The Dopamine API finally gives developers a tool to reliably and scalably increase retention and engagement through behavioral reinforcement techniques. It takes the guesswork out of so much of UX design that dictates an application's growth and staying power.
Obviously, I don't think that our product is subversive. I think that there is a moral imperative for our product. A huge percent of the world's human suffering stems from peoples' inability to do the 'simple' behaviors they wish they could (eat less, go to the gym more, study, relax when they get home from work).
We’re not stupid: app-based approaches are not silver bullets. But if we’re going to help people fix some of their major lifestyle-based problems we need to do so with a rigourized technology of human behavior. Less sophisticated versions of what we do are already being used - and mostly by gaming companies. We’re bringing the most powerful technologies for changing human behavior to bare on some of the biggest problems there are.
For example, medical interventions, (i.e. curative care) involving doctors, specialized clinics and other health professionals is very expensive and will only become more so in the future. Preventative care - especially through mobile devices and apps - is extremely important if we want to lower the cost of health care over the long term. The most effective approach to prevention is building better, healthier habits. And apps scale stupidly well.
We only make production keys available to teams that we're sure are working in their users' best interests.
He shows that all the same technologies that the characters of BNW use to chain people to the state, can also be used to set them free. Habits can be your best friend or your worst enemy. We can't unlearn what we know about how to change human behavior. But we can decide that while casinos and gaming companies are using these insights to numb people, we're going to use the same insights to help people become their best selves.
We only make production keys available to teams that we're sure are working in their users' best interests.
Thank you for taking the time to reply, I do hope you'll keep in mind the ominous aspect your product draws on. I wish you and your team all the best, but be careful where you go with this please.
Also my concern isn't about habits directly, more that you're selling the manipulation of those habits in an automated, generalisable way.
How did you figure out the optimal reinforcement schedule for behaviors? That doesn't seem like something you can just brute force with number crunching.
Right you are! This can't be bruit forced. We use brain wiring model and behavioral models to figure out when/how to reward users. We also run a lot of experiments when we have hunches about ways to improve the system.
Looks neat, and we might give you guys a try. I'm wondering about a couple of things that weren't super-obvious to me from the website and video alone:
* does it work with a web-app? I saw curl examples, so probably it does, but what about those reinforcements?
* speaking of which - apart from two simple popups on the video, is there a "library" of cool reinforcements to use?
* not trying to underestimate the science behind this, and just playing devil's advocate here - did you test the actual difference between your algorithm and a semi-random/ultra-basic reinforcement trigger? (something along the lines of `if reinforcements_today < 3 && rand() > 0.7 then reinforce()`)
hey @gingerlime, I'm Ramsay, Founder @ Dopamine Thanks for the feedback!
1.) We're releasing our web sdks next week: A PHP version and a Python version, each with a JS frontend wrapper. If you email me at ramsay@usedopamine.com I can poke you when it's shipped!
2.) We do! To get dev's off the ground we released a library of popups with different icon images like you saw. You can modify parameters like color, duration, copy.
AND you can arbitrarily call any UX-updating function from the code that handles the Dopamine API's response. This lets you map a response we give you to any function in your app. We're sprinting to deploy a way to do this more elegantly on the integration dashboard also.
3.) Yes, we did. TL;DR: We outperform alternatives that devs are resorting to right now. @tdaltonc: care to chime in?
Thanks for answering all my questions. I'll ping you guys and see about testing this out. We created a learning platform for anatomy[0], and gamification and motivation is one of the things we know we can and should improve. Sounds like dopamine can be a good fit.
On your final point: Totally reasonable question. It's one of the first ones we asked ourselves (no one wants to work on a random-number-generator-as-a-service). We've seen simple reinforcement schedules increase average engagement by 3-8%. Our optimized schedules get about 10x that and up to a 60% increase.
14 comments
[ 0.99 ms ] story [ 28.0 ms ] threadWe’re not stupid: app-based approaches are not silver bullets. But if we’re going to help people fix some of their major lifestyle-based problems we need to do so with a rigourized technology of human behavior. Less sophisticated versions of what we do are already being used - and mostly by gaming companies. We’re bringing the most powerful technologies for changing human behavior to bare on some of the biggest problems there are.
For example, medical interventions, (i.e. curative care) involving doctors, specialized clinics and other health professionals is very expensive and will only become more so in the future. Preventative care - especially through mobile devices and apps - is extremely important if we want to lower the cost of health care over the long term. The most effective approach to prevention is building better, healthier habits. And apps scale stupidly well.
We only make production keys available to teams that we're sure are working in their users' best interests.
If you're in to BNW I'd recommend Huxley's little known follow up 'Island'.
https://en.wikipedia.org/wiki/Island_(Huxley_novel)
He shows that all the same technologies that the characters of BNW use to chain people to the state, can also be used to set them free. Habits can be your best friend or your worst enemy. We can't unlearn what we know about how to change human behavior. But we can decide that while casinos and gaming companies are using these insights to numb people, we're going to use the same insights to help people become their best selves.
We only make production keys available to teams that we're sure are working in their users' best interests.
Also my concern isn't about habits directly, more that you're selling the manipulation of those habits in an automated, generalisable way.
* does it work with a web-app? I saw curl examples, so probably it does, but what about those reinforcements?
* speaking of which - apart from two simple popups on the video, is there a "library" of cool reinforcements to use?
* not trying to underestimate the science behind this, and just playing devil's advocate here - did you test the actual difference between your algorithm and a semi-random/ultra-basic reinforcement trigger? (something along the lines of `if reinforcements_today < 3 && rand() > 0.7 then reinforce()`)
1.) We're releasing our web sdks next week: A PHP version and a Python version, each with a JS frontend wrapper. If you email me at ramsay@usedopamine.com I can poke you when it's shipped!
2.) We do! To get dev's off the ground we released a library of popups with different icon images like you saw. You can modify parameters like color, duration, copy.
AND you can arbitrarily call any UX-updating function from the code that handles the Dopamine API's response. This lets you map a response we give you to any function in your app. We're sprinting to deploy a way to do this more elegantly on the integration dashboard also.
3.) Yes, we did. TL;DR: We outperform alternatives that devs are resorting to right now. @tdaltonc: care to chime in?
p.s. I really like the name!
[0] https://www.kenhub.com