Need free/trial anti-spam product with guaranteed ZERO false positives ASAP...

[CZroe]

I have 1,700 unread messages. I don't know why, but I get at somewhere around 25-50 spam messages an hour. There are probably three legitimate messages. I will not use most anti-spam products because, as you can guess, any false positives will only make the ratio of emails I have to search through EVEN MORE daunting 🙁 I usually don't even try until I'm expecting something and I have 3,000+ unread messages. Therefore, I frequently find myself searching for a needle in a hay-stack and nearly always it's found after reviewing my deleted emails (Meaning I have to search the haystack TWICE).

I know the tricks. I disable the preview pane to avoid unique "phone-home" images in HTML emails. I only wish Outlook had a "flag" for HTML email, or a way to view source before opening the message. There also appears to be no way tosort all HTML email to a seperate folder or disable links to external images/content. It's like MS WANTS spammers to do this!

I usually arange by name, and delete the spammers that send hoards of messages from the same address or the ones with corrupted-looking names (Most of it comes from China with Chinese fonts and the screwed-up characters seem to arrange together). Then I arrange by subject and delete the exorbant duplicants and garbage corrupted subjects there.
I then dupble-check the deleted messages and delete the sorted ranges as I write them off. It's tedious, but with this method eBay and PayPal messages sometimes get deleted and MUST be recovered (If you spend just a second longer to read the duplicate subjects, you increase the time it takes EXPONENTIALLY). This method still leaves me with 1,700 emails this time, and no real shortcuts.

I do not immediately delete any of the hundreds or "hey" "hello" "what's up" or blank-subject messages because too many times they turn out to be from someone important (People should know that it makes their messages look like spam. Stupid.). I lost $280 by ignoring a message from the sender "sales" and no subject line (Company went out of business and I never got the product or responded to get the refund). If I didn't ignore it, I would have spent DAYS searching through those messages anyway 🙁 Am I the only one with this problem?

After this, I usually unplug my broadband so I can be sure none of the spam I inevitably open will be able to phone-home. Then it's one by one for the rest of the day...

Today I don't have the time. I need something that can delete ONLY the KNOWN SPAM. With zero false-positives, I only expect ~40 but that would SIGNIFICANTLY reduce the workload. Perhaps some sort of fingerprint created by users flagging confirmed spam. I still worry that some people will flag things important to me that some may consider spam, like Nintendo or eBay newsletters and updates, but I'll deal with that.

Because I must do it today, I need something with a trial version.

Thnx!

 

[Need4Speed]

zero false positives is impossible. i use spamassassin, and after months of training and thousands of emails, i'd say i probably have 98% success at weeding out all spam

 

[igowerf]

Just get any spam filter and set it to less aggressive settings. That should filter the 1700 down a lot. Also, I use 2 email addresses. One for friends/family and online deals (FS/FT, ebay, etc) and another for when I sign up for things.

 

[ClueLis]

Try using Mozilla or its stand-alone twin, Thunderbird. Both have built-in junk controls. I can't garantee false positives, since the whole thing is based on teaching it, but it will not delete your email. It simply marks or hides junk (based on your request). I have found that it works flawlessly for me, but I don't get a whole lotta spam.

There is no way, unfortunately, to garantee no false positives in any anti-spam system that is at all effective.

 

[CZroe]

You people don't get it! 😉

Hearing "98% success" is like hearing "Makes the problem 98 TIMES worse"

The PROBLEM involves IMPORTANT messages (I RARELY receive personal email, nearly all legitimate messages are VERY important and nearly all are auto-generated and frequently misidentified)

At the MOST, I get about 13 legitimate messages per month (Expected RMA requests, support inquiries, eBay / PayPal receipts, etc) in addition to at LEAST 3,200 spam messages. Usually, the ratio is about 3 legitimate messages to 3,500 spam messages. Lets say, 1 legitimate message gets mixed in with the 3,500 filtered spam messages. IT HAS NOT SAVED ME ONE BIT OF WORK. I still have to go through every single message! What I want to do is REDUCE the number of messages I have to go through, NOT eliminate. I'd GLADLY go through a haystack of 2,000 spam message instead of 3,500.

As for zero false-positives being impossible: YOU ARE STILL THINKING ABOUT GENERIC FILTERS! Not filtering messages based on a database of CONFIRMED spam.

From http://www.ultrasoftware.net/spam/outlook%20spam%20filter.html regarding one of the methods used by their SPAM-OFF product:

Spam Signatures
are the most effective, zero- false positive and quickest method of detecting spam.
We use numerous emails on geographically distanced systems and actually promote the emails so that they could be picked up by new spammers. So by the time you open your e-mail box, Spam OFF starts updating and most probably it will filter out all those spam messages that appeared this day.
We use numerous emails on geographically distanced systems and actually promote the emails so that they could be picked up by new spammers. So by the time you open your e-mail box, Spam OFF starts updating and most probably it will filter out all those spam messages that appeared this day.
The only drawback Spam database has is that it requires a large amount of traffic and processing power on vendor's end. But you pay us for that, right?

I tried SPAM-OFF. You can not disable all other filters and exclusively use the signature / fingerprint method. Also, it appears to be seriously flawed. Many of my "Notification of Instant Purchase Payment Received" messages were identified as spam through this method (It modifies the header and tells you this). I can't see why, as the subject and return sender are formatted uniquely by PayPal (The user name is in the subject, the return address is the PayPal user's instead of PayPal's). This means that the "database" was not of confirmed spam! Their "method" described looks like it was DESIGNED to block important auto-generated emails!

Looks like I need to make a P2P email client somehow where each user reports spam signatures.

 

[PrincessGuard]

For the love of god, just get a new e-mail address, especially since all your important mails are automated. In the time you spent bitching here, you could have setup your own domain and changed your eBay/PayPal e-mail many times over.

No anti-spam system is perfect. Even ones that catalog existing spam can't possibly catch every variation of every piece of spam out there.

 

[Nothinman]

You people don't get it!

Hearing "98% success" is like hearing "Makes the problem 98 TIMES worse"

You're the one not getting it. As long as the mail filter only deletes mail it knows 100% is spam how can the problem be worse? Worst case it takes care of 98% of your spam and you have a few dozen messages to delete instead of a few thousand, what's the problem? You won't find your perfect spam solution, if it was available don't you think a lot of people would be raving about it here already?

 

[Sunner]

I'd just get a new mailaddy and never publish the address, then use some other address for stuff like forums, mailinglists, etc.

 

[oog]

I don't recall the name of it, but our company runs a spam filter on the email server that gives a score to each email to say how likely it is spam. The score is on a 10 asterisk scale, and I can write rules to search for any number of asterisks. The more asterisks, the more likely it is spam.

Personally, I use a two-level spam filter. The less aggressive spam identification is the one that I'm sure never has false-positives. I delete those automatically. The more aggressive spam filter is left with only 5-10% of all of the spam messages, so finding the false positives in there is a lot easier. It is not like finding two out of 100 messages -- it's more like finding 1 in 10, which is easier to identify. All told, I don't need to even look at the heading of most of the spam, which suits me fine.

I think a similar two-level filter will work for you as well, but I agree with some others here that it may just be better to get a new address and protect it from spam in the first place.

 

[manly]

Originally posted by: Nothinman

You people don't get it!

Hearing "98% success" is like hearing "Makes the problem 98 TIMES worse"

You're the one not getting it. As long as the mail filter only deletes mail it knows 100% is spam how can the problem be worse? Worst case it takes care of 98% of your spam and you have a few dozen messages to delete instead of a few thousand, what's the problem? You won't find your perfect spam solution, if it was available don't you think a lot of people would be raving about it here already?

Your explanation is rather confusing. If the spam filter has 0% false positives, then yeah you only have to manually filter any remaining messages in the Inbox (which could still be substantial, but still largely reduced).

If the spam filter has 2% false positives, with a substantial volume of mail, you still have the problem of searching through the "junked" messages just to locate just those false positives. You also have to search through the Inbox (as in the first case). So the worst case is hardly better than no spam filter. It's not 98 times worse, which is ridiculous.

I don't think Bayesian filtering, such as in Mozilla or MSN8, can bring the false positives down to zero. At the same time, I do agree with everybody else that the problem is being approached from the wrong angle. The suggestions to switch email address and mail client seem apropos.

The first part of any filtering system should be a whitelist. Sorting away known good senders (or sometimes, recipients) would greatly reduce the possibilities for false positives. Now if you conduct commerce with strangers (such as regular eBay selling), then it's likely no system would be near perfect. We all pretty much agree on the last part.

 

[Need4Speed]

a good white list is a great start. also, banning any know spamming domains will take a huge chunk out of spam as well, either through manual updating of the access files or using someting like realtime DNS blocking.

Filtering at the client end is much harder than at the server end IMHO.

Furthermore, there isnt a single spam package on the planet that can filter 100% of the spam. There are always going to false postives and false negatives. The only way is to sort it by hand. The best one can do, is to reduce the spam to where its manageable to sort manually using any of the methods above.

Furthermore, the best way to avoid spam is not to use youre email addy online for everything. dont post it on forums, dont use it to apply for stuff or download proggies.

I generally have one account on my mailserver that I use for online stuff, and then I rotate the address for that every 7 days.

I too used to get hundreds of spam messages on a daily basis until I started implemennting Realtime DNS blocking in Sendmail/Postfix and using spamassassin, and using two accounts. I'm now down to one or two spams a day that get through my filters.

Of course this doesnt happen over night. It took months to setup a useable database.

If you cant filter at the server end, its very easy to setup a nix box with fetchmail on it to grab all your mail from other pop accounts and do the filtering before it gets to outlook or whatever mail client you might be using.

 

[Cadaver]

I use McAfee's SpamKiller. After about one week's worth of tweeking, I've not had it give me a false positive since. Some spam still makes it through, but when it does I add it to the spam filter's database.
The program isn't an "intelligent" learning spam seeker per se, instead it's basically a huge database of filters. You can add/remove filters based on subject text, sender's email address, sender's domain name, country of origin, and messgae text. You can also add a "friend" filter to allow emails from people you know, regardless of the message or subject text (in case you have a friend or client who likes to talk about farm girlz humping super huge c0cks and such).
McAfee also uploads updated filter sets as well automatically if you wish (often things like filtering on message text like "cum girlz", etc).

It's worth a try.

However, I think there is NOTHING that will give you 100% accuracy. If there was, spam issues would not be up for congressional hearings.

 

[Nothinman]

a good white list is a great start.

This is probably the best solution. If you only have a handfull of addresses that send you mail put a rule to allow them in and one big rule to delete everything else.

 

[CZroe]

You're the one not getting it. As long as the mail filter only deletes mail it knows 100% is spam how can the problem be worse? Worst case it takes care of 98% of your spam and you have a few dozen messages to delete instead of a few thousand, what's the problem? You won't find your perfect spam solution, if it was available don't you think a lot of people would be raving about it here already?

No sir. I was responding to someone else's comment stating that zero false-positives is impossible, but 98% of spam can be caught. He DEFINATELY was not saying "98%" (His qupte) of the mail the filter "knows 100% is spam" (Your quote) so it looks like you got confused 🙂 I'm NOT looking for something that identifies even close to 98% of spam. Even 5-10% is OK with me as long as it is guaranteed to have zero false-positives. 5% *will* save me the trouble of digging through several hundred messages.

If the spam filter has 2% false positives, with a substantial volume of mail, you still have the problem of searching through the "junked" messages just to locate just those false positives. You also have to search through the Inbox (as in the first case). So the worst case is hardly better than no spam filter. It's not 98 times worse, which is ridiculous.

Exactly right. It does multiply the effort however. Searching through the previous ratio was so difficult that nearly half of the legitimate emails were deleted and only found the second time through. This INCREASES the slap-happy chance of missing it and because the spam will already be in the trash, it will have no room for error. Searching for 1 false-positive in 3,000 is much harder than searching for 5 in unfiltered 3,011 messages.

Furthermore, there isnt a single spam package on the planet that can filter 100% of the spam. There are always going to false postives and false negatives. The only way is to sort it by hand. The best one can do, is to reduce the spam to where its manageable to sort manually using any of the methods above.

A "100%" filter is exactly what I'm not looking for. I'm talking about only filtering the spam messages which are guaranteed to be spam. Even if it only catches 5%, that's hundreds less emails for me to go through. The "sorter by hand" method is EXACTLY what I'm talking about. If every message I sorted through and every message everyone else sorted through were added to a real-time database, they could be deleted by RECOGNIZED signatures. The fingerprinting Spam-OFF uses uses "bait" email accounts and registers it as spam when several of their servers receive the same message. Makes you wonder how they blocked a PayPal email with that method if they weren't using the addresses for online transactions

I use McAfee's SpamKiller. After about one week's worth of tweeking, I've not had it give me a false positive since. Some spam still makes it through, but when it does I add it to the spam filter's database.
The program isn't an "intelligent" learning spam seeker per se, instead it's basically a huge database of filters. You can add/remove filters based on subject text, sender's email address, sender's domain name, country of origin, and messgae text. You can also add a "friend" filter to allow emails from people you know, regardless of the message or subject text (in case you have a friend or client who likes to talk about farm girlz humping super huge c0cks and such).
McAfee also uploads updated filter sets as well automatically if you wish (often things like filtering on message text like "cum girlz", etc).

The filter method is the most notorious. Almost all legitimate automated mailings will be blocked for including words like "Click here to activate" (Forum sign-up) or "Click here to view" (eBay receipt). Grrr. I missed a Western Digital RMA because of this... Also, all of my Nintendo mailings were classified as spam for promoting each games external web site ("Click here"). It doesn't work except for those who don't use email for serious things. I believe the term is "chattyb*tches" in the FPS community 😉


I think there is a patentable idea here, but I'll go ahead and say it:

An email client. Outlook competitor. Flags HTML email and email with external content. Can allow you to open or check them without images or other HTML content loading. Can view source without opening the message (Outlook Express allows you to, but not the full application). Can filter messages based on fonts and language used (If I can't understand them, I can delete them. This has eliminated 15% of spam successfully for me using external applications. My one success! 😛). Messages CONFIRMED by users to be spam will be uploaded to a P2P network. All clients will share the spam signatures. Recent and common spam will be promoted in importance. Spam that is incorrectly identified by users as spam will depreciate the users influence when other users correct it. Spam signatures can be created on any part of a message instead of messages as a whole (Most spam is unique to phone home and confirm your address or mention you in the subject). It will use the method for identifying spam EXCLUSIVELY. Other rules can be made for sorting messages (Dumping all the Russian and Chinese language email in the trash for example, but not correctly or incorrectly identifying them as spam).

All of the concepts exist seperately. Of course, I suck and can't put it together. 🙁