The Eatery: Crowdsourcing Evaluation of Food Diaries

3 Nov

Previously, I suggested that other people can be a good source of feedback, among other things, to support self-regulation. Right on queue, two days ago The Eatery was released, the first iPhone app from MassiveHealth. The two biggest differences from other food logging applications out there are:

  1. You just take a photo of what you are about to eat; you don’t try to semantically tag it in a way that allows for calorie counting or ingredient analysis.
  2. Other people rate what you eat on a 10-point fit-or-fat scale.
The purpose, like other logging applications, is self-regulation. But the strategy is to support mindfulness, not to provide feedback on progress toward goals. It seems that their inspiration is that if most of the benefit of food diaries comes from recording what you eat, and not from tallying up the calories, why make the interface slow and cumbersome, as it would have to be to estimate calories.
The numeric ratings attached to photos still allow for some aggregation and tracking of progress over time. It generates a graph of average rating by day, which would presumably get more interesting if you starting seeing patterns, like Fridays always being a bad day. (Another app, called MakeIt10, skips the photos and just does ratings. Its website tagline is, “Counting Calories Sucks.” It only does self-rating, not crowd-sourcing.)
Now, why is it that ratings from other people are needed at all? Why not just self-rate? Presumably the reason is to avoid self-deception. It’s easy to talk ourselves into thinking that what we’re eating is OK, but others who have less vested interest in this dessert being healthy because it uses fruit juice instead of sugar for sweetening will call a spade a spade.
I have yet to use the “friends” feature to integrate my Facebook friends. I’m going to try that, since reviews on the App Store suggest there are more features that will unlock. But our previous research suggests that your overall Facebook network is not a group that one would want to participate with in an App like this, though perhaps other friends using the app might be a reasonable group to share with.
Right now, I am just getting the wisdom of the anonymous crowd right now. I suspect they may be using some recommender system techniques to match people with the right raters. At least they have the data that they could use to do so: after you snap a photo, you rate it yourself. And then they prompt to rate some other people’s photos. They could do some pretty quick calibration to figure out whose ratings you tend to agree with when you rate your own and others’ photos, and then ask those people to rate yours.
I probably won’t use thisapp  regularly, because improving my eating is not high on my self-improvement list right now. But here are some quick reactions from using it for a couple days:
  • Even during the excitement of startup phase, I haven’t remembered to take photos of everything I’ve eaten
  • It’s kind of fun to rate other people’s photos. I found myself rating four or five each time I got started and then having to consciously decide to stop. I have received about 20 ratings for each of my four photos, suggesting that other people also find it more fun to rate than to log (or else they’ve hired a bunch of people to rate in the first few days (Mechanical Turk?), to get things rolling.) Just rating others’ photos may be a useful food mindfulness exercise. It would be an interesting experiment to test the impacts of an intervention where people participate as raters for a few minutes every day, and don’t do any food logging for themselves.
  • You have to take the ratings with a grain of salt. I posted two photos of the same food, once when I had it for dinner and once when I had it for leftovers. I got overwhelmingly positive ratings once, mostly negative the other time. Admittedly, I wrote the ingredients once and took the picture before eating, so it looked appetizing, while the other time I didn’t write ingredients and took the picture after eating most of it (oops, it looked kind of gross, apologies to my raters). Ratings for a Yoplait lite yogurt were overwhelmingly positive, and I don’t think it’s actually very healthy. One commenter on the TechCrunch blog post about the app complained that he wanted expert evaluations, not the wisdom of his friends or the crowd (or even people who rate the same way I do, though he didn’t say that). But I think the crowd doesn’t have to be perfectly correct for the mindfulness aspect to work, or for the retrospective assessment of eating patterns to work.

One Response to “The Eatery: Crowdsourcing Evaluation of Food Diaries”

  1. Terry Ruth VanDuyn November 4, 2011 at 9:49 am #

    Great post!

    So far, I’ve only been using The Eatery for a few days, but I already have a lot of opinions.

    Similar to what you said, I think one of the major advantages of the app is providing accountability. So far, I’ve been more reluctant to grab a handful of candy from the jar at work since I would have to provide photographic evidence. On the other hand, I could see this leading to selective or biased posting (sure I’ll take a picture of the salad, but not the huge slice of cake I ate afterward), which might defeat the purpose.

    As much as I like the idea of a social aspect (even though only two or so of my friends are using it), it still feels a bit invasive. I got a notification yesterday about what my husband ate for dinner while out with his friends. It was kind of disconcerting and it felt like a slight violation of privacy. There’s also no real way to provide encouragement to my friends. Besides rating or publicly commenting on their foods, I haven’t found a way to acknowledge positive choices or give private feedback.

    I realize the point of the app isn’t necessarily to provide feedback or support progress towards goals, but it feels like a missed opportunity. This might just be my public health background talking, but I’m always seeking nudges toward positive behavior change. I think it’d be great to set goals and indicate if I’ve reached them. For example, I want to eat vegetables at least two of my meals. At the end of the day or week, I could be prompted to review the photos I took and see if I made progress toward my goal. Frankly, crowd ratings don’t seem like a good enough payoff to “snap” everything I eat. Without some feedback or goal tracking, I’m not sure I’d continue using the app long term.

    I would really like reminders. If I could tell it (or it could learn) that I always eat breakfast at 7:30 and I always eat lunch at 12:00, I wouldn’t mind notifications to prompt me to record the meal. Maybe I could even set additional reminders to have a snack at 3pm when I start getting cranky.

    Some of my other frustrations are that I find myself compelled to add long, in-depth descriptions to prove my food is healthy (e.g., whole wheat English muffin, egg, fat-free no-sugar-added Greek yogurt). This takes much more time, but I know I’ll get annoyed if I chose something healthy and it isn’t recognized as such.

    I also wish the feedback was faster. I’m still getting ratings on posts over a day old, but it takes several hours for new food to get rated. If the turn-around was in minutes rather than hours, it might give me an opportunity to chose something different or skip the side of fries.

    I’m still unclear how location data is used, so I haven’t been willing to share it. I’m not sure if it can be seen by my friends, by everyone, if it’s private, or if I can control it. It’s too bad too, because I think this data would provide a lot of insights.

    I’ll keep using it for the time being. And I’ll be especially interested to see new updates and features they add.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: