At http://whatdoestheinternetthink.net you can actually find out what the internet’s global opinion is on a given search term: positive, negative or indifferent.
The site was live in January 2009 and went on to go viral twice; June/July 2009 and April 2012 (following a re-design and re-coding).
Since launching, this has sparked some discussion as to how it all works. Well, you can understand that I don’t want to disclose too much of the ‘algorithm’ of the site. However: basically it searches based on associative (so far just English) sentences. The given search term is used in these sentences which are then sent off to the various search engines, counting the amount of results returned. (Sentences are double quoted before they are sent off, so as to make sure the search-engines search for occurrences of the *whole* sentence).
This, of course, produces questionable results which should not be taken very seriously. However, the more results (hits) returned, the more reliable these results can become. Do a search for George Bush and then Barack Obama, and you’ll see that the internet is certainly not far off – or perhaps even in-sync – with the result you had in mind.
Velar Trill has written a very nice article on whatdoestheinternetthink, which is spot on. I suggest you read it, as it explains more on the connotation of searches.
This is a fascinating little application that uses Microsoft’s Bing search service to analyze perception of phrases based on the context they occur in across the interwebs. It sounds pretty fascinating, but any amount of testing will immediately reveal seemingly contradictory results like “the Internet hates Hitler, but loves Adolf Hitler” or “the Internet dislikes marijuana, but loves weed.”
So is whatdoestheinternetthink.net laughably broken? Far from it! This results are actually exactly what you’d expect, and it’s pretty clear why they happen. What these two apparent errors are an example of is what my LING 201 professor called “unspeak,” a phenomenon where words that ostensibly refer to the same thing have radically different connotations.[..]
So while whatdoestheinternetthink.net isn’t that useful for figuring out what the internet thinks (although it’s dead on when it comes to cats and 4chan), it could be very useful to linguists charting the development of the English language, or other languages if it ever bothers to index them.
Articles on (or using) whatdoestheinternetthink.net:
Is this doing actual semantic sentiment analysis or is it just counting Google results?
A bit of both, actually
I’m wondering how much the bubble Bing puts you in might affect the quality of your searches? Google puts you in a bubble as well, though DuckDuckGo doesn’t. However, their crawler bots won’t return as many hits on all the random phrases, I fear.
Really now. The Internet thinks Adolf Hitler and pedophiles are the new craze.
It’s a good introduction to the topic of Social Media monitoring! Nobody’s going to take the data too seriously but it’s certainly a useful free tool.
@2: A lawsuit? Really? You do realise whatdoestheinternethink.net would win that lawsuit, right?
Some results are good, some are funny, some are just plain ridiculous.
However I think you need to finetune your search algorithm where it concerns one word searches.
I do think that lawyers from google, yahoo or bing might slap you with a mighty lawsuit when seeing the following results:
Conclusion: The internet is mainly positive on the subject of incest, according to Google.
Conclusion: Even the internet is not quite sure about rape, according to Yahoo.
Conclusion: The internet is mainly positive on the subject of torture, according to Bing.
Hi, I love the idea behind the website. I’ve been chatting about it with my friends concerning the attitude towards religion on the internet. And I think we may have found a bug.
If you search Barack Obama (I know, nothing to do with religion) you get back over 2.5 milllion results with almost all of them being “Don’t Care” and it saying that no one has an opinion about Barack Obama on Twitter (I find this hard to believe).
But if you search just Obama, it looks like more realistic results (it was like a 60:40 split between positive and negative).
Anyways, I think it is a great idea and a great discussion starter. Just thought I would point out some wonky results. Thanks again!