The Adblock Project Forum Index The Adblock Project
Pull up a seat ...stay a while.
 
 FAQFAQ   SearchSearch   MemberlistMemberlist   UsergroupsUsergroups   RegisterRegister 
 ProfileProfile   Log in to check your private messagesLog in to check your private messages   Log inLog in 

Features in Adblock 0.6?

 
Post new topic   Reply to topic    The Adblock Project Forum Index -> Main
View previous topic :: View next topic  
Author Message
TychoQuad
Guest





PostPosted: Fri Mar 18, 2005    Post subject: Features in Adblock 0.6? Reply with quote

Sorry if I've caused a mod to jump at this thread, but I'm not making this thread to ask what's in 0.6, or when it's coming.

What I am here to discuss, is what could they possibly impliment that would take so long? I'm not saying this in a negative way either, I want good examples of what would take a while to do, but would add greatly to AdBlock. I know Rue from use of SessionSaver, and have helped him diagnose some really nasty bugs in it, and watched as he implimented some brilliant changes in relatively a short amount of time.

So I'm going with the assumption that the changes in 0.6 are so huge that they almost require a complete rewrite, otherwise development on 0.5 wouldn't have come to such a screeching halt. (I'd imagine smaller releases adding small but cool features would be out by now if that was the case)

What I think are possible things for 0.6 are:
  • Full Proxomitron filter support. That would be cool, as there are stacks of good complex filters avaliable, and i wouldn't need to run an external program
  • Bayesian filtering like what Thunderbird does for spam. It would be brilliant to have your browser "learn" what is an ad, and block them automatically once you have properly taught it
  • Image placeholders. I like the idea that ads would be replaced with an image placeholder which would replace all ads with an image which could be used to add certain images to a whitelist or as another means to block them, a smaller feature than the other 2, but would be hard to do.


Of course this is all guesswork, but I would be happy to the moon if anything I've said comes true. It's also completely possible that Rue and co come out with something even better, we will have to wait and see. I'm personally very interested in what comes out! I don't believe for a second that AdBlock is dead.

-TychoQuad[at]yahoo.com.au
Back to top
G



Joined: 10 Oct 2004
Posts: 550

PostPosted: Fri Mar 18, 2005    Post subject: Reply with quote

1. Are you implying that there aren't good filterlists for Adblock?
2. Bayesian filtering is completely unnecessary. A good filterlist will remove 99% of ads, and most of what remains can't be dealt with by bayesian filters, since they contain exactly the same URL strings as legitimate content. The only way to deal with the last 1% is user-created site-specific filters (unless you want false positives).
3. I'm not sure what image placeholders have to do with whitelisting.
_________________
Filterset.G
Back to top
View user's profile Send private message Visit poster's website
TychoQuad
Guest





PostPosted: Fri Mar 18, 2005    Post subject: Reply with quote

1. Of course not. I use yours, and it does a brilliant job, and gets better every release. However, there are things you cannot do with mere url strings. For example, Proxomitron can block cookies, change referrers, click on ads, exetera. Downside is that it is harder to use of course. Every program has a way to get better, and AdBlock has too. this was only an idea. a possibility.

2. That's completely your opinion. I believe Bayesian filtering can exeed any other current form of ad killing, once properly trained, catch more than any blacklist. Of course there would be false positives, but that's something that can be solved with a SIMPLE whitelist, and identifying false positives is where idea 3 comes in.

3. Images that have been flagged as ads could be replaced with tiny placeholders, which remove the ad, but also help diagnose any problems with the site. If you think the site may be dodgily coded, you can tell if it was the fault of an ad or not. It also makes it easier to identify exactly what is being blocked when it shouldn't be. Clicking on one of these (or right clicking, whatever) could bring up the same entry you get from Object Tabs, allowing you to easily white/blacklist the selected site.
Back to top
TychoQuad
Guest





PostPosted: Fri Mar 18, 2005    Post subject: Reply with quote

Whoops, totally forgot this part:

You seem to have totally missed the point of this thread. It was NOT to pick holes in any of my dream features for the future of AdBlock, but to come up with equally outrageous ideas for features that could possibly be in the works. I don't think anyone will argue that 0.6 isn't taking it's time, but knowing Rue's work, I'm willing to bet that it's because there is something groundbreaking round the corner.

This is a speculation thread, not a difinitive feature request thread.
Back to top
G



Joined: 10 Oct 2004
Posts: 550

PostPosted: Fri Mar 18, 2005    Post subject: Reply with quote

http://www.example.com/images/1.gif is legitimate
http://www.example.com/images/2.gif is an ad
There is no consistant numbering pattern that determines what is an ad and what is not.

Please explain how a bayesian filter in conjunction with a simple whitelist will solve the problem.
_________________
Filterset.G
Back to top
View user's profile Send private message Visit poster's website
TychoQuad
Guest





PostPosted: Fri Mar 18, 2005    Post subject: Reply with quote

Oh, I'm sorry, did i mention somewhere down the track that all features currently in adblock would be senselessly destroyed? *checks* No, i didn't think so. You would of course use a normal ad filter in that case. Once again, you totally missed the point of this thread. Please refrain from futhur commenting unless you are going to present your idea for a dramatic change to AdBlock.
Back to top
G



Joined: 10 Oct 2004
Posts: 550

PostPosted: Fri Mar 18, 2005    Post subject: Reply with quote

My point is valid, whether you like it or not. There is simply no need for a bayesian filter.
_________________
Filterset.G
Back to top
View user's profile Send private message Visit poster's website
ShunterAlhena



Joined: 26 Oct 2004
Posts: 40

PostPosted: Sat Mar 19, 2005    Post subject: Reply with quote

Dear G,
I don't see why you are so hostile. The thread starter was very polite, even after you posted several unfriendly posts.
And yes, there IS need for bayesian filters. See, your Filterset is marvelous but it doesn't block Hungarian ads. (It has around 80% ratio.) And it's impossible for every nation to have a dedicated guy doing all these updates for free. But a Bayesian filter would learn what I need and would keep learning even without updates.

Also, I'll try not be off topic unlike some fellow posters. My wish (as I see little possibility that Bayesian filters will come Sad ) is a reliable filter updater. It's not some drastic change, but having the updater leave my custom filters alone and the updated ones deleted/changed as needed is something I think needs quite a few days of coding.

SA
_________________
"Look at you hacker
a petty creature of meat and bone
panting and sweating as you run through my corridors...
How can you challenge a perfect, immortal machine?"
SHODAN, System Shock
best game ever - the only one I ever bought as well
Back to top
View user's profile Send private message MSN Messenger
G



Joined: 10 Oct 2004
Posts: 550

PostPosted: Sat Mar 19, 2005    Post subject: Reply with quote

The problem is that a bayesian filter simply won't work for internet advertisments, at least not when dealing with URLs, image sizes, and positions alone. The reason it works for spam is that the content of spam differs in obvious ways from legitimate email. Internet ads are frequently inseparable from legitimate content when analyzed on anything other than the interpreted content (image analysis is the only pathway to this). Bayesian filters, when applied to internet advertisements, will result in a large number of false positives. In the end, they'll be restricted in precisely the same way Filterset.G is.

Two questions for you:
1. Why is it "impossible for every nation to have a dedicated guy doing all these updates for free"?
2. Filterset.G does block Hungarian ads. I receive emails from all over the world every day, and there are dozens of filters which specifically target ads on non-English sites.
_________________
Filterset.G
Back to top
View user's profile Send private message Visit poster's website
ShunterAlhena



Joined: 26 Oct 2004
Posts: 40

PostPosted: Sat Mar 19, 2005    Post subject: Reply with quote

You might have a point there. Indeed, many URLs don't indicate the content of what they'll display. But this image analysis seems cool: if there is a project dedicated to image analysis, can you post a link where it (at least some prerelease theory Smile) is explained?

Q1: I simply don't believe that there is a man for every nation who has your skill and dedication to maintain a quality filterset for a free extension of a free browser with (relatively) small penetration.

Q2: This is great to hear. Still, I encounter some ads while surfing in the hu domain. If I send the raw URLs to you, can you include them in the filterset (probably by tweaking some existing rules... it's your field of knowledge:))
_________________
"Look at you hacker
a petty creature of meat and bone
panting and sweating as you run through my corridors...
How can you challenge a perfect, immortal machine?"
SHODAN, System Shock
best game ever - the only one I ever bought as well
Back to top
View user's profile Send private message MSN Messenger
G



Joined: 10 Oct 2004
Posts: 550

PostPosted: Sat Mar 19, 2005    Post subject: Reply with quote

My skill comes from experience - prior to Adblock, I had never even heard of Regular Expressions.

Send the URL of the page on which you encounter the ad, not just the URL of the ad-object.
_________________
Filterset.G
Back to top
View user's profile Send private message Visit poster's website
s33m33



Joined: 08 Jan 2005
Posts: 59

PostPosted: Sun Mar 20, 2005    Post subject: Reply with quote

hey tycho, this came off the adblock homepage
Code:
And we're targeting these items for future builds:

    * Bayesian probability analysis, for minimal-interation blocking


at least one of ur prayers has been answered Wink
Back to top
View user's profile Send private message
TychoQuad
Guest





PostPosted: Mon Mar 21, 2005    Post subject: Reply with quote

Wow, I haven't looked at the homepage for ages, I wasn't even aware it was there. As I said earlier though, this is not a hope, or a prayer, but would no doubt be really cool, and possible explination on what is taking all the development time. I have no doubt whatever it is, it will be really cool, and I am still yet to hear anyone else's ideas, as is the point of this thread.

G, since you ignore my requests, I will ask you directly, rather than using the same question i've aimed at the general readers here: "What do you think would be a positive improvement for AdBlock that may take longer to build than your average release?"*

This thread was not entitled "what do you think of these ideas for AdBlock" or "do you think these would work for AdBlock" but "what are your ideas"
Back to top
G



Joined: 10 Oct 2004
Posts: 550

PostPosted: Mon Mar 21, 2005    Post subject: Reply with quote

Since you asked: I do not think that this is a positive pursuit. The inevitable outcome is that people will get their hopes up for features that aren't in 0.6, and are let down. We don't need to pressure the developers into feeling like they need to put in more features.

In response to your annoying fine print: anyone who feels the need to stipulate that their ideas should not be carefully considered or critiqued should not discuss them in public. A forum is, by definition, a place for open discussion.
_________________
Filterset.G
Back to top
View user's profile Send private message Visit poster's website
TychoQuad
Guest





PostPosted: Tue Mar 22, 2005    Post subject: Reply with quote

Yeah, your probably right about that. I was trying to stimulate positive conversation in here rather than repetitive negative, but it seems I have failed miserably in many respects. Good day everyone.
Back to top
rue
Developer


Joined: 22 Oct 2003
Posts: 752

PostPosted: Tue Mar 22, 2005    Post subject: Reply with quote

Tycho:
I think what G meant was: "please stop suggesting ideas that will put me out of business. P-lease."
..In fact, it was more of a "Psh!- puhlEEze " ..with a flick of the wrist, and marvellous dash of accent.
.
Not like I was there.

.
ps: all is, as you suspect, far-from-dead
Back to top
View user's profile Send private message
G



Joined: 10 Oct 2004
Posts: 550

PostPosted: Tue Mar 22, 2005    Post subject: Reply with quote

On the contrary, rue. I've simply been fighting false positives since day 1. Note that it was I who suggested image analysis in this thread as a solution to the Bayesian problem - I'm not against being replaced by something better.
_________________
Filterset.G
Back to top
View user's profile Send private message Visit poster's website
s33m33



Joined: 08 Jan 2005
Posts: 59

PostPosted: Tue Mar 22, 2005    Post subject: Reply with quote

dont feel bad G. humans have been replaced by machines in many areas.

just hang in there Smile
Back to top
View user's profile Send private message
Display posts from previous:   
Post new topic   Reply to topic    The Adblock Project Forum Index -> Main All times are GMT + 1 Hour
Page 1 of 1

 
Jump to:  
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum


Powered by phpBB © 2001, 2005 phpBB Group