Date: 2008-10-17 22:34:00
journal pollution
I came home today and my email inbox contained about a half dozen notifications of new comments posted on some past LJ entries. At first I thought that the LJ site itself had gone haywire and randomly posted somebody else's comment thread to my journal, because it almost appeared to be snippets of a conversation between a few people. On further investigation though, the posting accounts were all recently created (within the past day), had no LJ friends, no userpics, and spammy nonsense in their journal. Furthermore, the text posted in the comment was found to exist on other random web sites (forums, blogs, etc). I've seen probably 20 of these comments so far, and the text in the comment is uncannily relevant to the rest of the post, yet was originally written in some totally different context!

For example, here is a sample comment:

The text appears to have come from an article titled Basic Writing 1000 (Google search to find it). This comment was posted on goulo's entry which happens to be about word usage. There must be some kind of matching algorithm going on to pick relevant text to post in a comment.

You will probably notice this sort of activity in your journal, if LJ doesn't implement some technical measure to stop it. The spambots are getting smarter, they are learning how to post to your journal with real-looking accounts. I don't have any clue what the ultimate goal for them is, unless it's just to practice making a smart bot to be used later for spamming.

I have anonymous comments on my journal screened automatically so I have to manually approve them. Up until today, most of the anonymous comments have been spam. If this keeps up, I will probably upgrade my default settings so comments from all non-friends are screened until I approve them. It's a sad state of affairs but at least some tools exist.

One thing I would like is the ability to close comments for a post. I have 454 past entries, all of which are potentially open for random drive-by spamming. If anybody from LJ happens to be listening (ha ha), please add a "closed to new comments" flag!
There was a wave of these a month or two ago as well. I assume the ultimate goal is to create more links to their bogus journal with it spam links, thus trying to appear higher in search engines results.

I've not had enough spam to make me start moderating more severely, though. I just delete the occasional spam comments. I wouldn't like to disable comments on old posts because occasionally real and interesting comments are made on them, by people who stumble onto them while googling for something that I happened to have blogged about.

If anyone from LJ is listening, they could probably get some mileage out of simply raising a flag for accounts that are created with journal entries full of dozens of links and comments on very old entries by other people, all in a short time span. Have a moderator proactively look at such accounts and (probably 99% of the time) delete them as evident spam activity.
I would turn off old comments on an as-needed basis. I seem to have a few entries that have been linked to externally, which increases the probability that google will find them. Those seem to be the ones that attract the most spam.
The only thing that happened to me today was that I had an account add me as a friend but, between then and when I checked email, the account had been deleted. Perhaps it was a spammer?

Incidentally, do you use LiveJournal in French?

Oui, je changer la langue de temps en temps. :)
I too got a flurry ofthese a while back. I simply set the comments to screened for the post and ignred any new ones. It's strange, but they were targetting very specific, old entries. I'm guessing that they met some criteria for a high rank on a search engine.
I get dozens of these weekly. It's really really annoying.
This reminds me of the spammers a long while back who were grabbing text from my blog and inserting it into spam emails to get past filters with real text.
I noticed that kind of posts too in my journal. What puzzled me at first, was that they were all in esperanto. But on second though, it's not that surprising assuming the bot matches text using a Google search.
Greg Hewgill <>