaking 
their code. If it was a server to server issue, I'd expect things to be 
consistent across all groups and instances and they are not. There's a 
lot of custom code for Usenet out there, especially when it comes to 
posting. There were obfuscated posts long before the usual suspects, Nyuu 
and nGpost, were publicly available. I myself am using a highly 
customized build of Pan and I'm far from the most adept of coders. And 
these days you can get AI to write your code for you. So I'm more 
inclined to think someone out there is doing this. Why is an open 
question. One possible motivation for this is just screwing around with 
usenet code, doing it just because you can. I've seen some obfuscated 
garbage reposted along with comics and movies so I don't think what the 
actual content has anything to do with what gets reposted. It just takes 
a group with a lot of old content and not every group with content from 
2010-2017 is getting hit with the zombie posts. Another possibility that 
I've thought of is resentment against the Omnicron backbone providers who 
have been very aggressive in their business practices and have a monopoly 
on retention beyond 2016. Reposting their old content could be seen as 
some sort of attack on the crown jewel of their business empire. That 
being said, I'm just idly speculating and the more I think about it the 
less likely a possibility I think it is. Even if someone was doing this 
to screw with Omicron I don't think it's going to do much for that cause. 
There's just too much content to be reposted and randomly reposting some 
of it doesn't do much more than give a taste of what's there like giving 
free samples of what you can get if you get a Newshosting subscription. 
You might as well pay the price and get the full deal. I'd much rather 
have groups nicely indexed in chronological order from their backbone 
than just potshot reposts on another. Even for files post 2016, I've 
found a lot of the other backbones will just have the headers for them, 
but only Omnicron has the actual files. That's petabytes of data that 
would have to be posted to change this. 

As for the effects, the unintended consequences, it complicates indexing 
for sure. I've had to make adjustments and redo indexes, in some cases to 
my benefit. I've found that some of my early attempts at deep indexing 
had flaws I wasn't aware of and I now have access to old posts I didn't 
see before. It's also effected the online indexers, Nzbindex and 
Binsearch are filled with the zombie posts. It's getting hard to see new 
posts with all the regurgitated posts in either indexer but on the other 
hand, a lot of old content is reappearing when there's not all the much 
new content being posted and a lot of that is obfuscated. Unfortunately, 
some of the early obfuscated posts are also reappearing. Example groups 
for this are alt.binaries.comics and alt.binaries.comics.repost where 
garbage from 2017 is starting to reappear. So we are getting a taste of 
the old content posted years ago that includes some spam and garbage too. 
On the positive side, I can't see DMCA being applied to the zombie posts. 
Most of the posters are long gone and they can't get slapped with a DMCA 
notice. So they are among the immortal undead and can't be stopped. 


The main tools for researching this are with my two Pan builds and noting 
what gets through on my deep index build. As previously mentioned, in 
some cases, I've found issues with some of my indexes, they were 
incomplete for some reason or another. Once I redid them, I was left with 
just a few zombie posts and these differed somehow from the original. In 
one specific example, in alt.binaries.comics.dcp some of the reposts are 
being reposted without the article part structure so what was originally 
posted as a binary file with one header is reposted a series of text 
posts with the same name and a lot of headers, one for each part of the 
binary file. This is annoying, it takes up a lot of header real estate 
and you can't download the file unless you get the original post from 
years ago. It's also a good example of variation in the zombie posts. 
It's not happening in any other groups I've checked. Some groups with a 
lot of old content are not getting hit with this at all while others are 
getting it daily. 

Addendum 7-11-25
Ok, so the pace in some groups is really picking up, the worst right now 
is alt.binaries.comics.dcp where the zombie posts that have the article 
part structure removed or are different in some way from the original are 
frequent so I'm seeing a lot of zombies when I load new headers in my 
deep index build. I'm not even bothering with my other build which would 
spend an hour indexing all of them. Binsearch has a huge number of them 
with scattered dates. The newest date I've seen is 2024 in this group. 
Second group that's bad is alt.binaries.e-book.magazines where SinnerG's 
huge archive is being regurgitated. Other groups are lightly or 
unaffected and I'm more convinced that this is deliberate and not just a 
server sync issue. It's starting to look like a zombie apocalypse and all 
the old posters are rising from the grave in these groups. It's getting 
hard to find anything anyone's posted recently in these groups with all 
the zombie posts. 

This post is an ongoing draft. As this new Usenet phenomenon is ongoing, 
I'm getting more and more examples daily to take notes on. 




-- 
PX=Paradox