CDR: Re: Ho to KICK OUT Junkbusters users

Greg Newby gbnewby at ils.unc.edu
Sat Oct 28 17:15:22 PDT 2000


On Sat, Oct 28, 2000 at 11:19:52AM -0500, Igor Chudov wrote:
> 
> I have a website (www.algebra.com) that makes money from banners. I have
> a suspicion that a small percentage of my users uses Junkbusters proxy
> in order to avoid seeing my banners.
> 
> I do not want to serve such users at all and I do not want them to use
> my bandwidth.

I don't think you're going to get a lot of sympathy by wanting to
force people to view ads.  Although it's your bandwidth at the server
end, it's the USER's bandwidth (& time) at their end.  I nearly always
leave images turned off in Netscape so I don't need to wait for ads to
download via a modem.

> Is there any way to detect a user of Junkbusters in a CGI/mod_perl script?

Good question, I'd say the answer is 'yes' but it's but probably not
worth the effort.  Your basis would be what files they download (and
yes, it seems you could automate assessing this).

It seems that you could configure your Web server software to stop
serving data to hosts that don't also download the banner images (if
the images are coming from YOUR server; this isn't clear -- if not,
you might be able to redirect through your server to get the images).

This would also block:
	- people using Lynx (e.g., many blind users or people at 
public access text-only stations)
	- people with their images turned off
	- people who press their browser's STOP button before 
everything is downloaded

Let's assume you could catch Junkbusters users this way.  Their
workaround might be to GET the banner file, but abort the connection
before the file transfers.  Much tougher to catch..  You're be burning
far more CPU cycles than just for regular Web serving.  It's not clear
to me that the current Junkbusters (or their Guidescope) software
could do be modified to work around such blocking.

Your question was whether you could do this type of detection using a
CGI/mod_perl script.  It seems you could "tail" the server log file to
get the data you'd need (& combine with the regular client info you
can get from the Web server).  You'd also need to maintain a state
variable across multiple hits to the server so you know what clients
have downloaded the whole set of files you're testing.  Seems
feasible, but potentially error-prone.

  -- Greg





More information about the cypherpunks-legacy mailing list