In previous posts, we looked at what happens when people judged strangers by their pictures. This time we looked at how people would judge strangers by their twitter feeds.
We took the top 200 twitterers from http://twitterholic.com/ and asked people on Amazon Mechanical Turk to judge how smart, interesting, trustworthy, etc. they thought they were.
Here’s a scatterplot of the top twitterers showing the smart vs. interesting values. You can see someone like Matt Cutts was thought to be smart but not entertaining, while Jimmy Fallow was judged as entertaining but not as smart.
If you mouse over a feed you can see what turkers said in the free response field.
PS Without asking permission, I subjected some of the D.L. people to the same treatment… Chris was tagged as “rocker” and “sarcastic” (dead on). Travis was tagged as “chatty” and “sarcastic” (starting to see a pattern :) ). Steph was tagged as “productive” and “good”. Brendan was tagged as “confusing” and “busy”. The author got “dramatic” and “rambling”, which was nicer than I expected.
If you’re feeling brave and you want to be included in the next batch send a message to @doloreslabs on twitter…
And since I can’t bear to pass up an opportunity to make a classic powerlaw graph in R, here is a graph of tags and their frequencies. The x-axis is subsampled substantially.:
(Shout out to Travis for the idea of this post.)
Update
- In a late-night fit of inspiration, Chris made the above chart zoom-able (with the scroll-wheel) and search-able.
- For those of you that want a simpler interface to looks at the results here it is.
- To answer some of the questions in the comments: the turkers could see the username of the person twittering, so certainly that affected the results.
- If you want to be included in further experiments, please follow doloreslabs on Twitter.
- If there is interest, we could post the raw data for people to build their own visualizations, shoot us an email if you want access.
-Lukas
Mark Johnson
01/22/09
Are you using the entire twitterstream, i.e. regular tweets + @ tweets, or just the regular tweets? @ tweets tend to be more conversational. Also, do you want a direct message on Twitter or an @ reply? In any case, include @philosophygeek on your next round =)
Justin Wehr
01/22/09
DL does it again! Very cool.
Could you please show us plots of how smart, interesting, trustworthy, etc. they are versus their popularity?
Thanks much!
–JW
Benjamin Wiederkehr
01/23/09
Great analysis. Simple, understandable but none the less compelling. The request from Justin Wehr would be a nice addition and bring even deeper insight in the correlations.
Thanks,
–Benjamin
Kim
01/23/09
I chuckled to see one person I dropped as dumber and less entertaining. I couldn’t agree more, so I’ll assume the rest of these results as fairly accurate as well.
badmacktuck
01/23/09
Whoa. This is rad. I also would like to know if this included @replies. The replies can mean any number of things that they may or may not be, taken outside of the context of the original message. If they included the replies, was the turker given access to the other feed as well?
y’all are gettin’ a new follower. well done.
Dave Winer
01/23/09
Is there a non graphic readout? Can’t find my name by rollover, number 93 on the list.
Phil Wilson
01/23/09
Dave it isn’t an image, view the source of http://assets.doloreslabs.com/blog/toptwitter.html – it says you’re “serious, Average, entertaining, well-informed, well-informed”
Andy Eggers
01/23/09
Did the turkers see the twitterers’ usernames? If so, algore and stephencolbert and other celebrities were being judged not just on their tweets but on their general reputation.
Foodimentary Guy
01/23/09
Wow.
This is a neat post.
Thanks for letting me know about it.
Rudy
01/23/09
Algore is smarter? I agree he’s more entertaining. Definitely not smarter.
Dave Rutledge
01/24/09
I’m bummed that we didn’t hit the bottom-left corner. Seriously, someone thinks the Woot Twitter feed is ‘smart’ or ‘entertaining’? Probably the same person who commented “Boot”.
Diana
01/27/09
as with most surveys the value will come from the way the group filling the survey out was chosen. If I can give an example: a younger person might find an older person smarter because they do not understand the subject the person tweets about as well. They may also find another person of similar age group more entertaining than someone in another demographic because of shared interests and life experience.
These are minor things though, as what we are talking about here. i guess, is perceptions. i find it quite interesting to see how twitter is attracting so much of this type of study and how these types of studies are influencing the behaviour of the people using twitter who relate back to them.
I find the whole culture quite fascinating.
thanks for the post!
Real Estate Space
01/27/09
While Borat may be one of the dumbest twitterers of the lot, he is way more entertaining than most of those posted here IMHO (yet placed in the bottom left); e.g. starbucks? wordpress? come one, when is the last time those guys made you laugh?
Seriously, great post!
Peter Skomoroch
02/04/09
Great stuff guys. Did you do the scatter plot straight up in javascript & html or use some templating / plotting toolkit?
Chris Van Pelt
05/12/09
Hey Peter, we got really ambitious on an idle monday night and made the sweetest interactive scatter plot imaginable with a little help from the http://mootools.net framework.
Razik
06/10/09
Google Webmaster Youtube channel mentioned about this page :)
http://www.youtube.com/watch?v=2QIxTI59r5o
Austin Belts
11/02/11
Thank you, I have just been searching for info approximately this subject for a while and yours is the best I have found out so far. However, what about the bottom line? Are you certain concerning the supply?