* or anything. The *s are ignored altogether. Try it (both ways).
This is ~another~ of Google's maddening limitations. A sales rep told me wild card's not there because they have found that it doesn't increase search effectiveness by double -- whatever that means -- and that's their standard!
Sheesh! Tawkabout monopolies!
That's how it's used in practice. I don't know what the original idea was for differentiation, but it clearly hasn't worked out. The only hitch is MS's claim as a software brand. But they're already facing that, and it's clearly unreasonable for them to expect anything special for the tld.
So, sounds like a slam-dunk. If it goes differently, we'll know the fix is in.
Any filter based strictly on message content is all but useless in the long run. Why? Three reasons: false positives, false positives and false positives!
If there is a reasonable chance of losing even one real message, then I have to comb the filtered messages anyway, no matter how they got segregated. So absolutely nothing is gained in the end.
For an example in the extreme, what if a good friend forwards you a particularly juicy piece of spam with a commentary to make some point? Bam! Any content-based filter will rate it high and filter it. Message lost... unless you comb the rejects anyway.
Is there a filtering method with no reasonable chance of false positives? Yes, actually, the bait account, distribution-based, signature filtering represented well by BrightMail (I have no affiliation with BrightMail). That approach actually uses the very definition of spam, namely ~unsolicited~ mail sent by strangers to large numbers of recipients, plus blacklisting.
BrightMail claims false positives are 1 in 100,000, but it's probably even smaller than that. Even 1/100,000 is small enough that I don't feel a need to scan the filtered messages for false positives.
And, if you are unhappy with the less-than-100% filtering of something like BrightMasil, then you can apply other methods as well. At least you'll have less purported junk to scan for false positives.
Actually, the best combo would be an automatic whitelist acceptance (anybody you've ever mailed to or accepted mail from) followed by BrightMail (or equivalent) followed by a good content-based filter. Why nobody's done this yet is beyond me!
I sure wish there were a consumer version of BrightMail.
Slashdot Mirror
* or anything. The *s are ignored altogether. Try it (both ways). This is ~another~ of Google's maddening limitations. A sales rep told me wild card's not there because they have found that it doesn't increase search effectiveness by double -- whatever that means -- and that's their standard! Sheesh! Tawkabout monopolies!
So, sounds like a slam-dunk. If it goes differently, we'll know the fix is in.
Any filter based strictly on message content is all but useless in the long run. Why? Three reasons: false positives, false positives and false positives!
If there is a reasonable chance of losing even one real message, then I have to comb the filtered messages anyway, no matter how they got segregated. So absolutely nothing is gained in the end.
For an example in the extreme, what if a good friend forwards you a particularly juicy piece of spam with a commentary to make some point? Bam! Any content-based filter will rate it high and filter it. Message lost... unless you comb the rejects anyway.
Is there a filtering method with no reasonable chance of false positives? Yes, actually, the bait account, distribution-based, signature filtering represented well by BrightMail (I have no affiliation with BrightMail). That approach actually uses the very definition of spam, namely ~unsolicited~ mail sent by strangers to large numbers of recipients, plus blacklisting.
BrightMail claims false positives are 1 in 100,000, but it's probably even smaller than that. Even 1/100,000 is small enough that I don't feel a need to scan the filtered messages for false positives.
And, if you are unhappy with the less-than-100% filtering of something like BrightMasil, then you can apply other methods as well. At least you'll have less purported junk to scan for false positives.
Actually, the best combo would be an automatic whitelist acceptance (anybody you've ever mailed to or accepted mail from) followed by BrightMail (or equivalent) followed by a good content-based filter. Why nobody's done this yet is beyond me!
I sure wish there were a consumer version of BrightMail.
--David