Google as empirical/statistical analysis tool.

[Google]

This is something I’ve seen popping up for a while now, and had explosive growth in the past few months: Google being used as analysis tool. People now use Google to measure simple statistics, and expect to get decent results.
People, quit it.

In short, this is the idea: People use Google to measure how many times a certain string is mentioned on the Internet, and based on that, they draw conclusions.
An example of this technique. Lets say, we want to measure how many people like Coca Cola, and how many people like Pepsi. Okay, easy enough. Lets see…
Query one: “I like coca cola”. Results*: 7.370 hits.
Query two: “I like pepsi”. Results*: 13.600 hits.
So, this means that about 65% of.. the internet(?) drinks Pepsi.
Wait, WRONG!.
Of course, you’ve realised by now how stupid this is. The dataset you are getting your facts from is very select, very small. They are mostly Western, internet-connected posting-online humans ages varying mostly from 14 to 40 (or something). Oh, and they’re ALL saying exactly, “I like coca cola/pepsi”. EXACTLY. Statistical analysis my ass, this is clearly bullshit. Google can’t be used to measure “complicated” stats like soda love. Google does loads of things (things you aren’t even aware of, probably) but serving stats is NOT a possibility. Don’t ever use it this way. You can’t just search for indexed-by-luck, perfectly-matching strings.
The Google Search Engine is not smart. It is actually, one of the most stupid crawlers ever. Sorry Google, nothing personal (cough). Anyway, it’s not really the crawling that makes Google “stupid”, it is interpretation that is being a *****. Google doesn’t know what ObamafanX319 meant when he posted a sarcastic sneer aimed at a few Pepsi lovers when he said “**** yeah drinking liquefied nutsacks rocks, I like Pepsi!. Google can’t interpret this. Google does _not_ interpret anything!
When I google “I love niggers” (wait for it) the third result returned actually is this little gem: “I love niggers. I think everybody should own at least one.” Yeah, racial jest aside, the web isn’t semantic. The web is a sarcastic piece of shit, you can’t search through piles of junk for usable empirical data this way. Not today. Not through Google.
You know how many documents there are online? How many duplicates? How many quotations? The vast amount of “bullshit” indexes Google has will scare anyone. Even as sole collection of random strings, (the) Google (database) is corrupted. Yes, you can find song lyrics, old mailing list entries and correct-spellign-of-words on Google. The data you find is relevant. The metadata you (seem to) find is useless. To be discarded.
I’ve seen people win debates (xkcd) by applying this hideous method of measuring popularity. Just because Google doesn’t index the Bob’s opinion, doesn’t mean Alice is right.
The Google Search Engine is a great tool, which can used to determine location of documents, location of insecure webapps and your sick habits and fetishes, it is NOT a way of measuring and finding empirical evidence. Please, part of the internet, cut it out!
*may vary depending on your geographical location


This article is written by spyware.


What do you think?

[Kingroudy]

rahib
he's right, sarcasm is everywhere and google can't interpret things.

[Google]

Originally Posted by Kingroudy

rahib

Yes, he is. He's one of my mates at HBH.

[P337]

Very true stuff.