Slashdot Mirror
You've Got Spam: AOL Blocks 1/2 Trillion Spam
yohaas writes "Yahoo! News is reporting that AOL blocked more than 500 billion spam messages for its users in 2003. That comes to 40 messages a day per user. The company regularly blocks 75-80% of all incoming mail as spam! The article also lists the top 10 spam phrases for the year, including such come-ons as: 'Viagra online', 'Online pharmacy', 'Get out of debt' and 'Get bigger'."
It's been suggested in nanae that as a brutal display of the efficacy of spam-fighting and, most importantly, blocklisting, major ISPs all simultaenously turn off their spam defenses for a day to show users just how much UCE spew is clogging the internet every day.
Of course, the idea is repeatedly turned down for its utter lack of pragmatism.
But damn, 500 billion spams, and that's only to AOL.
Just imagine.
The instant clogging of mail-servers around the world and subsequent technological disruption might actually get the general computer-using public to take more of an interest in the fact that around 200 gangs of people are effectively raping and pillaging the Internet right under their eyes.
But then again, what can one do when faced with the Tragedy of the Commons?
I just took a gander at my logs on my postfix-amavisd-spamassassin front ends for one of my smaller ISP's and after doing the math, it's blocking ~36 spam/user/day on average (with spamassassin only blocking at score 9+). It doesn't surprise me that AOL is getting somewhere around ~40spam/user/day as it is more widely visible and the userbase as a whole is generally a lot more likely to do things that would encourage spammers . . .
In 2003 Spamprobe blocked just over 12000 on my personal domain, which is low compared to many others.
If tits were wings it'd be flying around.
iiNet is one of the largest ISPs in Australia (third or fourth now, I think). I got an advisory yesterday saying AOL and RR had both blocked all inbound mail from iinet as 'spam' They can crow about 500 billion mails all they like, but if a lot of it involves turning off mail from whole slabs of legitimate users, then it's not much of a service. The other thing is, if spammers are using trojans to create spam relays, then it's a bit hard to blame a particular ISP if a bunch of their users have been infected with this stuff. iiNet has a policy of advising users when they appear to be infected, they're cluey people too, they run everything on Debian as far as I can tell, and they have local mirrors for many Linux distros etc. I guess what I'm saying is that if you're going to block an ISP's mail you'd start with clueless behemoths who don't give a damn. Anyway, they appear to have a work-around in place, but RR is still blocking. Simon
Hal Spacejock: Science Fiction with Nuts
Slackware, what else when it must be secure, stable, and easy?
Although funny, it is also true. AOL has been randomly blocking entire ISPs - my hosting service's outgoing SMTP server was arbitrarily blocked by AOL for a total of about a month back around October. My hosting service had absolutely no violations of any kind, and after 2-3 weeks of bitcing and voice-mail-hell, AOL did finally respond, agree that they were not big-bad-purveyors-of-donkey-dick and unblocked them... Only to reblock them again in about 10 days, at which point my hosting service had to start all over again with them. It seems like the second time was the charm since I just sent email to an AOL user today and it didn't bounce (maybe AOL is now silently eating email instead of bouncing, that sure wouldn't make my life easier).
Anyway, from what I read on the net my hosting provider's experience with AOL's blocking of incoming SMTP connections is not out of the ordinary, many, maybe hundreds, of "little guys" have had the same experience. Makes me want to know the false positive rate for their spam blocking -- I'm willing to bet that AOL themselves don't even know the answer to that one.
When information is power, privacy is freedom.
Note: I did some thinking earlier on spam, and I figured I would post this the next time slashdot does a story on spam... You can find a link to this at:
http://sillygoth.com/journal/21669
This is my writing... I just want some feedback on it from the slashdot crowd.
Okay...
One of the things that I've been tired of recently is dealing with lots and lots of spam in my inbox. I've become even more tired of hearing about how there's a lack of solutions for dealing with it. It's one of the things that slashdot has been endlessly parading about.
To me, the primarily problem with spam is that emails are too easily spoofable. Solve this, and spam will become *much* more managable.
So, what technology is there right now that deals with certifying legitimacy?
Digital Certificates!
When you go to a site that's protected with https, the owners of the site usually have to get a certificate from a trusted source (Verisign, Thawte, etc) signifying that the site is legitimate (so that you don't end up giving credit card information to someone fronting for that company).
You actually *can* get a digital certificate for your email, but it costs money. Plus, to make something like that mandatory, each user would have to set up a certificate individually. Evil.
Why not move authentication to the domain itself? When accounts are setup on a user's machine, create an RSA public / private key per account. Simple enough.
When a user sends an email, force this user to relay the email through the mail server rather than directly from his/her computer. Force the user to authenticate their email / password to send the message. Some servers already force this, I believe.
When the user authenticates him/herself, encode a confirmation id using some elements of the email (first xx characters of message, subject, date, etc) using the RSA private key and attach it to the message.
Here's what should change with the receiving server... When a mail server receives the message, the mail server should initiate a separate connection that looks up the domain's MX server, and communicates with it.
This MX server should then provide the RSA public key for the account listed. The public key will then be used to decrypt the stamp that the MX server included with the message. If the stamp is legitimate, deliver the message to the inbox.
If a stamp is not legitimate, or there's no stamp, simply don't deliver the message. Simple enough.
This method has its series of strengths:
There would be absolutely no point in spammers taking over people's machines with viruses in order to send email if email must be sent through a qualified mail server. It's possible that worms could be written to auto-send messages through these relays, but at least then the mail server could detect it and shut the person out.
If mail sent is authenticated from a domain, people would then have the option to blacklist domains that aren't responsible for keeping tabs on its users.
Mail *will* come from where it says it's coming from. If not from the exact user on the domain, it'll come from that particular machine.
Of course, there are possible weaknesses to this strategy too.
If the mail server is hacked, hackers would be able to still send mail from it using the private key. Fortunately, they would only be able to send from email addresses listed under domains they own.
Spam software like SpamCop / Spamassassin / etc would be able to keep tabs on servers that exhibit hacked behavior, and temporarily blacklist these servers until resolved.
This doesn't necessarily stop users with legitimate email addresses from sending spam. Someone with a legitimate email address can still be spammed.
But at least when you block their email address or domain, it'll be a real email address, and a real domain name.
This method is not 100% in eliminating spam. But it's a damn good start.
/^[A-Z0-9._%+-]+@[A-Z0-9.-]+\.[A-Z]{2,4}$/i
Whitelisting makes sense--trusting certain mailservers more and not bothering with intense heuristics on mail coming from them. But blacklisting anyone you don't know makes none. The Internet is too vast to really implement something like this without huge costs and huge losses; I think solutions like this likely do far more to Balkanize the Internet than to protect it.
The solution mentioned in a previous Slashdot article a few days ago of making SMTP servers run a small computation per e-mail makes much more sense. This allows you to impose restrictions on non-whitelisted servers without completly ignoring them, either.
But when you talk about the anonymity preferred by the spammers, you ignore the fact that they are, in fact, selling a product. Forget the spammers. Track down their clients, the ones paying for the ads. Problem solved.
if you couldn't send anonymous snail mail.
Or anonymous e-mail. That's where this "signed" e-mail crap is going.
Imagine every message you send being tracible right back to you.
But hey, what's the trashing of rights in the name of convienience.
If you can send e-mails without being traced, so can spammers.
If spammers can't send e-mails without being traced, neither can you.
"Spammers are most afraid of being tracked and identified. "
Yeah, and nobody has a legitimate reason to not want to be traced.
I spent all of 2 hours modifying RinetD to do proper logging in between senders and my mail server. I spent another 3 hours writting a simple program to parse that log pulling out who a message is from, who it's going to, the subject line and what links it contains and the domains of those links.
Any entry "to" entry that isn't one of my e-mail addresses is deleted. The remaining are then examined for spam domains by looking at the froms and subject lines and the domains themselves.
A short list:
If expression both matches "*imgehost.com*" Delete ""
If expression both matches "*mydailyoffer.com*" Delete ""
If expression both matches "*topofferz.net*" Delete ""
If expression both matches "*adweawen.biz*" Delete ""
If expression both matches "*divineprice.com*" Delete ""
If expression both matches "*stamps.com*" Delete ""
And poof, no more ads from those companies and nobody's right to privacy is infringed. If they happen to have multiple domains for the same campaign I'll catch them as they come.
I will not support a means to subvert my right to privacy over some stupid ads.
How much are your rights worth to you? Not much apparently.
Terrorists blow up buildings and we get the patriot act. "terrorists" flood inboxes and you demand tracable e-mail.
Get bent.
Ben
Work Safe Porn
I'm a student at Notre Dame and work for the IT people and get to go clean compromised machines. generally any machine spewing spam gets picked up by university sniffers relatively quickly and their machine is disconnected before much harm could be done. also anything reported as spamming would be disconnected as well. they keep mac address records and such so that finding the computers is more or less easy. of course a lot of the stuff the IT people do is ass backwards at times and i'm sure at an engineering school like purdue they tend to do things a bit more sensically, so the chances of spam originating from a university with any sense at all is extremely small.
I never said I was smart, I just said I was smarter than you
How do you infer that? They block half a trillion messages. How many of those are legitimate e-mail? I have a great idea.... just send all your mail to /dev/null. You'll block 100% of spam. Might have the ocassional false-positive, though.
Our mail server has somehow erroneously been blacklisted and so we have added about 100 emails of that "Spam" to that half a trillion. I'm sure we're not alone.
The blacklists aren't infallible and get messed up and tend to be very slow to respond to errors or worse just don't bother (or even worse demand money to be removed in one noteable case).
What the article should say is that AOL blocked half a trillion emails, god knows how many of them were legit emails or how many really were spam...
Every now and then we'll wake up to find one or more of our servers blocked by aol, you can test it quickly by telnetting to port 25 on one of their MX's and it'll tell you right away if you're blocked.
Call, stay on hold 45 minutes, and you get "white listed" for 30 days and they ask you to setup a special email to send you spam complaints to if that IP becomes a problem again in the future. Sounds good right? I mean we host nearly 13,000 web sites for over 6000 customers, we DO get some spam sent through us once in a while (open formmail.php is the worst) and we handle it the second it's noticed.
HOWEVER we have YET to recieve ONE, and I mean that as in a SINGLE complaint from AOL for ANY of our ips. Yet 7 times now we've been blocked. Luckily it hasn't happened in a few weeks.
Do you know how annoying it is when 13,000 web sites become unable to talk to aol? Jesus christ.
Here's the funny part, often times it's only 1 or 2 of the (best I can tell) 4 main MX servers blocking us, so much for keeping those in sync.
I applaud them for trying to curb the incoming spam but goddamnit make it POSSIBLE to work with and if you block someone TELL THEM WHY and maybe a little warning please! If I'm notified of a problem I'll GLADLY nuke the spammers ass, or if it's just an open script, we can help the customer secure it, but if we're not informed what can we do? At least spamcop sends us emails with headers of the spam so we can take care of it.
So I gotta wonder how many of that half trillion is REALLY spam and how much is erroneous blocking.
--- www.f-theocean.com
I'm still trying to figure out what they aren't blocking. They block emails from mac.com even though a valid name, address and credit card number are required for a .mac email account, but they don't block free services like fastmail.fm or mailhaven.com.
If they really want to get a handle on spam, fwd:fwd:fwd Urban folklore... they should really block *@aol.com.