Slashdot Mirror
Addendum to The Slashdot Effect Internet Paper
Spock_NPA writes "An addendum to the Slashdot Effect Internet Paper is available here. It details the effect of Slashdot on the Slashdot Effect Internet Paper."
← Back to Stories (view on slashdot.org)
Seriously, the white paper and the addendum are interesting looks into many abject statistics...bundled up and analyzed to make sense against the fact that "[insert Slashdot featured server name here]'s connection is crap" or that their machines can't handle the load. I shouldn't complain, my sites are run on a shared server that wouldn't dream of handling Slashdot-caliber traffic just for a single hosting client.
Interesting white paper, I enjoyed reading it.
--------
Oscarfish.com: tropical fish with attitude. Way t
I'd love to know the answers to the following:
How many registered people are there on Slashdot ?
How many unregisterered people use Slashdot ?
Maybe Slashdot ought to have a statistics page, showing the effect of the Slashdot effect on Slashdot!
Donte Alistair Anderson Roberts - hi son!
Karma: Chameleon
Isn't it time for the slashdot crew to write a statistical analysis of the reading frequency of the slashdot stories about the slashdot effect papers? If everybody keeps this up, we'll have a new academic subdiscipline in no time! "slashdot science", maybe, or "statistical geekology". If we get the terms 'postmodern' or 'cultural imperialism' into the title as well, we'll be rolling in grant money...
Trust the Computer. The Computer is your friend.
"The Slashdot Effect as documented on the addendum documenting the Slashdot Effect on the original article on the Slashdot Effect."
Some
people
have
too
much
time
on
their
hands.
"Everybody's Got Something to Hide Except for me and my monkey" - The Beatles "If you're not part of the solution, you'
What about including a click-through counter in every link that's posted on ./? And then have a top ten ranking which links got clicked most in the last month or so (or maybe ranked by clicks/time unit).
That would also be interesting to get an idea what load these servers have to deal with. I am not sure if the Slashdot Effect paper is really very representative.
EagerEyes.org: Visualization and Visual Communication
Yesterday's Red Hat article doesn't seem to have any first posts. It was released in the middle of the night after the kids had been sent to bed.
Well we can't say when, but the addendum does address the question:
250 hits per minute. Now we know how big the slashdot effect is. Has anyone else seen surges this big (from 30/min to 250/min) from other sources other that /.?
There is a second resurgence, on a much smaller scale, two day's later which peaks at about 9:30am, with an abrupt fall off and then a re-resurgence around 6pm that day. This is seen in Figure 4. plot. The author scanned the various Linux news web pages for posting of the /. effect article but found none. This small resurgence occurring 2 days after the initial posting by /. is unexplained and open to interpretation.
.htaccess/cgi hacks. I've found some interesting things this way, including links from news sites/mailing-lists/newsgroups from over the world that I wouldn't have found otherwise. A link to the author's paper could have been posted on a russian linux news site, for example.
.html, this isn't a problem as you can have a .htaccess file that specifies something like "AddHandler cgi-script .html" or you can point people to the directory and respecify the DirectoryIndex.
;print;
It's easy to log the referrer in Apache. I do it using a combination of
If you're curious as to how this can be done under Apache, here's a rough description (apologies in advance for any ugly or careless code and any "Slashdot munging" in indentation, etc that may occur).
(1) Point people to a CGI script, not the direct page. If you want this file to end with
(2) Now here's the fun part. The script dumps the page to the requester then logs the interesting environment variables set by Apache.
Example script:
#!/usr/bin/perl
# Navindra Umanee <navindra@cs.mcgill.ca>
# Give them the page.
open(indexPage, "/path/to/real/content.html");
@fileStats = stat(indexPage);
$modifiedTime = gmtime($fileStats[9]);
$modifiedTime =~ s/^(...) (...) (\d\d?) (\d\d:\d\d:\d\d) (\d\d\d\d)/$1\, $3 $2 $5 $4 GMT/;
# first, print HTTP header
print "Content-Type: text/html\n";
print "Last-Modified: $modifiedTime\n";
print "Content-Length: $fileStats[7]\n";
print "Accept-Ranges: bytes\n\n";
# then, print body
while(){
}
close(indexPage);
# Now log the information.
$log="../khtmltest.log";
$localtime = `date`;
open(logFile, ">>$log");
print logFile "$localtime";
print logFile "$ENV{'REMOTE_ADDR'} - $ENV{'REMOTE_HOST'}\n";
print logFile "$ENV{'HTTP_USER_AGENT'}\n";
print logFile "$ENV{'HTTP_REFERER'}\n";
print logFile "\n";
close(logFile);
Cheers,
Navin.
I'm not sure if I'm babbeling sh*t here or not, but I am ABSOLUTLY SURE I have seen that paper before, reported on /. 6 or 12 months ago. It *MAY* have been the origional but I am fairly certain it was an addendum to the origional reporting the effects of the /. effect on the report of the /. effect.
/. etc... I dunno. Comments included "what next, a report on the /. effect on the report of the report of the report of the /. effect" etc.
The graphs had lines saying when it was posted on betanews,
Again, I may just be dribbling sh*t (or dajavu or something)
anyway.
Cyas
Ah suck, Slashdot did munge the stuff very slightly despite my best efforts and I managed to press "submit" instead of "preview" before fixing some of the code and removing that extraneous ';' before one of the print statements there.
Oh well, if you have trouble figuring it out, you can always email me and ask.
uptime: 50 days, 21:18, 1 user,
load average: 0.49, 0.35, 0.24
processes: 132
yesterday: 76525
today: 1
ever: 203679322
These stats are shown in a 'slashbox' for registered users. A Custom Page that show these stats along with many other items.
What's missing from the /. stats is the average ratio between 'proccesses' and 'unique visits'. This number would prove valuable. Rob? are you reading this ? :)
My name is Dr. Vinnyboombas and I seek funding for the following research:
Building a quantitative model of Postmodern Slashdot Societies effect on Cultural Imperialism and Statistical Geekology. and ummm oh yea..can I have a supercolider and a mindstorm kit?
Thank you Santa, I've been a good boy this year. :)
Even geeks sleep in on Sunday.
The two-days-later surge would have been easily explained if he had looked at a referer log.
It's the time when the story hits the top of the "older stories" list.
Harald, counter.li.org slashdot victim/beneficary
But look at it another way:
A moderator, who has only five points to begin with, moderates your post to -1, using one of his points.
Because of _you_, another more deserving post doesn't get moderated up.
So even if it seems like it doesn't hurt someone...
it does.
I would also posture a guess that at least 25% of the people experiencing the /. effect are indeed experiencing normal lag (packets dropped by routers a few hops away) combined with bad timing.
For example: Prime time (United state 7PM to 10PM) evening net use is 'spotty at best'.
The "today" stat is broken. It always says 1.
-jacob
That would imply, though, that /. gets ~ 1 hit/minute, which isn't very plausible. I've actually never seen that counter be anything other than 1.
On the other hand- as I've been checking, the "yesterday" figure keeps increasing. Probably the slashbox is just broken so that yesterday is today's stat and today points off into never never land somewhere.
-jacob
> actually i think all this "ph1r5t p05t" stuff is rather funny.
/. is if you step back and look at it. There's also the meta-moderation that started after the moderation scheme was set up. And then there is the meta-meta-moderation comments that people started making after they noticed the meta-moderation. (It was amazing how quickly the term "meta-moderation" sprang up and came into common use.) Now I'm sure someone will point out that my commenting on meta-meta-moderation may qualify as meta-meta-meta-moderation; and of course you know what the comment on that will qualify for.
A lot of stuff on
Other things are the numerous "that's not worth posting" posts, which always strike me as really ironic, since someone is taking the trouble to comment on something that they say wasn't worth mentioning in the first place.
And then there are the obligatory "that's been mentioned before" posts.
And the "just got to work my political views in" posts, where people can't resist mentioning the moral status of [Bill_Clinton, George_Bush] (almost always in the subject line, since it doesn't do much good to hide inflammatory views down in the body of a post).
I'm sure there are more that I haven't noticed. Perhaps this would be a suitable topic for a companion paper, or at least a Guide to Slashdot.
All in all, I can imagine that this is really an entertaining site for those lurkers with a sense of humor about human nature.
--
It's October 6th. Where's W2K? Over the horizon again, eh?
Sheesh, evil *and* a jerk. -- Jade
Actually, I seem to recall seeing the counter be greater than 1. Before Rob got the new servers, there were days that it would be 2, 3, 4, or even 5. I never saw it larger than 5, though.
I had always assumed it to be current users (or current instances of root). This conclusion came from the fact that the more other people posted (Hemos, Cliff, Robin, et al), the higher the number was...
'Course, I too, could be full of it...
Jedi Hacker (Apprentice) and Code Poet
censorship is a form of noise, which actively seeks to drown out content with silence - Crash Culligan
I am such a geek that I found so FUNNY! Nothing like a bit a chemistry humor on a Sunday afternoon to brighten my day...
Teacher: "Is the bond covalent or ionic?"
Student1: "Isn't it ionic?"
Student2: "Don't you think?"
"Don't mind me cutting myself on Occam's Razor"
It occurred to me that ad banner companies could make a bit of extra cash from the slashdot effect. Suppose you were an ad company with banners at many different sites, if you kept track of which sites were referenced on /., you could sell advertising on those sites at a premium during the /. effect. It would be like buying a spot during the Super Bowl. You're guaranteed to have a whole bunch of eyes look at your ad.
:-)
Of course it probably won't work too well if you're trying to advertise MS Windows 2000
Laissez lire, et laissez danser; ces deux amusements ne feront jamais de mal au monde. - Voltaire
I propose that we launch a massive Slashdot statistics program. It would involve a significant amount of invasion of "privacy", i.e., almost every link off of /. would be logged, but we could get a lot of raw data to look at in interesting ways. Here are a few of my ideas:
There are more! Let someone else suggest them, though. I think that analyzing the raw data gained by these approaches would be a good statistics project for any upper high school or college student. And hey, I like Slashdot, and if I get credit for it, even better! (Well, it would be mostly raw data, not the real site, but still...)
Ken
PS - I would do such a project. Anybody who wouldn't?