Rhetorica: Press-Politics Journal

The Firefox plug-in called SpinSpotter was just introduced this week, and it’s already creating a stir. Can software “de-spin the news, expose the slant and bias, separate the facts from axe-grinding opinion”?

I think the answer is clearly no. At least software cannot do such a thing until someone figures out a way to make software think–in this case, make software understand meaning and intention.

I just watched a presentation on SpinSpotter by founder Todd Herman at a “technology summit” that is part of the Missouri School of Journalism Centennial. There will be Q&A time later today. I plan to ask Herman about the specific criticisms made by the writers of Language Log.

For example, Mark Lieberman wrote:

This might be an unusual type of demoware …, one that is released for general use in the hope that enough people will submit their proposed spin-spots to give the company enough free training data to actually develop some of the technology that they pretended to have in the first place.

It turns out he is correct. One of the first things Herman said this morning was that SpinSpotter is “very very beta” and “we want people to teach the technology.” His video (not live) demonstration clearly showed that the software really can’t do anything until people start using it, i.e. start flagging spin themselves.

What is not clear at all is what SpinSpotter will do with the data it collects. It looks like just a glorified word/phrase flagger. What a word/phrase means in the context of a particular article is something no software can determine. But the social networking aspect of the software is being used to convince users that it can be determined.

Users mark news articles and readers of the same articles will apparently be able to use the input of other users in real time. For example, user A says story X contains spin point Y so other readers of the same article will see spin point Y flagged. What happens when reader B reads story Z with spin point Y used in a different context?

As I said before, flagging words/phrases can certainly be useful to the extent that flagging helps one find what one is looking for. That’s just the first step. The crucial step is the act of interpretation. That requires a human mind.

The trouble here is that Herman is making claims about what SpinSpotter can do that are clearly not possible. Consider this example from the rules of spin (in this case the rule of “Reporter’s Voice”) offered to guide users’ flagging:

The reporter employs language (in the form of adjectives, adverbs, verbs, or superlatives) that conveys meaning beyond the supporting evidence provided in the article, and begs the question: In who’s opinion and by what objective standard?

Flagging words is one thing. No software, however, can determine if the “evidence” supports the use of a particular adjective. Software can’t do this because software doesn’t know…anything.

SpinSpotter is really social-networking / cooperating software. To the extent that users flag and interpret the same article in real time it may be entertaining if not actually useful in combating spin.

UPDATE: I guess it’s official. There is no spin to be found anywhere in the opinion section of The New York Times! I’m “testing” SpinSpotter for myself. And, just as Mark Lieberman said, it doesn’t do anything. No spin will be spotted until users spot the spin first. If users can already spot the spin for themselves, then what need do they have for SpinSpotter?

UPDATE: My day did not work out as planned. I was unable to attend the Q&A session.

UPDATE: After more playtime with the software, and actually finding tags on a story about SpinSpotter, it is now (somewhat) clear to me that the SpinSpotter folks simply need to do a better job of explaining what the software does and how the software works. From what I can tell at this point, it only works on (and is only intended to work on) articles that have been flagged by a user. All this talk of algorithms makes it sound like the software is going to be making the judgments. I’m not at all sure this is what the SpinSpotter folks intend to convey.

Here’s an example statement from the SpinSpotter: “The neat thing about the adaptive SpinSpotter technology is the ability to filter and identify the presence of spin in any news article, web site, press release, or thinly disguised political talk sheet.” This seems to suggest that the software is going to do something I believe is impossible. Herman needs to take his own advice (from the rules of spin) and be clear about agency: It ain’t the software that’s going to “filter and identify the presence of spin in any news article.” The users are going to do it with the help of the software.

Tags: journalism, rhetoric, politics