Wednesday, January 06, 2010

Google Analytics, and sometimes I can't do math

First, my apologies to rich people: You're getting at most a $65 tax break over last year, not the 5% I promised you. I didn't account for the extra tax brackets that came with the new withholding tables. There are nine tax brackets now rather than seven, and things line up less disturbingly than I thought. Still, poor people have a higher tax obligation than last year, rich people a smaller one. The amounts are now insignificant, but the symbolism is still bad.

Here are the corrected charts, first, how much normal people will be paying extra, then how much less rich people will be paying. I'm including the new amounts for people filing as married, as well.

With that out of the way, I decided a month ago to use Google Analytics to track people who read my blog. Because I'm nosy. Also because I miss my web server logs from, the domain I used to run out of a closet in my house but let expire after the experiment had run its course.

The web server logs (apache on Debian, in my case [pre-Ubuntu]) contained nothing more than a date, an IP address, and the http request (the page being asked for plus whatever extra info your browser volunteers). With just those things you can infer a lot of information. You can find what company owns the IP address, which in some cases can give geography or a specific owner, often neither, sometimes both. Browsers often volunteer extra info in the http request, including the referring page, and what browser and operating system they are (or say they are, sometimes they lie). If a search engine was the referring page, you often get the search terms they used.

So in most cases you can find out roughly who or where they are, what they searched for, what they found, and some tidbits about them. I once found an inexplicably big influx in hits from a small college, for example. No explanation. I saw a lot of hacking attempts, and spent a lot of time blocking IP addresses, mostly from other countries, as they attempted (futily) to find old backdoors and security holes in my server. In 2002 I wrote the following quick blurb about just that in my blog:

Most of the attacks are attempts at buffer overflow, exploiting formmail, or are Microsoft IIS-specific. I quickly learned how to compile fw_ipchains support into my Kernel (running 2.2.20 on the web server machine) so I could filter out most of them. I wasn't in any real danger of being hacked, but my access.log file was getting huge.

Look how cute I was! All proud of being a site admin! Running Linux Kernel 2.2 back in 2002 was kind of backwards, though, but I was new. Anyway, in a fit of nostalgia, I signed up for Google Analytics to try to get some info on who is hitting this blog.

Google Analytics is a very nice interface, but doesn't give you access to the raw server logs, sadly. But what it does give you is very nice. Here's some snippets:

I have had 170 visits to my blog in the last month, most from people who stumbled onto the site from various Google searches. Here are the top entries that they linked to:

Farmtown : 31 hits (including the badly phrased google search "farm town: how to prevents people in market to coming onto your farm")
Mainframe Cobol Geekout : 26
Bread and Puppet : 16
My two entries about Facebook puzzles : 11
My last post, the "Rich getting richer" entry : 8

Here are some odd search terms that showed up:
eminent demise (awesome, more people should find me that way)
perky high class (I think this may have been the type of porn they were searching for)
short sci-fi story about rich wives donating organs in hospital (no idea. Anyone ever heard of this story?)
talent hunt in medicine grout (no freakin' clue what this guy was looking for)
zombie girlfriend flash (Sweet! I think they were looking for a flash-based video game)

Browser comparison:
Firefox: 96
IE : 34
Chrome : 23

Internet Explorer fails it! In fact, the new Chrome is almost as popular with my readers.

Geography. I'm making wild assumptions about who from each is reading, not accounting for random search-engine hits from the same area:

Ohio: 19 - My posse. 'sup? This includes a visit from Blacklick where 9 pages were viewed. I'm guessing this is my new FB bud who was my boss at an old job. I was always jealous because all the girls swooned over him. He had hair and wasn't an asshole, which made him more desirable than me somehow.

California: 5 - My buddy Craig from high school. Thanks for reading, man. And congrats on the marriage!

Virginia: 5 - Possibly family, more likely the feds updating my terror status

Wisconsin: 4 - The ex in-laws. My ex-wife's dad, Don, is pretty cool. We did geek stuff together a long time ago.

Florida: 3 - The Auterys and ex-Auterys. We all occasionally flip out and act irrationally toward each other. I blame Vietnam. Long story.

Vermont: 3 - Dominique and clan (Aunt Mimi to Scout)

So that's that. Lots of graphs and charts to geek out about. And it's free. If you keep a blog, I recommend it, not as a way to tweak your site to make money if you're a greedy little shit, but as a toy or a gauge of how popular you are compared to your buddies. Being that I'm paranoid, that's quite the boon.

No comments:

Post a Comment