Catch up on stories from the past week (and beyond) at the Slashdot story archive

 


Forgot your password?
Close
typodupeerror
×
Spam Government Politics

Risks of Partisan Spam Filtering? 63

Posted by Cliff from the filter-fairly-if-you-filter-at-all dept.
Mike1024 asks: "Pete Klammer reported in RISKS 23.95 about spam filtering software filtering political e-mails - including Postini blocking certain anti-Schwarzenegger URLs and Comcast blocking e-mails mentioning afterdowningstreet.org. This could be caused by malicious action, misreporting of spam, 'joe jobs', or actual spamming. With many people using their ISP's default settings, and manual spam filtering being impractical for many users, what can be done to avoid giving ISPs and anti-spam companies extensive, fully automated censorship abilities?"
This discussion has been archived. No new comments can be posted.

Risks of Partisan Spam Filtering? More

Risks of Partisan Spam Filtering?

Comments Filter:

  • Gee, stop asking for donations? (Score:4, Insightful)

    by Marxist Hacker 42 ( 638312 ) * <seebert42@gmail.com> on Wednesday August 03, 2005 @02:24PM (#13232602) Homepage Journal
    That would seem to me to be the #1 cause of political spam being filtered. #2 would be the outlandish use of HTML when a text message would do just as well. If they stopped just those two behaviors, most of the spam filters would let the messages through just fine.
    • #1. The sending address is on a blacklist because that address sent a message to a spam trap.

      #2. The sending address is on a blacklist because people received the email and submitted it as spam.

      #3. If #1 or #2 apply, web links inside the message can be classified as "spammy" content. This includes links to graphics, articles, and so forth.

      #4. See #3, but key words can be classified as "spammy" content. The easy way around this is to use the "this is not spam" function that should be available to you.

      So, the

      • #1. The sending address is on a blacklist because that address sent a message to a spam trap.

        It's not so simple since many return addresses are forged. And unless the spam gets bounced to the forged return address, the owner is likely unaware that their address is being forged and can't pursue corrective measures.

        #4. See #3, but key words can be classified as "spammy" content. The easy way around this is to use the "this is not spam" function that should be available to you.

        All this filtering d

        • It's not so simple since many return addresses are forged. And unless the spam gets bounced to the forged return address, the owner is likely unaware that their address is being forged and can't pursue corrective measures.

          Sorry, I should have said "sending IP address". There is no way to forge the IP address of the machine sending you the message and still receive the message.

          All this filtering does is selectively breed spam generators with more realistic content, sort of like a slow, manually operated g

          • By realistic, I mean that the headers are now mostly RFC compliant. In 1999, nearly 100% of the spam I filtered had malformed headers. However, the filler in the body is getting more realistic too, or at least difficult to filter automagically.

            Yes, most spam is probably sent via zombies, but that's not who I recommend going after. (Though a blanket ban on MS-Windows by all ISPs would help there.) Who I do recommend going after in my earlier post are the businesses advertising via spam. If they want

        • That stopped the junk fax problem which some here may be old enough to recall.

          It did? I commonly get 3-4 pieces of fax spam a week, despite all my phone numbers being on the federal "do not call" list. And the station numbers always lead to disconnected lines.

          • I commonly get 3-4 pieces of fax spam a week, despite all my phone numbers being on the federal "do not call" list. And the station numbers always lead to disconnected lines.

            Then you can start collecting money from them, if you're both in the US. Junk fax is illegal according to 47USC227 [gpo.gov], so look into "(b) Restrictions on use of automated telephone equipment". IANAL, but maybe small claims court is the way to go. If they don't show up, you win by default.

            • Good luck trying to find out who sent the spam though- if the station ID information is spoofed and the 1-800 number denies sending the spam, who do you collect FROM? Who do you SUE?
              • I use a "mark" function (usually *67) to flag the last incoming call as harrassing. Then after a certain number of times you've been faxed, wardialed, etc by the same company, call the cops and tell them to get in touch with the phone-company, which at that point has to submit to the police the list of "flagged" phone numbers. I'd like to see the 1-800 # come up with an explanation/denial at that point...

                My worry is that with VoIP, how easy will it be to spam call without being traced?
                • Very easy- and the standard caller ID system (my phone company doesn't include *67- but others do) is pretty easy to spoof also, all you have to do is program a computer to count the rings and send a burst of static at RE6 voltage down the line at the proper second after the 2nd ring, at which point all the recieving phoneco's computers (and any other caller ID equipment) recieves is "O" for "Out of Area". I know this because I keep a database log of all phone calls recieved at my house.
                  • Interesting. I don't think I've ever gotten an "out of area" listing on my caller id. "unknown #" sometimes, but that usually means it's a private #, or being blocked by the caller. But in those cases the phone company is still recording the originating phone call's # (otherwise how would it be able to connect the call?)

                    I wonder why the phone company has never fixed this? You'd think they'd get in trouble from police departments trying to trace harrasing calls...
                    • Depending on your equipment, there are three possibilities for any given caller ID record data: O, P, and an up to 32 char string. The first is Out of Area, the second is Privacy Mode, the third of course is just the standard caller ID data. How your machine interprets it is up to the manufacturer- Unknown # could be either an O or a P. You can look at it raw with any caller-ID aware modem.

                      O calls are by defintion coming from a different exchange than your local phone company- a different town, say, or
                    • Wow, I didn't realise the phone companies were so far behind when it came to preventing fax spam. Maybe the phone companies need to start collaborating on a sort of IPv6 equivalent...
                      *choke*snort*spewcoffee* sorry, I can't believe I actually suggested that phonecos would collaborate to improve security...

                      Anyway, you'd think the phonecos (the major ones at least) would want to be able to track the phone calls coming in, otherwise who would they bill for those long distance calls? I can't believe the phoneco'
                    • Anyway, you'd think the phonecos (the major ones at least) would want to be able to track the phone calls coming in, otherwise who would they bill for those long distance calls? I can't believe the phoneco's wouldn't care who was using their phone-lines.

                      They simply charge the next phone company up the line- who charges the phone company up the line from them. It's basically a huge game of "not my problem", until you get to the final one- and even then the system that records long distance billing informa
                    • I wonder how hard it would be to build a better phone, one that could "ring" silently, until it got the caller id, and then if the id says "O" or if it's not on your whitelist or if it's on your blacklist (however you choose to set it up), it would just keep ringing silently until the machine at the other end gave up.

                      Same way I deal with spam. My whitelist is now at the point where if it does go into my "junkmail" box, I likely didn't ask for it to be sent to me.
                    • And the phone system is over 100 years old. All the phone company cared about in the early days was one-way- they cared that you accessed the long distance line, they didn't record who the call was TO. Besides, when these mechanical switches were created- long distance operators were still *manual* switches. You'd call the operator and ask her to connect you and she'd write down the connection. No computer records at all- and at best tracing backwards you might find that five or six people on a given e
                    • And the phone system is over 100 years old.

                      IPv6 doesn't have any plans for IP packet origin verification, does it? Just curious because I wonder if phone/fax spam will get worse once VoIP becomes popular.
                    • I wonder how hard it would be to build a better phone, one that could "ring" silently, until it got the caller id, and then if the id says "O" or if it's not on your whitelist or if it's on your blacklist (however you choose to set it up), it would just keep ringing silently until the machine at the other end gave up.

                      These days- not to hard. You can do it with software and a voice modem. My only problem is getting it to *intercept* the call before it gets to other phones in the house or the fax machine.
                    • IP packet orgin is different- if it's TCP protocol anyway instead of UDP. TCP/IP *requires* handshaking, which means at worst you might get a syn flood or an ack flood, but unless all addresses are correct you won't get the syn/act in the correct order to produce a complete connection to let the VOIP call through. IPV4 implements this quite nicely. It doesn't stop all attacks- but it does stop any attack that has information in it. I haven't messed around with IPV6- but I imagine it's similar. Of cours
                    • My only problem is getting it to *intercept* the call before it gets to other phones in the house or the fax machine.

                      I wonder how hard it would be to set up a digital "phone switching" service, just like you get most businesses. Something you could set up where the phone line enteres the house, and then any authorized calls it would pass on to the rest of the house.
                    • Less hard than *expensive*- your cheapest PBX system runs to about $300 and it's not the type of thing you're going to put together out of spare parts. Plus there's the danger of that whole "dial 9 to get out thing"- training kids not to try to transfer calls to extensions begining with 9 (because they don't exist), teaching the whole family that the real number for emergency is 9911, etc.
                    • Too bad it's so expensive. Although I imagine I'd just disable the whole "dial 9 to get out" (my work system doesn't require it), as well as disable call transferring. It would be nice if they came out with a linux app that could do it with 2 modems. I wonder how hard it would be...

                      I've heard of devices that send a certain voltage back down the phone line when the phone rings that somehow confuses modems into thinking they've been disconnected, ever heard of it?
                    • 2 such items- one you can build yourself from any spark-gap cigarate lighter (such as the script electro), and the other is based on the standard tones for a disconnected phone line.
    • So if I delete any email with HTML tags, am I a biased polititian?

      (I don't mind being biased, just interested to know)

      Secondly, since US political parties are, I believe, "licensed to spam", surely that means I delete them on principle anyway (as legalised spammers) regardless of the validity of their viewpoint (if any)

      i.e. if a group are given legal dispensation to spam, I'll do what I can to delete anything they send, simply because they might be spammers.

      (works for terrorists, may as well work for politi
  • What with the modding down of any conservative opinion and what not.

  • What the hell is the big deal? (Score:3, Insightful)

    by pclminion ( 145572 ) on Wednesday August 03, 2005 @02:32PM (#13232679)
    I use Thunderbird's Bayesian spam filter, and guess what? It fucks up sometimes. Usually, it screws up on messages that contain HTML elements (the VAST majority of HTML email I receive is spam, so this is no surprise) or "spammy" words.

    Does political mail look like spam? Oh God, yes. Again, no surprise that a Bayesian style filter might get confused.

    Should we chuck automated email filtering? This problem has always existed. Important messages have a small chance of being miscategorized. If that's not acceptable to you, don't use those filters, or switch to an email provider that doesn't filter your mail that way.

    "Partisan" spam filtering is a farce. What the hell could the ISP possibly gain by surpressing political viewpoints? It's a software fuckup.

    • False-positives are the reason why Thunderbird sends spam to the Junk folder and not directly to the Trash. Yeah, it sucks having to go through that stuff to make sure nothing accidentally got bundled in there, but it's better than a) having it all in the Inbox, and b) losing something important. But the good thing is, it's my choice. If the ISP starts blocking it, I may never know that I lost something that was supposed to get through. At work, that could cost money. Not good.
      • But the good thing is, it's my choice. If the ISP starts blocking it, I may never know that I lost something that was supposed to get through. At work, that could cost money. Not good.

        Very true, but once again, it's merely a failure of the system. There's absolutely no reason to assume that ISPs are purposefully blocking certain political viewpoints. It's ludicrous.

        • Very true, but once again, it's merely a failure of the system. There's absolutely no reason to assume that ISPs are purposefully blocking certain political viewpoints. It's ludicrous.

          I agree. But the second some political mass-mailer or form letter gets dropped or bounced, the assumption will be conspiracy.

          Besides, I wasn't even speaking to political mail. At work, we had a machine running Norton AntiSpam and Thunderbird Junk controls. Norton was overly aggressive, flagging nearly everything as spam

          • I agree. But the second some political mass-mailer or form letter gets dropped or bounced, the assumption will be conspiracy. Well, that should tell you all you need to know about the source of the emails in question. I don't see spam in any form as any different than a guy walking up to you and sticking a flyer in your pocket, and I don't care whether I share in the political sentiment involved or not. In both politics and religion, those who want to hear what you have to say will seek you out, those wh
          • Absolutely. A few years ago, someone posted to the razor-users mailing list that everything MoveOn.org sent out got tagged as spam by Razor. (Vipul's Razor is the original open-source client for what has since become CloudMark SpamNet.) The poster was concerned that Razor was unwittingly being turned into a censorship tool.

            As I recall, it degenerated quickly into a... discussion of whether it was just people putting old addresses on auto-report, or whether the site was actually sending out unsolicited mail
    • Does political mail look like spam? Oh God, yes. Again, no surprise that a Bayesian style filter might get confused.

      Perhaps the sub was unclear. What I'm particularly interested in is ISPs blocking e-mail based on lists of spamvertised URLs.

      It would be pretty trivial for an ISP employee to add arbitrary URLs to such lists. Someone in AOL's anti-spam department could stop AOL users receiving e-mails referencing www.anncoulter.com or www.michaelmoore.com.

      The referenced article implies that this has happened:-
    • What the hell could the ISP possibly gain by surpressing political viewpoints?

      Perhaps you missed this [slashdot.org]?

  • It wasn't that long ago... (Score:3, Interesting)

    by krewemaynard ( 665044 ) <krewemaynard@noSpAm.gmail.com> on Wednesday August 03, 2005 @02:35PM (#13232712)
    ...that folks were clamoring for ISPs to do MORE to prevent spam. Now they're doing more, and we're wondering if it's too much and they should do less. decicive bunch, aren't we?
    • People aren't indecisive. They just have a short attention span. It's like that episode of the Simpsons where a bear gets loose in Springfield, and people storm the Mayor's Office insisting that he make sure it never happens again (never mind that it never happened before). So he spends a lot of money on an Anti-Bear patrol, passes a tax to pay for it -- and they storm his office again, demanding that he do something about high taxes.

      I forget how the episode ends, but it probably wasn't pleasant.

  • ISPs shouldn't block anything at all except by protocol/standard - for example dropping mail not properly sent when using SPF or similar. But blocking based on content, etc. is not proper.
    • ISPs shouldn't block anything at all except by protocol/standard - for example dropping mail not properly sent when using SPF or similar. But blocking based on content, etc. is not proper.

      Oh, so now you're the judge of what is and is not a "proper" service for a company to offer to its customers? I believe Bayesian mail filtering is a perfectly reasonable service to offer to those customers who want it. But hey, according to you that isn't "proper," so I must be a moron.

      If the customer doesn't like it,

    • Content filtering is by far the most effective mechanism. Looking for known spam URLS via SURBL (for example) is Highly accurate. 419 scams generally come from REAL accounts on REAL email servers (mostly the free ones...) Are you saying that those shouldn't be filtered? That's nuts!

  • It's only email that gets filtered like this (Score:3, Interesting)

    by badfish99 ( 826052 ) on Wednesday August 03, 2005 @02:39PM (#13232750)
    Suppose I set up a service described like this:
    All your (snail) mail gets delivered to me. I will read it, and throw away anything that looks like junk, and then send the good stuff on to you. I won't tell you what I've thrown away, or what criteria I use.

    You wouldn't sign up for that, would you? So why do people put up with third-party spam filters that do the same thing? I know that spam is a big problem, but everyone ought to take control over their own communications. It's one thing for me to decide to throw some email away unread; it's quite a different thing to give someone else that power over me.

    • You wouldn't sign up for that, would you? So why do people put up with third-party spam filters that do the same thing?

      You said you won't explain what criteria you use. There are some mail filtering systems where this indeed is the case, but a Bayesian or other statistical system is open for examination. The criteria are probabilities. As somebody who has designed these kinds of filtering systems, I have absolutely no problem handing the decision-making over. I know how it works. And I know how it screws

    • Spam is also different then most snail mail. Snail mail is delivered once a day and most people don't get 1/10th junk mail that they do spam. If you got the amount of junk mail that you do spam and it was delivered even 4 times a day where the post man opened your door and dropped it on the kitchen counter, most people would pay for a service to filter.

      Remember, ISPs are doing spam filtering because customers ask for it. You may not use it but other people will and are.

  • Nothing? (Score:1, Interesting)

    by Anonymous Coward
    Spam filtering is by its very nature censorship. You're agreeing to have your incoming email censored so you don't have to deal with spam. If you don't like the terms of that, well, it's up to you. Set up your own mail server and customize the spam settings how you see fit. Or even find a decent host with SpamAssassin and tweak that to your liking. Or just find an ISP/mail host with no spam filtering, and handle it on your client with Bayesian training.
  • With many people using their ISP's default settings, and manual spam filtering being impractical for many users, what can be done to avoid giving ISPs and anti-spam companies extensive, fully automated censorship abilities?"

    Well, there you go.

    They shouldn't use the default settings.

    The Masses are getting what they want. The least amount of spam possible.

    If some people are really interested in their email then they would be well advised to spend a few minutes looking at the filtering options.

  • Answer (Score:2)

    by 4of12 ( 97621 )

    what can be done to avoid giving ISPs and anti-spam companies extensive, fully automated censorship abilities?

    Making them openly disclose exactly what spam filtering policies they use.

    After all, if people have no idea about what they're not getting, they won't even know.

    For example, I lost out on a wonderful opportunity to help out a government official in Nigeria that my friends are pursuing right now.

  • I don't fully understand how the spam filters work, but I assume it has something to do with a certain percentage of users reporting that an email was spam that caused an ISP to begin filtering it as spam. If its not and the ISP is filtering certain domains because of its politics and not allowing their user base to decide, then this should be reported on and allow users to make up their own minds and allow capitalism to work. I no longer use google because of their filtering of conservative ads. I'd do the

  • Obvious (Score:1, Flamebait)

    by Intron ( 870560 )
    Just put spam filtering in the hands of a large, trusted organization with experience in secure systems such as Diebold.

  • the political lists could help out (Score:4, Insightful)

    by snorklewacker ( 836663 ) on Wednesday August 03, 2005 @03:30PM (#13233352)
    Starting by running some decent lists. These folks run some amazingly slipshod lists, usually unconfirmed optin, and sometimes outright buying addresses. What do they expect? They're spamming!

    But they figure that since they're not commercial, that they have a cause and a message, that they don't really have to pay attention to running a clean list, and anyone who blocks them must be a political censor.

    I'm a left liberal, but I find myself blocking the DNC more often than the RNC, simply because the former has less stringent mail practices.
  • Unsolicited e-mail from someone not on my address book has a huge likelihood of being spam.

    The fact that a bulk-mailing to a bunch of people happened to be political in nature doesn't change the fact that those messages smell shockingly like spam.
  • Come on, anytime you hire someone else to handle your responsibilities, they are going to screw up sometimes.

    You could hire someone to point out all of the interesting TV shows for you. They will miss some good ones and they will give you some that are boring.

    You could hire someone to push the FF button on the remote to skip the commercials on the DVR. Will they skip over the 1 in 100 that is actually advertising something interesting? Yes.

    People do hire others to teach their kids and listen to the noise ab

  • If your not on my list your probably SPAM. (Score:3, Informative)

    by Shivetya ( 243324 ) on Wednesday August 03, 2005 @05:32PM (#13234994) Homepage Journal
    The spam system I use is the challenge/response type. You get told you can ask for access to my mailbox and I can decide to allow it.

    As for other automated systems like Postini, which we use at work it all comes down to the content. A lot of time configuration errors will creep in as well like the time all mail a guy named Dick H... was being deleted without notice! You can guess why.

    PostIni has actually blocked some political mail to me and I can give you a few hints.

    Certain political emails contain verbage that gets labeled as porn or hate speech. That second category is so broad and abused that companies will throw about anything that might hint of insensitivity under it just to keep themseleves safe.

    The second category they got caught by? Money. There is a $$$ category that attempts to weed out money requests and the like.

    Do I care one way or another? No, I particulary did not like how the government excluded the rules of spam from acting on elected officials and I really couldn't care less about any unsolicited spam from either aisle.

    • "The spam system I use is the challenge/response type"

      Good luck buying anything from a shop which emails you once to say "can we check your address?"

      I was looking at the greylist system, until I realised how many false-positives it would generate. Even today, when I auto-delete about 99.8% of my email, I still have to fish-out internet shops and add them to the relevant whitelists...

      Maybe I should allocate one To: address for all the shops, rather than a different one for each shop (which tells me who leak
  • I seem to recall reading stories that most cyber-nanny type products block access to liberal sites like NOW and PETA and bloack access to democratic candidates, but never block traffic to sites like the NRA...

    Can't find an original story right now, but http://www.washingtonfreepress.org/46/urban_work.h tml [washingtonfreepress.org]

  • Tag spams, don't drop them (Score:3, Interesting)

    by Motherfucking Shit ( 636021 ) on Wednesday August 03, 2005 @11:55PM (#13237438) Journal
    If you're worried that you're missing emails of any sort (politics is a red herring as far as I'm concerned), go with an email provider that tags messages instead of deleting them. This is a simple configuration in SpamAssassin, and any competent mailhost will set it up on a user-configurable basis. I'm not saying that ISPs will be so kind as to offer this option, but if you can afford it, you've probably already got your own domain at a webhost somewhere.

    The webhosting company I use for my personal stuff does this (not going to mention names, because I don't want this to sound like a plug). I can set up unlimited POPs and forwarders, and adjust the SpamAssassin settings on each one; not only the scoring gestalts, but also whether messages that are considered spam get deleted, or simply have a [SPAM] tag added in the subject line.

    I opt for the tagging. This means that I get all of the email, but with a single filter rule on my mail client (i.e. matching "[SPAM]" in the subject header), I can sort all the questionable stuff to a "Junk" folder and go through it later.

    If your ISP drops or deletes spam mail, set your mail up elsewhere, or bitch heavily to your ISP requesting that your spam be tagged instead of nuked. I hate spam as much as the next guy, but there's no reason that any ISP should be risking false positives and having their customers lose legit email.
  • Now if I can just convince the USPS to filter out the ads that clog my mail box.
  • If the question is:
    what can be done to avoid giving ISPs and anti-spam companies extensive, fully automated censorship abilities?

    Well.. I would have to say one of the services TFA complains about is actually an example of an acceptable way for an ISP to have balance. Postini does not 'delete' any spam, it only quarantines it. And not only that, but when you release a quarantined message, it asks if you wish to make the sender 'approved'. Not only that, but it correctly handles mailings lists because it

Slashdot Top Deals

E = MC ** 2 +- 3db

Close