AOL Blocks Links from LiveJournal

Evan Martin writes "LiveJournal.com is an open-source weblog site with over a million users, some of whom use AOL. Last week, AOL began blocking all HTTP requests with "www.livejournal.com" Referer headers. This is a common practice by image hosting sites to prevent off-site linking of their images and 'bandwidth theft'. However, in AOL's case, they're blocking everything, not just images, effectively breaking all links to any AOL member's site--but only from LiveJournal. To be clear: nobody on LiveJournal can even make a link to any AOL member site without getting a '404 Not Found' error. We've also heard reports of the same thing happening on AOL properties (Netscape, Compuserve). This concerns us because we have to deal with the support requests: it worked in the past for our users, and it continues to work for other sites, so our users think it's our fault."

Martin continues: "We've tried to contact AOL three different ways, all without success. We've also told our users to contact their tech support. At one point, an AOL staffer pointed out that FTP access still worked (which is probably because FTP has no "Referrer" concept), and so, as an interim fix, we're rewriting all HTTP URLs to use FTP on the AOL properties where that works instead. This means that users can again host their images on the AOL webspace they're paying for, but more importantly, it means they can simply link to their webpage.

We wouldn't be so upset if they were simply blocking images. Bandwidth use is a valid concern, after all, and we even provide step-by-step instructions for people to configure their webservers to prevent image "theft". However, because they're blocking all access, including regular links, this looks like it's either a mistake, or something more insidious (the conspiracy theorists have pointed out that AOL has just launched their own competing weblog product, also based on "journals").

Although CI Host sued AOL recently for being blocked, we really don't want to do that. We still suspect that this was all just a mistake, and hopefully, by making this public, we'll manage to get their attention, since all our previous attempts have failed."

Comments

Re:Quick fix for HREFs viewed by MSIE

[cpeterso]

People keep posting that the whitespace insertion is a fix for looong lines. If so, then how did the "I love wiiiiide posts" troll make really wide posts? Was that an IE bug? I haven't seen the wide posts troll in a while.. I miss him. :-)

Re:Wait a minute

[Phroggy]

I regret to inform you that all Slashdot comments are not posted by the same person. Not only are there different people on Slashdot, some of them have different opinions. It's a shocking revelation, I know.

That's not true! You're wrong. Each Slashdotter is part of the herd; we all think alike, and agree on everything. Why do you repeat such lies?

Re:Wait a minute

[DAldredge]

Wrong. It's just me and you. I created all the other 600,000+ accounts because I was board and needed someone to talk to.

Sorry you got caught up in the joke.

So when I suggested it, I was 'an ass'?

[The Monster]

(The space added to the URL you pasted in is added to every long word at the 50-character mark, to make sure idiots can't break your browser rendering by typing very long words into their comments.)

I think Taco owes me an apology. I suggested something QUITE similar

as a first cut at it, I said to try this Perl fragment as an example of the logic needed to take care of 'page-widening' trolls:

s/(\S{80})/\$1 /g;

OK I used 80 instead of 50 as the magic number, and I know that would break really long URLS, (not as many as 50 though) but so many people block links from Slashdot anyway that you probably have to fiddle with the Address field just to get around that...

in email over a year ago, and his response was to call me an ass

I guess I don't have to worry about moderating ever again...

Re:So when I suggested it, I was 'an ass'?

[The Monster]

The page widening trolls had nothing to do with spaces, they exploited a browser's (broken) line-wrap logic by prefixing every word with a dot.

But if you insert a space, between two non-whitespace characters, then you turn one word into two, and unless the second word happens to begin with a dot, (The trick would be to force the breakpoint to the left if the following character is a dot) the page-widening troll is foiled. With a little fine-tuning of the regexp, it should be possible to get something that works, and can't be circumvented. I suspect that's exactly what's been done here, as I no longer see that particular troll technique.