Comments on: Putting Cool Ahead of Science: TweetPsych

By: girl

girl — Wed, 25 Nov 2009 04:17:41 +0000

I’m not sure why anyone would take this more seriously than those online personality tests, horoscope predictions, career assessments, etc.

By: DrDNickelson

DrDNickelson — Tue, 24 Nov 2009 20:45:21 +0000

John, as always an insightful and thoughtful analysis. My take is that some of the claims — and in particular the name — are less than ideal; I imagine they are made and the name chosen specifically because of the valence the term psychology has in our culture. Putting those things aside (and Dan, I think you could and still be successful), what I like about the tool (and a few others, like Klout) is that they are stepping away from the old ways of measuring things. That has it’s drawbacks, but so does having to defend a web-log generated measure that one could argue has even less direct correlation to the kinds of behaviors that seem to be populating the 2.0 space. So Dan, a little more transparency and maybe a little less trading on psychology as your anchor point, but keep trying to push the envelope re: measurement of online social behavior.

By: My concerns about TweetPsych | The Tyler Hayes

My concerns about TweetPsych | The Tyler Hayes — Mon, 13 Jul 2009 04:47:36 +0000

[...] TweetPsych. If you’re looking for real medical commentary & opinion, check it out: “Putting Cool Ahead of Science: TweetPsych.” I highly recommend [...]

By: Naru

Naru — Wed, 24 Jun 2009 14:52:22 +0000

I get annoyed when people use the word psychology as something anyone can do. No. It is a four year plus degree at university and even then you need more training. No random person can just randomly say ‘I’m going to psychoanalyse you and tell you your personality’. It doesn’t work like that.

By: Daniel R.

Daniel R. — Wed, 24 Jun 2009 13:58:09 +0000

Wow, a real psychologist versus one of a million web developers who are running a “business” out of their parents’ basement. Professionals don’t understand the internet, or kids these days (and who are so wrapped up in themselves and their media-gluttony-induced abstractions of the world that they who don’t give a damn about the former generation’s idea of protocol). Kids these days, especially web developers, are trying to strike it rich with silly little programs, but for every Twitter, there are a thousand emulators, and for every Twitter aggregation service, there are a thousand web developers who should be working at Burger King instead of spewing their code all over the internet.

Final take: Professionals don’t understand that young people don’t take anything seriously. (Come on, this is the generation AFTER Generation X. The post-nihilists.) And the liberating, wonderful, decentralized nature of the internet lets every other hack with a downloaded copy of Photoshop set up a “service” to get his fifteen minutes of fame.

By: Jessica

Jessica — Mon, 22 Jun 2009 18:49:48 +0000

Really great dialogue about this new tool. John, you bring up some excellent points and criticisms about TweetPsych. It is good to be skeptical. It will be interesting to see where this goes.

By: Have You TweetPsyched Yourself Today? | Celebrity Psychings

Have You TweetPsyched Yourself Today? | Celebrity Psychings — Fri, 19 Jun 2009 15:45:08 +0000

[...] it turns out, TweetPsych goes a little bit beyond “for entertainment purposes only”; according to Psych Central’s Dr. John Grohol (whose main points are below in bold), [...]

By: John M Grohol PsyD

John M Grohol PsyD — Fri, 19 Jun 2009 12:36:45 +0000

PS – Let me again emphasize the valuable contribution Zarrella has made with the introduction of this service. Despite its current limitations, he has shown the potential of what could be done with this sort of analysis, and that’s a very important step.

Dan, I’d love to see you release the analysis engine component itself as open source. You could open the door to other researchers conducting their own analyses on the tweetstream.

By: John M Grohol PsyD

John M Grohol PsyD — Fri, 19 Jun 2009 12:21:53 +0000

@Dan – As I said, I wouldn’t have mentioned your background except that you’re making specific scientific claims regarding the service. So one would expect such claims be based in the scientific method. It’s not my definition, it’s the definition that’s widely accepted within society. (Like social networking itself and its rebranding of “friends,” you’re welcomed to redefine words as you like. But then don’t act surprised when others ask, “So what’s your scientific training and background? Published anything in a journal?”)

What “baseline”? From all users who try the service? From some other population? From a random sampling of “normal” twitter users? Could you describe the specific methodology a little more, because I’m not sure I understand what you’re referring to here.

I think it’s fine to keep refining the analysis engine and what-not as you’re in “beta,” but wouldn’t it be nice to let your users know that you’re the one actually manipulating their scores at this point — that it’s not a result of their additional tweets or anything. A content analysis service is a lot less meaningful if someone is constantly tweaking the analysis algorithm which can result in such significant changes in one day.

Last, I didn’t add this to the post because it wasn’t central to my analysis, but I noticed you don’t have a privacy policy on the site either. Are you keeping a copy of every analysis done? And if so, are you using said data (or plan to use said data) for other marketing purposes? Such a database could be invaluable to a marketing company…

By: Dan

Dan — Fri, 19 Jun 2009 02:17:34 +0000

John:
Ad hominems aside (my apologies for not conforming to your defintion of what a scientist should be) I appreciate the time you’ve put in analyzing tweetpsych, and I apologize for not responding right away to your offers to help (I’ve been busy trying to make the app scale to the traffic its been getting).

A few points of clarification. Tweetpsych compares individual user’s results with a baseline and displays the variation of each user from that baseline, so naturaly the output of LIWC2007 would not be the same.

Scores are changing because I’ve made changes to how the baseline is calculated as well as some minor changes to improve how I was applying the two dictionaries.

I completely agree that the service is a “freshman” effort because its beta and I plan on evolving it. I started to add some descriptions for each of the codes, and will continue to do so as I make the system stable at high traffic levels.

I do plan to follow up with you on your offer to help make the output more readable.

Thank you again for your thoughts, critical attention is invaluable to improving the service.

By: Tyler Hayes

Tyler Hayes — Fri, 19 Jun 2009 00:12:10 +0000

Thanks for the credit at the end John!

Your analysis & commentary is way more in-depth than mine, and I wish I could add something beneficial to it. But, you nailed it! Spot on good sir.