> It doesn't need to do 100% filtering - it certainly does well enough,
> though.  But you've decided it's "intractably difficult" so it's not
> worth bothering with, even though it's very popular and lots of people
> have great success with it.
>
> Which planet are you on again? ;)

Ah yes, the argumentum ad populum. You realise that *millions* of people 
use Internet Explorer, right? So that *must* mean it's the best web 
browser. :-P

>> In other news, I have yet to see a spam filter which actually filters
>> out spam and nothing else. (Then again, I haven't searched extensively
>> for one...)
>
> None of them make that claim.

And now I must ask which planet *you* are on. This is /the definition/ 
of what a spam filter is. And, just like moisturising creams that are 
"guaranteed" to make you look 20 years younger, people make all sorts of 
completely unsubstantiated claims about spam filters.

> But by and large, spam filters do a pretty decent job.

By this I can only assume that you mean "the spam filters /which I have 
used/ do a pretty decent job". Because the spam filters that *I* have 
used all do a pitifully awful job.

For example, my ISP filters all my incoming email. After 5 years, it has 
/still/ failed to realise that anything containing the words "free meds" 
or "PayPal" is spam. And it has also failed to realise that anything 
containing the word "Haskell" is ham. This, after my repeatedly marking 
the spam and ham for its "learning algorithm". Unfortunately, this 
program's idea of "learning" is to blacklist or whitelist each 
individual sender, each time I press a button. How pathetic.

My employer used a system called "barracuda" I believe, and that was 
similarly hopeless. In fairness, the genuine email we receive has 
several fairly unusual characteristics. Even so, people really shouldn't 
be receiving emails with viral attachments, or links to exploit sites, 
or those fake "delivery failure notification" emails, or...

Post a reply to this message

On Mon, 11 Jul 2011 10:29:49 +0100, Invisible wrote:

>> It doesn't need to do 100% filtering - it certainly does well enough,
>> though.  But you've decided it's "intractably difficult" so it's not
>> worth bothering with, even though it's very popular and lots of people
>> have great success with it.
>>
>> Which planet are you on again? ;)
> 
> Ah yes, the argumentum ad populum. You realise that *millions* of people
> use Internet Explorer, right? So that *must* mean it's the best web
> browser. :-P

False equivalency, my friend.  IE is the default browser installed in 
Windows.  Many people use that by default.

Firefox does not ship with AdBlock Pro installed.  People download it, 
install it, and use it successfully.  I happen to be one of those people.

You, instead, have assumed "it can't possibly work" and so haven't even 
tried it.

Now, given the empirical evidence of having used it and having it work or 
the *assumption* that it isn't going to work so why bother....I think the 
empirical evidence is going to work.

>>> In other news, I have yet to see a spam filter which actually filters
>>> out spam and nothing else. (Then again, I haven't searched extensively
>>> for one...)
>>
>> None of them make that claim.
> 
> And now I must ask which planet *you* are on. This is /the definition/
> of what a spam filter is. And, just like moisturising creams that are
> "guaranteed" to make you look 20 years younger, people make all sorts of
> completely unsubstantiated claims about spam filters.

I'm on Earth.

Spam filters work to a significant extent.  Again, empirical evidence.  
100% isn't necessary to declare success or failure.

You're declaring failure because 1 message in "n" (for sufficiently large 
values of 'n') gets through.  Fact is, spam filters aren't supposed to 
block all spam.  They're supposed to reduce it, and for most people, they 
do the job properly.

>> But by and large, spam filters do a pretty decent job.
> 
> By this I can only assume that you mean "the spam filters /which I have
> used/ do a pretty decent job". Because the spam filters that *I* have
> used all do a pitifully awful job.

Fair enough point.

> For example, my ISP filters all my incoming email. After 5 years, it has
> /still/ failed to realise that anything containing the words "free meds"
> or "PayPal" is spam. And it has also failed to realise that anything
> containing the word "Haskell" is ham. This, after my repeatedly marking
> the spam and ham for its "learning algorithm". Unfortunately, this
> program's idea of "learning" is to blacklist or whitelist each
> individual sender, each time I press a button. How pathetic.
> 
> My employer used a system called "barracuda" I believe, and that was
> similarly hopeless. In fairness, the genuine email we receive has
> several fairly unusual characteristics. Even so, people really shouldn't
> be receiving emails with viral attachments, or links to exploit sites,
> or those fake "delivery failure notification" emails, or...

Have you tried something like Spamassassin?

And yes, you do work in an industry where the normal triggers for spam 
might actually be legitimate business communications - as you work in the 
pharma industry IIRC.

That doesn't mean spam filters are useless.

You'll note that I didn't say *all* spam filters were wonderful and 
flawless.  I said that spam filtering generally works well.

Jim

Post a reply to this message

> Firefox does not ship with AdBlock Pro installed.  People download it,
> install it, and use it successfully.  I happen to be one of those people.
>
> You, instead, have assumed "it can't possibly work" and so haven't even
> tried it.

Before we get too far into this, I would like to re-emphasize what I 
actually said:

I said I had never bothered trying such software under the /assumption/ 
that it won't work properly. I said that it looks intractably difficult. 
Then again, so does writing non-trivial software using something as 
primitive as C, and apparently people manage to do that.

In short, I would be surprised if ad-blocking technology can work. Not 
astonished, but definitely surprised.

> Spam filters work to a significant extent.  Again, empirical evidence.
> 100% isn't necessary to declare success or failure.
>
> You're declaring failure because 1 message in "n" (for sufficiently large
> values of 'n') gets through.  Fact is, spam filters aren't supposed to
> block all spam.  They're supposed to reduce it, and for most people, they
> do the job properly.

In my book, if I use a spam filter and still receive unacceptable 
quantities of spam, or have unacceptable quantities of genuine mail 
filtered, then the filter is "not working". I agree that 100% filtration 
would be almost impossible, but (for example) 3% filtration is useless.

> Have you tried something like Spamassassin?

Nope. If fact, I've never actually installed spam filtering technology 
myself. I've only had it forced upon me by whoever is operating the mail 
server.

> And yes, you do work in an industry where the normal triggers for spam
> might actually be legitimate business communications.

Quite. This has no bearing on the spam filter at home being similarly 
useless, however.

> That doesn't mean spam filters are useless.
>
> You'll note that I didn't say *all* spam filters were wonderful and
> flawless.  I said that spam filtering generally works well.

Even I didn't say that *all* spam filters are useless. I merely said 
that I've yet to see one that isn't. (I even qualified it by saying I 
haven't seen all that many.)

Post a reply to this message

On Mon, 11 Jul 2011 11:47:31 +0100, Invisible wrote:

>> Firefox does not ship with AdBlock Pro installed.  People download it,
>> install it, and use it successfully.  I happen to be one of those
>> people.
>>
>> You, instead, have assumed "it can't possibly work" and so haven't even
>> tried it.
> 
> Before we get too far into this, I would like to re-emphasize what I
> actually said:
> 
> I said I had never bothered trying such software under the /assumption/
> that it won't work properly. I said that it looks intractably difficult.
> Then again, so does writing non-trivial software using something as
> primitive as C, and apparently people manage to do that.
> 
> In short, I would be surprised if ad-blocking technology can work. Not
> astonished, but definitely surprised.

Prepare yourself to be surprised, then. ;)

It works pretty well.  Not 100%, but IME better than 90%.

>> You're declaring failure because 1 message in "n" (for sufficiently
>> large values of 'n') gets through.  Fact is, spam filters aren't
>> supposed to block all spam.  They're supposed to reduce it, and for
>> most people, they do the job properly.
> 
> In my book, if I use a spam filter and still receive unacceptable
> quantities of spam, or have unacceptable quantities of genuine mail
> filtered, then the filter is "not working". I agree that 100% filtration
> would be almost impossible, but (for example) 3% filtration is useless.

We can agree on that last sentence.  Now, using various spam filtering 
technologies myself over the years, I've seen some good ones, and some 
excellent ones.

3% filtration is not what I've seen with any solution that I've used.

>> Have you tried something like Spamassassin?
> 
> Nope. If fact, I've never actually installed spam filtering technology
> myself. I've only had it forced upon me by whoever is operating the mail
> server.

One of the benefits of using something like fetchmail to pull from an 
external server to your own is that you can do this.  I use fetchmail to 
pull from my Earthlink account (which has its own spam filtering 
enabled), and fetchmail delivers to my local mailbox after running it 
through spamassassin to catch the ones that got through the Earthlink 
filters.

>> And yes, you do work in an industry where the normal triggers for spam
>> might actually be legitimate business communications.
> 
> Quite. This has no bearing on the spam filter at home being similarly
> useless, however.

No, that has to do with incompetence at your mail provider or an improper 
configuration on your account.  Since I don't know what they use, I 
couldn't comment on that.

>> That doesn't mean spam filters are useless.
>>
>> You'll note that I didn't say *all* spam filters were wonderful and
>> flawless.  I said that spam filtering generally works well.
> 
> Even I didn't say that *all* spam filters are useless. I merely said
> that I've yet to see one that isn't. (I even qualified it by saying I
> haven't seen all that many.)

Fair enough.  You seemed to be implying it in your original post, but re-
reading it, I see that I inferred something that perhaps wasn't there.

Jim

Post a reply to this message

On 7/11/2011 2:29, Invisible wrote:
> Ah yes, the argumentum ad populum. You realise that *millions* of people use
> Internet Explorer, right? So that *must* mean it's the best web browser. :-P

No, it means the web browser works. What you're missing is the fact that 
people complain they have to do something special to make IE6 work. But 
until very recently, everyone *was* doing something special to make IE6 
work, and hence IE6 works.

-- 
Darren New, San Diego CA, USA (PST)
   "Coding without comments is like
    driving without turn signals."

Post a reply to this message

On 11-7-2011 11:10, Jim Henderson wrote:

> Which planet are you on again? ;)

Andy is from Milton Keynes. I am not sure if that is a whole planet or 
tha last village populated by irreducible Celts that still resists to 
the Google invaders.

Post a reply to this message

On Mon, 11 Jul 2011 18:03:44 +0200, andrel wrote:

> On 11-7-2011 11:10, Jim Henderson wrote:
> 
>> Which planet are you on again? ;)
> 
> Andy is from Milton Keynes. I am not sure if that is a whole planet or
> tha last village populated by irreducible Celts that still resists to
> the Google invaders.

LOL

Post a reply to this message

>> AdBlock actualy prevent the ads from even starting to load.
>> It's realy a "Can't live without" addon.
>
> I've never installed this or any similar system under the assumption
> that it is almost guaranteed not to work.
>
> Designing a mere algorithm which can determine, with 100% accuracy, the
> difference between vital content and useless advertising garbage looks
> almost intractably difficult. Thus, the result would almost certainly be
> a system which either fails to block the majority of ads, or blocks the
> majority of useful content. (Or possibly both, at the same time.)
>
> In other news, I have yet to see a spam filter which actually filters
> out spam and nothing else. (Then again, I haven't searched extensively
> for one...)

The spam filter on Thunderbird is prety good and becomes beter with 
time. It literaly learn what is spam and what is not.

The spam filter on GMail is also exedingly good and effecient.

It's been a LOOONG time the last time ether of those missed a spam or 
mistoke legitimate mail for spam.


Adblock use a black/white list aproach. The black list contains an 
extensive list of known add servers.
If it misses some adds, you can right click it and set it to be blocked.
It keeps a log of blocked elements.


Alain

Post a reply to this message

Invisible <voi### [at] devnull> wrote:
> I've never installed this or any similar system under the assumption 
> that it is almost guaranteed not to work.

  Just try it. You'll be pleasantly surprised.

> Designing a mere algorithm which can determine, with 100% accuracy, the 
> difference between vital content and useless advertising garbage looks 
> almost intractably difficult. Thus, the result would almost certainly be 
> a system which either fails to block the majority of ads, or blocks the 
> majority of useful content. (Or possibly both, at the same time.)

  It doesn't need to do smart filtering. It only needs a blacklist. Said
list updates automatically.

-- 
                                                          - Warp

Post a reply to this message

On 12/07/2011 07:15 AM, Warp wrote:
> Invisible<voi### [at] devnull>  wrote:
>> I've never installed this or any similar system under the assumption
>> that it is almost guaranteed not to work.
>
>    Just try it. You'll be pleasantly surprised.

Well, yes, there is that...

(In a stunning turn of events, I actually have some work to do today. 
Yeah, I know. Imagine that...)

>    It doesn't need to do smart filtering. It only needs a blacklist. Said
> list updates automatically.

I don't get how that can work.

If it blacklists the server IP addresses, all you need to do is copy the 
ads to another server and the blacklist is defeated. As a bonus, if you 
put the ads on the same server as the genuine content, then you *can't* 
blacklist the server or you'll be blocking genuine content.

If it blacklists the MD5 hash of the ad files, just take each image and 
change 1 pixel. The MD5 hash is now completely different, but the ad 
still works just as well.

If it blacklists the file /name/, it's trivial to rename it.

If it blacklists the code fragments used to link to the adverts, just 
make any trivial cosmetic change to the code and you've defeated the 
blacklist.

All of this on top of the fact that blacklists are generally ineffective 
anyway. (If you filter anything - spam, malware, adverts, whatever - 
based on a blacklist, then that blacklist will always be out of date and 
you need an army of people to constantly update it for you.)

Post a reply to this message