httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Andrew <and...@aaaaaaaa.demon.co.uk>
Subject Re: idea for new module...
Date Thu, 25 Jan 1996 12:48:20 GMT
> What do you say?
> 
> 	Brian
> 
> <!-- Anna Nicole Smith Anna Nicole Smith Anna Nicole Smith Anna Nicole Smith
> Anna Nicole Smith Anna Nicole Smith Anna Nicole Smith Anna Nicole Smith Anna
> Nicole Smith Anna Nicole Smith Anna Nicole Smith Anna Nicole Smith -->

[snip]

A little story about the webserver survey to further illustrate
this point...


The survey's been running for about 6 months now and since its
initial announcement the hit count against the pages has been rising
slowly.  As expected there's a sub exponential feel to the rise in
hits, it's catching on which is encouraging.  Every month we see
a peak, round about the 1st when the new figures come out and then
it rapidly decreases back to a steadily increasing background level.

One week strange things started happening, our monthly peak didn't
fade and instead showed no signs of decreasing 2 weeks after the
'new survey is ready' announcement.  I started fiddling with the
stats routines filtering out robots (32 of them for god's sake)
and still the figures seemed weird.  A check to the referer log
showed we were taking a lot of traffic from the page:

	http://www.tcp.com/~jlick/

Looking at that page we saw a link to the survey's 'whats' script,
which tells you what server is being run on such and such a host.
Ok, fine, 'whats' was written as a form which used GET rather then
POST so that people could embed its URL in their pages.  It was a
deliberate ploy to make the survey more accessible.  Only how come
some guy's home page is getting > 1million hits a month?  Who is
he?  Elvis lives, and works at tcp.com?  Weird.

Ok, looking at ~jlick's page we see that he's also clued into the
presently 'mindless' way that robots index the page content they've
found.  ~jlick's explaination is more entertaining than I can manage
right now, but to paraphrase:  a bunch of his friends sat down one
day and wrote out all the rude words they could think of and then
typed them into his home page.  After the robots visited his own
hit count went ballistic because (prepare to be shocked dear reader)
most people on the internet spend all their time searching for any
and all things to do with sex.

Imagine little Johnny's distaste when, after his parents have gone
out for the evening, he steals into his dad's study, fires up the
p5 and begins his lifetimes work of crawling for porn, when ...
HORROR!  His innocent eyes are assaulted by the gratuitous wise-guydom
and wanton abuse of the internet perpetuated by ~jlick's latest
crafty prank.

The tale doesn't end there.  Little Johnny (and a million other
likeminded individuals) figures this must be a crafty ruse.  Surely
there's something here he can play with?  What does this link do?...
Wow, look, another form to fill in!  Ok, what do I type?  Mmmm....

	host: hot+babes
	port: 80

Er, nope, nothing there.

	host: tight+leathuur+pants
	port: 69

Damn, what a fraud!

*sigh*

Ok, so now back at the bat cave our access stats are being skewed
all ways to christmas by this new flood of dorkdom.  I was initially
so pissed by this wanton abuse of my latest toy that I hacked on
the 'whats' script to use 'Location:' to bounce the browser all
over the net whenever $ENV{HTTP_REFERER} =~ /www.tcp.com/ !  After
a while I relented (hey I'm only human) because:

a)	it's kind of nice to get those hits and one of these people
	might just be some important company per$on taking a lunch break
b)	it's actually pretty uncool to loose your temper when you
	can make some advantage

So now 'whats' just adds a polite little extra bit of verbiage
explaining what the survey's all about if $ENV{HTTP_REFERER} !~
/www.netcraft/ .  Which means that if you already know about the
survey then you don't get any of the additional promotional noise.
Perhaps one in 20 dorks turns out to be curious enough to visit
the survey proper.

So, what does this prove?  Basically that if all you care about is
hits then don't try to be clever.  Just use the apache mod_hotbabes.c
module and watch your access stats explode.  Snake oil for the
internet.

Cheers,
Ay.

Mime
View raw message