Results tagged “ stats” from kwc blog

Slashdot aftermath


Yesterday's Slashdotting only brought in about 4-5x the normal daily traffic. As it turns out, I was actually getting hit by a one-two punch: two days ago was a big traffic spike due to a Obi-Wan Kenobi Valentine's image I posted from Something Awful (people were loading a 150K category archive page).

Looking at the number of visits:

08 Feb 2007 3832
13 Feb 2007 4782
14 Feb 2007 7946

15 Feb 2007 18078

18,000+ isn't that terrible in comparison to the normal 3,000-4,000 daily visits. It certainly would have been much higher had my server not been burning toast.

Yesterday was actually a good day from a bandwidth perspective (300-450MB is normal).

08 Feb 2007 412.45 MB
13 Feb 2007 658.47 MB
14 Feb 2007 946.67 MB
15 Feb 2007 457.37 MB

NOTE: The heavy bandwidth on the 13th and 14th was due to the Valentine's traffic.

# of pages and # of hits was only about 2x normal.

So, from a traffic perspective, it wasn't really the DSL line that was at issue. My Apache server (for whatever reason) wasn't able to handle the # of requests coming in. I blame Windows/Microsoft, as always (not my lack of skills in configuring an Apache conf file ;) ). Looking at my network utilization graph, the Apache server would handle the incoming requests well for a couple of minutes without fully saturating the link. Then there would be a sharp spike and the graph would flatline as the Apache server became unresponsive.

Perhaps the more interesting statistic was how I did with my Google ads. I believe I'm not allowed to share those directly, but I can make the following summarizations:

  • click-through ratios went down on my blog. Slashdot traffic is not ad-friendly.
  • click-through ratios on everything else went up about 3-4x.
  • yesterday still was not a 'banner day' with respect to ads

One nefarious conclusion one can draw from this is that slow site = better ad sales. This makes sense: people see the ad load, but the rest of the content is slow in coming and they decide to leave.

Speaking of stats (Tuning Apache)

| has slowly been degrading in performance, but thanks to some Apache tuning slides and some followup with bp I think I have bought some more time. The key was turning off the AllowOverride for nearly every directory on this site. Way back when had very little traffic, I had created an htaccess file to help me migrate from an older MovableType installation. It turns out that a large htaccess file and a sudden influx of traffic can bring your Apache to a halt.

I've wondered what the breaking point of my setup would be. I run on my home Windows desktop over 802.11g and DSL. Eventually it was going to start showing cracks. January was the first month that this site eclipsed 50,000 visitors and there is a slight chance of breaking 60,000 this month. This isn't very large compared to other sites: Alexa ranks somewhere between 116,013 and 383,693, which is not in the range that Alexa actually considers trackable. However, is now serving 6KB/s on average, with much higher traffic during peak times and as much as 800MB in a single day. Anywhere between 10-20% of my DSL bandwidth is being eaten up, so I may have to get more creative in the future to keep things running. I've also noticed MovableType degrading, often failing to rebuild files on the first try. It will be a race between MT and my DSL line to see which requires attention next.



I broke the magical $10 barrier with my Google adwords today. It took about a month to reach that mark, though it took about a week of fidgeting with the ad layout to figure out what was the best placement.

I monitored mybloglog to figure out which entries to place ads on, eventually choosing about 15% of my entries to mark with ads -- this results in about a third of my incoming traffic is being served ads right now. I would get a higher percentage if I placed ads on my category archives, but I'm not willing to do that right now.

Of course, given that In-n-Out pays $10/hr starting, I would make more money flipping burgers, but experiments are experiments. If only I could start getting some of the home loan ads or trial lawyer ads, I would really be rolling in the dough.

I've placed ads on this entry if you're curious to see them (if you don't see them, click on the entry title to go to the individual entry page). Search Keyword Fight I


In the spirit of mining stats to do auto-captioning, and inspired by wombat's Googlefight model of decision making, I bring you the latest to be scraped from my server logs: Search Keyword Fight I (roman numerals indicate the potential for future fights). Rules: each term below is given a score indicating the number of times it was used in a search query (Google/Yahoo/etc...) that ended up on For example:

  • mit 9
  • stanford 7

means that "MIT" was used in 9 search queries, and "Stanford" was used 7 times.

Round 1: Gender superiority

Hard to judge this round: man has a slight lead in the individual scoring, but Batman was a suprise third.

  • man 22
  • woman 16
  • batman 10
  • stickman 8
  • manchuria 7

Women shutout the men in the team competition, and the co-ed X-Men team posted a respectable score for homo superior.

  • women 34
  • x-men 6
  • men 0

Round 2: Warrior supreme

  • samurai 50
  • pirates 9
  • ninjas 6

Note to Hollywood: do more samurai films (w/o Tom Cruise).

Round 3: Computer Religion

I couldn't help but notice these two fighting neck-and-neck for position among my stats:

  • apple 67
  • microsoft 67

Round 4: Morality

  • evil 33
  • hate 13
  • good 9
  • love 0

I'm not sure how to interpret this round. Is it my site is evil -- I do write about evil bunnies from time to time -- or is it that people who use search engines are looking for evil/hateful content?

Bonus Round

This is a search query stat fight, so we'll end the scoring with a fight between the search engines themselves:

  • google 125
  • yahoo 11

My site's Amazon page


This is sort of a strange cross-over. As a result of visit Joi Ito's site, I found out that Amazon is now putting up listings for Web pages using Alexa data. This listings follow most of the normal convention for books, musics, and other products, including info such as: * Traffic rank (instead of sales rank) * Contact Info * Thumbnail of your site * People who visit this site also visit * Ratings

The latter two are apparently only available to high-ranking sites. When your traffic rank is in the millions like mine, they only provide basic info. I can't seem to find an easy way to browse info for different Web sites, so it seems that the easiest way to get info on your site is to go to Amazon's A9 and type in the URL there, then click on "Site Info" next to the appropriate search result. * website info: * website info: bOING bOING for an example of what a popular site looks like on Amazon

Samplin' the google rankin'


I found an online tool that returns the GoogleRank of your site (via blogdex). I posted some samples here, mainly because I thought it was interesting to see how easily blogs can destroy search engine results. As you can see, meta's and my blogs are only one pagerank away from Neil Gaiman, which really just doesn't seem appropriate, because he's way more popular, his blog is far more interesting, and has been around much longer. Also, if you were to look at our Technorati Cosmos, we're nowhere close to the sites that Google puts us next to.

I could not figure out any URLs that were a 1, nor could I find a Web site that was legitimately a 2. For the site that came up a 2, I added "www" to the front of it, and it became a 3. 0/10 3/10 3/10 4/10 5/10 5/10 5/10 5/10 6/10 6/10 6/10 7/10 7/10 7/10 7/10 8/10 8/10 9/10 9/10 9/10 10/10 10/10

google, you're _too_ good to me


While I was sick, the traffic on this site suddenly quadrupled. Normally, this would be an exciting event for me, and make me wonder how I ever managed to be so clever, but it turns out that it was most likely because Google image search for "gollum" was returning a screencap off of this site (no, not the sketch you see here in this entry). Also, people were inlining the image in their Web pages straight from this site. It wasn't enough to hurt yet, but it was such a skew from the normal I had to take preventative action.

pqbon and this useful guide both deserve some Google juice; pqbon gets it for telling me that I could block the inlining practice using a .htaccess file, and the guide gets it for giving me specific instructions for using Apache's mod_rewrite to get it done. Two things that I wished the guide mentioned, though, were these two steps:

1) change AllowOverride None to AllowOverride All in httpd.conf so that .htaccess files are enabled
2) uncomment the load mod_rewrite line in httpd.conf so that the rewrite module is actually loaded

If this were four years ago when I was regularly editing httpd.conf, I probably would have remembered those two ditties, but my old age makes me forget things nowadays.

gollum sketchGoogle: It was an honor, really, to have such a cool search term top my lists. I brought back my Gollum sketch for this entry to keep me company now that all the Gollum fans won't be siphoning my bandwidth anymore. Please respider this page one last time, read my new robots.txt, and eviscerate that image from your brain.

Xanga people I don't even know: I'm sorry -- I know you can't upload photos, and I normally would't mind, but it was too much. Maybe xanga2mt will be of some use to you.

This page should show you what happens when you try to inline now: a tiny blue box. zeitgeist


Google posted their year-end Zeitgeist, which is always a fun perspective on the content the Web provides. The Tour de France made the top ten, woohoo! Orlando Bloom made #3 on the popular men search, which is partly due to the efforts of metamanda (I wonder if they know to combine "elfin nugget" with his results yet).

I'm going to demonstrate some bad manners and post my own year-end stats here, not as an act of self-promotion (because the numbers are kinda pathetic), but because this is entry #800, and if there's ever a time to draw attention to yourself, nice multiples of a hundred are it. Also, I find it interesting to see how the number of visits to this site is steadily increasing even though the set of regular readers has not. This increase in hits is 99% due to search engine links. Apparently Google thinks I am an expert on Khleo generics (~300-400 searches/month), and my article mentioning YzDocklet is quickly gaining steam. Also, 61 people visited my site last month because the "Redskins Suck."

Unique Visitors 2003
May 2003: 280
Jun 2003: 709
Jul 2003: 1542
Aug 2003: 2160
Sep 2003: 3251
Oct 2003: 4237
Nov 2003: 4914
Dec 2003: 6129

Top Fifteen Search Topics Dec 2003
khleo generics/movie holes 364
statue venus london 145
ipod stuff 135
yzdock 118 78

atari adventure 71
cool inventions 62
redskins suck 61
terrell owens sharpie 51
frank gehry disney 51

internet blockers 45
wireless power 38
callipygous 35
photo stitch 33
stupid stuff 25

Top Ten Search Phrases for September


I have nothing else to post, so I thought I would show off my stat-hoor side and list the top ten search phrases that lead people to this site in September (according to awstats). Hopefully, by listing these search terms I will reinforce their ranking and thus skew the results even more in their favor:

ultimate fantastic four (74)
the movie holes (46)
movie holes (41)
wavecrest bike (33)
picasso guitarist (32)
david statue (31)
internet blockers (29)
venus statue (28)
exoskeletons (27)
waterclock (24)

The top keyword for this site, excluding common words (of, the) was... 'japan'