Obesity is a contributor to many different negative health outcomes. The simplest way to prevent obesity is to eat less. Unfortunately, it is inconvenient to track how much you are eating. A cheap device that passively tracks how often you snack could help the world at scale.

The hardest, riskiest part of this project is writing an algorithm that satisfies the following criteria.
1. Can run all day on a microcontroller.
2. Can learn to identify new gestures from a small quantity of data.
3. Can be sold cheaply.

I built such a device. As proof, here is a video of the device in action. The streaming numbers are acceleration, gyroscope and two intermediate features the device calculates.

 

I have completed the hardest, riskiest parts of this project. The only part left (besides retraining the algorithm with a wider variety of data) is "make a bracelet and sell it"—which is something I have already done. (I did it with a team of three people and a total budget of $96,000.)

I shut down the food tracking project because venture capital fundraising was pushing the project in a profit-oriented direction I didn't like. It has more recently been drawn to my attention that Effective Altruism (or a sister organization) might be able to fund it with a grant instead. I would happily resurrect the project tomorrow if I could get proper funding.

I have written a lot about rationality but I don't know much about the EA ecosystem. Is my project appropriate for EA? If so, how do I go about applying for funding.


The best way of contacting me is email.

5 comments, sorted by Click to highlight new comments since: Today at 12:27 AM
New Comment

Some questions I would have if I was an EA grantmaker:

  • Is this really super-scalable?  How many people would buy a dedicated gesture-detecting device?  Would it be better to write software for a device like a Fitbit or Apple Watch, which millions of people already own?
  • Wouldn't people learn to ignore the notifications over time?  If I put a post-it note on my fridge saying "stop snacking!", that might cause me to think twice a few times, but eventually I might just start ignoring the post-it.
  • Even if wearing the device was 100% effective at eliminating unconscious snacking, would this make a dent in obesity?  Wouldn't people just get hungrier and then eat more at meals?  The path between "use your willpower to snack a bit less" and "actually lose weight and keep it off" is absolutely notorious for being convoluted, impenetrable, and largely uncharted by modern scientific understanding.  My prior on proposed obesity interventions actually working is very low.

Is this really super-scalable?

Yes. We could sell these things for <$50.

How many people would buy a dedicated gesture-detecting device?

More than 1/100 people. We have done market research.

Would it be better to write software for a device like a Fitbit or Apple Watch, which millions of people already own?

Surprisingly, the answer is no. The Apple Watch (last time we checked) didn't support this kind of software. But more importantly, most of our potential customers own no smartwatches or Fitbits at all. Also, the smartwrist ecosystem is fragmented so we'd have to write new apps for every different model and update them whenever the manufacturer changes something.

In our trichotillomania project, we first tried putting software on others' devices. It was a disaster. When we started making our own wearables everything just worked.

Wouldn't people learn to ignore the notifications over time? If I put a post-it note on my fridge saying "stop snacking!", that might cause me to think twice a few times, but eventually I might just start ignoring the post-it.

Our experience talking to people and selling similar devices for trichotillomania suggests the answer is close enough to "no" to make our device worthwhile.

Even if wearing the device was 100% effective at eliminating unconscious snacking, would this make a dent in obesity? Wouldn't people just get hungrier and then eat more at meals?

The device also counts how many bites you take during a meal. Users would have to change what they eat or how much they eat per bite in order to hack Snackwatch's metric.

proper funding

Can you give a range? Also, how much would you be happy to sell a prototype for?

how do I go about applying for funding

I would say, come up with a Fermi estimate of where the value proposition is coming from, e.g. from preventing obesity, from being a good investment, etc. Then apply to the either the relevant EA Fund or to the Future Fund

I would also be curious if you can come up with a project you'd be more excited to lead.