The ethics of weight and body fat on Twitter

Update 12-16-2010:

I have just purchased a Withings scale. Indeed, as I discussed in my original article, it will only let me twitter my weight instead of my body fat. How frustrating. But, you can customize the static portion of each tweet. I will be tweeting my weight with a bit.ly link ( http://bit.ly/gS05Nz )to this article and @withings in each post.  Eventually, they will get tired of this and reply to my blog post or otherwise acknowledge this issue.

They also ask for my twitter password, which is no longer the right way to authenticate…

I originally blogged this while hanging out with Alan from videnitity.com ( originally met at health 2.0 ) at an mhealth conference. I should have mentioned videntity before.. they do cool work with Django and python for life streaming. In fact the discussion that I am about to go over prompted him to cover, in detail, some of the work he did using the Wii balance board to lifestream your weight. Hardware hacking for healthcare. Pretty cool stuff.

Almost immediately on seeing Alan again, our conversation turned to our mutual interest in quantified self. We are both interested wifi enabled scales.

But almost immediately I stumbled upon a mutual frustration. The most popular and well-known wifi scale in the space is the withings scale. The withings scale measures both weight and body fat percentage (impedance method). We were both frustrated with the default ability for the device to post to Twitter.

Take a look at the search for http://withings.com on twitter. Notice anything? Thats right folks… the withings scale posts weight. Just weight.

I do not need to bother to tell my readership ( clinicians and health interested IT folks) the problem with this. Our culture continues to have an obsession with weight at runs contrary to health. My own life is a great example. I weigh about 270 pounds. I am overweight, but my relatively high level of physical activity ensures that much of my weight is muscle. If I could manage to lose 10 pounds of fat and gain 10 pounds of muscle it would be better for me long term (as my muscle helped accelerate further weight lose) than just losing 10 pounds of fat. There is some debate about whether BMI or body composition is a better measure, but pretty much everyone agrees that thinking in terms of either BMI or body composition are vastly superior to simple weight measurements, because both take height into account (explicitly or implicitly).

So here we have the basic ethical quandary.

1. Focusing on weight contributes to an unhealthy obsession with a single number that cripples our ability to compare two people effectively. This obsession can merely be inconvenient for people like me, who are capable of seeing past the number but are frustrated that I have to constantly do that work… Or it can be dangerous for those with certain eating disorders.
2. The withings scale is capable of publishing both the body fat percentage/BMI and/or weight to Twitter providing social pressure for those who seek to manage their health.
3. The withings scale chooses to publish only weight to Twitter.
4. To add insult to injury, salt to the wound, spittle to the slap, and gratuitous cliches to the sentence: Withings knows that its users are requesting body fat percent in the tweet stream, they know they are requesting full templating of tweets, and they are taking a poll on what users want, but the poll does not have full templating (which would allow tweeting of body fat) or simple body fat tweeting as poll options.

Withings is playing “the users are too stupid” card on this one. From the comments of the blog post in question, in response to Paul who suggested a templating system:

Hi Paul, we first wanted to do so but as I said in the post above, we noticed that fully customizable tweets are too confusing for non computer-skilled users.

Moreover, a fully cuztomizable tweet can lead to unrelevant tweets.
Lets say youre trying to lose weight and your tweet is my current weight is %wc% and I only have %wo% to lose to reach my objective.
If ever you pass your objective on a specific weigh-in, the tweet will be unrelevant (it will say you still have -2 lbs to lose for instance) unless you think of changing the default tweet message just before weighing in…

Oh thank you pointy haired bosses at Withings. You have answered a tremendous consumer demand with a technology that further perpetuates fundamental healthcare mythologies. You have given us something at is 95% similar to what we need, but now serves a subtle destructive force instead of what you could have done which would be to apply social pressure to the right problem…and now as your user community suggests methods to allow us to fix your mistake… You tell us that we are not smart enough to do handle the tools we need to do that… Oh thank you from saving ourselves from ourselves!

This is the fundamental problem with Health IT today: we as health IT programmers are constantly making very subtle ethical decisions, and we regularly flub them up. More importantly, we ignore our users when they urge us to fix ourselves…

Withings made an ethical mistake in having the default tweetable data weight instead of body fat. Their UX justifications would be valid if they had made the right ethical decision to start. If I could only get body fat percent… That would be a frustration, but I could deal… As it stands they are just wrong.

But I understand why. Imagine the meeting:

Developer: we should integrate with Twitter
Pointy haired boss: cool do it.
Developer: we do not have time to do a complex integration. We can use weight, body fat percent or BMI…
Pointy haired boss: no one understands body fat percentage or BMI. This is a weight scale. People understand that… Lets go with that….
Developer: but body fat percentage….
Pointy haired boss: would not be as popular. This is a marketing and time question. We only have time for one of the two and most people will want weight…

The pointy haired boss is right. The fact is that people do understand weight better. Withings has probably sold more scales by choosing just to only tweet weight. This is not just a principle vs. profit issue. Assuming scales make a difference for people, more scales equals more difference. Frankly I do not blame Withings for starting with this decision. They were wrong, but their heart was in the right place.

But the arrogant position on templating systems and the fact that the poll they setup on the issues did not even give an option to chose body fat percent as an option is pretty unacceptable.

The other problem that Alan pointed out is that to get at the Withings scale data, you currently had to integrate with the Withings server rather than with the scale itself. So you have a device that records data about you, and then to get at that data, you need the by-your-leave of the device manufacturer… Not OK.

This device should be a “home health appliance”, and it should integrate with the Withings server -as a convenience- not as a requirement. I should be able to point this system anywhere I like. I should be able to easily point the data that the device generates anywhere I want to.

Withings need to take the following steps, and soon.

Support templating as a fundamental method of data export.
Support sending that templated data to any Internet location via simple POST, without going through Withings servers
Support full XML and JSON data export with those posts
Make body fat percentage the default measure sent to Twitter and give users the option to change it back to weight

If you support this idea… Please leave me a comment here, and retweet this article, with the @withings tag. I am pretty sure withings watches twitter.

-ft