Wednesday, May 29, 2013

Thoughts on a cool data project

A friend on facebook asked my thoughts on SwayWhat, a new social site that aspires to be the youtube of graphs, particularly on social or political issues. It's sort of like FRED, but for non-economists and with more social media attributes. For example, anyone can upload charts with a short summary. You can "follow" and "trust" certain users (two different things that are important to distinguish when dealing with data). Ultimately the hope is that it will be an easy place to go and grab data to help you make your point. The creators are asking for feedback, and I've got one major thought but feel free to add thoughts of your own in the comments.

The creators spend a lot of effort talking about the potential problem of fabricated data being posted. That's potentially an issue, but since you are expected to source the data I doubt it'll be a big problem. A much bigger issue, I think, is presenting data that is accurate but may be misleading by virtue of its framing. Or maybe not even misleading but that unintentionally presents less than the whole story. Take this chart posted by Eric S. (it says "Obama's Economy: Recovery for the Rich"):

The text with it reads: "The post 2008 policies coming from Washington DC under the Obama administration have not helped working class Americans recover, but they have helped the top 7% grow. There's nothing un-American about the wealthy prospering, but our politicians seem to be helping only those who can fund their campaigns."

So a big problem with this is that it doesn't even say what the variable is in the summary (it does on the vertical axis) - it's net worth. I'd think the net worth of the lowest 93% has not recovered as well as the top 7% because almost all of the lowest 93%'s net worth comes from housing, and in case you live under a rock, house prices haven't done so great (recent Case-Shiller data notwithstanding - note this goes to 2011!). The wealthiest aren't doing better because Obama's Washington is pro-rich (although they may be), they're primarily doing better because their portfolios are different and this financial crisis happened to have had a lot to do with the housing market. So to me this data is framed with an extremely political summary that isn't obviously supported by the data presented. This one on the federal debt is also just awful.

I'd worry that this is going to come to dominate the site. There are two options in social media for this sort of thing: have it heavily moderated/policed or let it evolve naturally. Both have their merits but if SwayWhat opts for the latter I'd expect that you're going to have people with different political views creating echo-chambers and there will be a lot of worthless charts drowning out the good ones.

So far Eric S. seems to be the only user with problems with this and the only one that tries to offer much of an interpretation in the summary. However, if the site grows I imagine there will be many more.

The solution, I think, is to not have much in the way of interpretation in the summaries - mainly reserve that for description of the data and let readers do their own interpretation. A wiki approach is another option, but without dedicated editors that could just turn into a flame war. Perhaps the best approach is to have comments on the graphs so that people can make these sorts of criticisms and suggest something better. This may result in echo-chambers too, of course - but perhaps there's a way to deal with that. Separate comments agreeing or disagreeing? Only allow registered users who do not "trust" the author to comment? Designate registered users to post these comments but not necessarily to act as full blown editors?

I think this sort of issue could make the site pretty useless if it got out of hand, so it's important to think about.


  1. Hi Daniel,
    (Noah, founder of SwayWhat, here.)

    Great critiques and definitely items we are spending a lot of time on. Some of your concerns I expect to address in the features that will be rolling out shortly. It is worth noting that the site is only 5 weeks old. I consider us to be in alpha mode currently with beta (i.e., testing the functionality we want in place for our public launch) scheduled to begin in July. So please do be patient with our current limited feature set.

    In the next few weeks we expect to add comments and rating functions as well as pages for specific tags. It is our hope that while there may be some poorly constructed or misleading charts, the best charts will rise to the surface. Thus, users could review not just a single chart but the best 2 or 5 or 10 charts on a topic. Each chart page will also show related charts and even enable chart responses to allow users to refute erroneous or misleading data.

    Our mission is to increase the use of facts in the way people understand the world and seek to influence others. "Facts" can definitely be constructed in misleading ways. But we believe that starting with a review of different sets of facts before we start slinging rhetoric will make for better discourse and better outcomes.

    Please do let me know what you think. We are hungry for feedback and would love to welcome you and your readers to the SwayWhat community.

    Be well,
    Noah Blumenthal

    Less rhetoric. More facts.

  2. They should just call the site "Troll Bait" and be done with it.

    1. OK I have to admit - this made me laugh a lot. "Lol" isn't always literal, but in this case it was.

      Nevertheless - I HOPE this works out well. It could be really cool.

      So I am a glass half full guy but one that still appreciates glass half empty derived jokes :)

    2. I also have to admit - that was a good one, Absalon! XD


All anonymous comments will be deleted. Consistent pseudonyms are fine.