Slashdot Mirror
How Good is Gmail's Spam Filter?
juglugs asks: "I've been using Gmail since the first round of invites on Blogger. Tonight I received my very first spam email. It was one of the ones offering me some product (I didn't read it too much) that would increase my manhood. It didn't trouble me too much as I just had to hit the 'Report Spam' button and off it went. But how good is their spam filter? Does anyone else get much spam? Why didn't this get recognized as spam - it had all the usual 'keywords' that you'd normally associate with it."
Not Ask Slashdot.
How Good is Gmail's Spam Filter?
Who knows?
Will it get better? Will it get worse?
Who knows?
Part of the convenience of using GMail -- or any other email service -- is that that service filters spam for you.
But that means you don't get to do your own filtering.
Which is why I don't rely on my ISP to filer, and indeed, asked them explicitly to not filter my mail when they began to do so.
I'm too paranoid about false positives causing me to miss an important email (eventually all those girls who dumped me will wise up and beg my forgiveness, right? Right?), and I figure I can do a better filtering job on the client side. And indeed, I can even use a chain of multiple filters, or roll my own filter.
Currently I'm using SpamBayes, and it works well enough. Could it be better? Sure, it does miss several spams a day. But, it also filters many more than it misses, I'm not worried about false positives, and I can always hack the source if I need to (already did so to work around some MS Outlook stupidity, in fact).
Opinions on the Twiddler2 hand-held keyboard?
Friend who sent me an invite reported a 100% drop in spam from 50 a day to 0.
I have a hotmail account for spam and it goes through just fine thank you.
Gmail is still in it's beta form. The company is still working out all the kinks in it. That's why you have to go through the process of getting invited before you can set up an account. Spam is going to fall through the cracks for a while until they finish fine-tuning the filter.
As long as it still says beta at the top of the screen when you log in, you should expect things like that. Click report spam and do your part to help the filter get finished.
"Well kids, you tried your best, and you failed. The lesson is, never try." -Homer Simpson
Perhaps when you signed up you gave something away and the Gmail filter is that smart that it knew to consider penis enlargement emails as legitimate?
In a word, it's terrible.
I forward all my email from my existing account (which I've been using for 5 years and gets a ton of spam) to my gmail account and spam always slips through. I've been using gmail since shortly after it was announced and I have seen it improve, but I'm still getting the same format spam slip through every day.
I pick up the same mail in Mac OS X Mail, and the combination of POPfile and the Mail spam filter gets it all.
Of course, the only email I've gotten is the welcoming message. I suppose I should sign up for a few mailing lists. Easier than getting actual friends.
Would you rather have an ISP do group heuristics, potentially marking email you want as spam, or individual heuristics, forcing you to identify email spam yourself?
It's a trade-off... on the one hand you get much quicker and more compehensive spam detection by using a group level rule but then you have to check your spam folder to see if it incorrectly marked good email as spam and on the other hand you have individual rules which must be generated for each account based on individual opinions.
Neither is perfect.
Bottom line is that you're using a free service, if you don't like it you can move on w/o expense incurred.
A fool throws a stone into a well and a thousand sages can not remove it.
That's sick dude.
"He who would learn astronomy, and other recondite arts, let him go elsewhere. " -- John Calvin, commenting on Genesis 1
Well give me your e-mail address and i'll send you an invite.
Well, actually - it's HORRIBLE.
So far, no spam whatsoever has found its way into my inbox. However, the amount of false positives filtered into the spam folder is overwhelming.
For a while I wondered why I only got reports by email about 30-40% of my finished online auctions (link omitted, no free advertising here). Last week I accidentaly clicked on the spam folder, and there it was, dozens of FALSE POSITIVES.
And yeah, there is NO INDICATION AT ALL of mail in the spam folder, one have to explicitly look in it to see if there are any e-mails there...
Sure, mod me troll if you like. I've been using gmail since the first blogger.com-invitations, and am very happy with it (and have more invites to give out than people to give them to. I tried gmailswap for a while but soon got bored).
Still, far to many false positives. I have no idea why some auction-results were treated as spam, and others not. They're almost identical. Or perhaps it was exactly that which caused the problem, several near identical mails in a short period of time...?
I get over 800 spam a day to my domain, which I now have forwarded to my gmail account.
The gmail spam filter is knocking out about 80% of it. I haven't bothered to check for false positives as it's enough of a hassle getting rid of the approximately 160 spams that get through to my inbox.
Gmail has only caught about 5 of the 30 spams I've gotten so far. Although I haven't seen any false positives yet.
It was a pub based on a previous search query.
After reading this thread I look at my spam folder in my gmail account. Argh !! What a surprise 6 conversations from different developer mailing list where marked as SPAM. In this particular case their filter did a really poor job : 100% of the mail was valid and should not have been considered as spam.
What a disapointment !!!!!
I received about 100 e-mails in my gmail account, and almost all mails were marked as spam. And I didn`t received a single spam in this account yet.
I'm surprised at the range of people saying its good to the people saying its terrible. I was going to say its pretty darn excellent. I use a catch-all address for my domain which I forward to wherever. Filtering on X-Forward I use to be able to catch most of the spam. However, some were so badly formatted that the headers weren't processed properly.
When I switched to using GMail I didn't realize it at first. I was still getting 2-3 spams a day (which is was what I use to get.) Then I realized I never set up any filters or anything! On the 2nd day using it I opened the Spam "folder" and found over 80.
I've checked the folder every so often and haven't found any false positives.
Personally, I'm impressed with GMail's spam filter capability.
just fucking google it? ;)
It is horrible. I've had my email account since 1995 and get a lot of spam. Gmail catches the minority of it.
Another problem is that gmail rejects incoming email virii when receiving over smtp. Normally great (who wants virus mail) but when you are forwarding to gmail from another account, it causes a bounce back to your forwarding account, then your forwarding account tries to forward again, and.... you get an email loop between smtp servers. I had one piece of email with megabytes of received headers before I realized what was going on. Now I fetch locally, filter using clamav, then forward to gmail. I'm considering also filtering for spam to get a result that is closer to spam free.
I signed my gmail account up for some mailing lists. The other day, I had 12 messages marked as spam. 8 of them were legitimate messages. That gives a 66% false positive rate. There have also been a bunch of false negatives (but I don't have numbers for that). So, gmail's spam filters need a lot of work.
A bunch of stuff is also filtered at the SMTP level. Anything with an executable attachment is dropped before you ever see it.. this is not so good for some of the security mailing lists like full-disclosure and others discussing trojans and worms.
-molo
Using your sig line to advertise for friends is lame.
i forwarded gmail -every- email that i received for several weeks. they caught ~50% of the spam. spamassassin catches 90-95%. how they can be so bad when they have 10,000 machines on which they could run spamassassin on all incoming mail is sad.
in my header :) (don't forget to unspam)
I'M NOT ANGRY!
Fuck you Comcast stop selling email addresses
-- I was raised on the command line, bitch
Now, about that invite...
If there's anyone that I think can create some software to intelligently determine if an email is spam or not, my first guess would be google. They seem to have a good track record of being able to sort the mess we call the Internet. Just give them some time. It's still beta after all.
Therefore, the amount of active users should affect the precision of the filtering, since it's the users who report the spam. Since Gmail has relatively few users right now, the filtering might be relatively poor in comparison to what it will become. At least that's what my sense of logic tells me now.
For the record... Currently, I have 0 ham reported as spam, 7 spam reported as spam, and 1 spam reported as ham. Not many know of my Gmail address at all yet.
Beware: In C++, your friends can see your privates!
Google's spam filtering better be good. Do a search on google with site:google.com inurl:gmail +"a-". Guess what you get? That's right, a huge list of gmail invites for easy spam harvesting. Way to go google.
It seems good. One question that's been on my mind since I started using Gmail is: "Where are the relevant text ads I've been hearing about?"
Since I got Gmail I've been emailing my programmer friends about projects like Quake 2, Blender 3D, GTK Radiant and the like... one email we were even joking about Viagra and Penis Enlargement pills setting off spam alerts. Those keywords, I would think, would have sparked a shitload of text ads. None, though. I have yet to see a text based ad in my Gmail portal.
Not even in my signature is there a "This Email Brought To You By The Friendly People At Google."
I have several e-mail accounts that receive roughly 2,500 messages a day, each. I could forward them to GMail and tell you -- if you'd invite me =)
mark (at) signal42 (dot) com
TIA.
Comment removed based on user account deletion
I will second what Richard5mith said. I also forward a copy of my mail to GMail to test the spam filtering and labeling features.
What I have found is that the spam filtering is less effective than Outlook Spam Filter, often letting more 30-50% of spam through to my inbox.
I must be the only geek without a gmail account. Hook me up? stryc9_AT_shaw.ca
www.madeofwinandawesome.com
I forwarded my email frrom my usual forwarder to gmail. I have gotten two false positives from commercial email that I subscribe to, and several spams have slipped into my inbox. All the missed spams are very similar, having a random dollar amount in the subject line, offering to refinance my house, then wrapping up with a few lines of madlib-like text.
I have to say, some of these madlib-like phrases are quite amusing. "Japanese requester non-proprietary doesn't" anyone?
I also got a casino offer signed:
Best regards,
Jamie Zawinsky
Oh the irony.
Well, I have over 16,000 (yes, you read that right) spam messages in my spam folder. Why so many? Because I can't 'empty spam folder'. I've used about 70M of space on spam alone. The filter is *okay* compared to how much spam is being received each day but there are about 25-70 spams that get through on average and I receive about 500+ spams a day. It isn't great but hopefully it will get better.
BTW, I redirect all unrouted mail from my domain to my gmail account, which is where I get all the spam from...
Large print giveth, and the small print taketh away