The Need for Human Driven Development

We are wasting our time and energy by not understanding the high variable cost we pay pay daily to technology that could be avoided by learning these skills ourselves instead of offloading them to “apps”.

Human technology optimizes itself to teach the user a skill instead of just providing answers. The user of human technology becomes less reliant with use.

There is an app for that. This mantra is becoming more and more detrimental to the general health of society as we choose to offload more and more tasks to apps. The best example of this is the widespread adoption of GPS. Now that GPS is not only in your car but also in your pocket with Google Maps, Waze, or Apple Maps people have died choosing to rely on their phone then rather than commonsense. Instead of acting as a supplementary guide in an unknown city we have stopped learning the basic skills of navigation, location awareness, and are fully dependent on GPS to get around even in our own towns.

These applications also do not interact and share information in a human manner. If someone asked you directions to the nearest gas station you wouldn’t give them a turn by turn direction list with street names and miles to the 2nd decimal place. No, you would tell them to “head west on main, you’ll see a Mcdonalds, its another 3 blocks on the right”. With out even realizing it we’ve compressed the 1 page long 12 turn directions into a single sentence. We need to develop technology that communicates not only with a realistic sound (Alexa/Siri) but in a human manner.

“Uber driver - Do you now a good route or should I follow GPS?” “You - Oh….just follow GPS..”

Another great example is spelling and spellcheck. This tool introduced at a young age along with the new mediums of texting, twitter, and emojis has made learning how to spell a lost art. People will choose to type the same word wrong everyday and right click to correct than just learning the correct spelling. This phenomenon is called temporal discounting. Roughly, we are predisposed to highly value pleasure today (GPS/Spellchecker) and to deeply discount future pain, especially the more distant it is[^1].

This isn’t the highway! Apple Maps beta tester

Current Incentives

Today, most companies rely on ad revenue. Creating huge incentives to increase app usage and clicks. This creates a cycle that pushes design patterns to make the user “addicted” to the application and not tied to positive outcomes for the user. I believe that this model is not sustainable and more human model needs to be developed.

But Danny technology is amazing! It gives you super powers! They’re called Smart Phones for a reason. /s

Some technology is spectacular, but some is garbage. For instance, even with cloud synced multitouch mobile apps people are still as disorganized as ever. Recently, I’ve switched to a good old analog notepad to keep track of my tasks and schedule and I’ve seen great improvements not only in my efficiency but in my own wellbeing. Taking away all the “Fluffy Tech” of push notifications, crazy transitions, and multiple accounts I’m able to get to the heart of the issue of organizing my thoughts into coherent action. It has allowed me to flex my mental muscles and improve my organization and time managment skills.

As developers we create applications with the assumption that our user is dumb as dirt. That might turn out to be a self fulfilling prophecy

We under estimate the extreme efficiency of the human brain. As developers we create applications with the assumption that our user is dumb as dirt. That might turn out to be a self fulfilling prophecy.

Machine Teaching

While Machine Learning and Deep Neural Networks are the hot buzzwords of today I believe that we must push towards Machine Teaching to maximize our human/technology interactions in the future.

Machine learning is the process of creating a mathematical model by that uses a training dataset to “learn” the optimal model to map inputs to a output dataset. Machine teaching is the inverse of this process. You start with the optimal output dataset (answers dataset) and you know the users learning algorithm. Machine teaching will derive the optimal training dataset for the user. Through the users interactions with the application the application will be able to adapt itself to optimize itself for that specific user[^2].

You could imagine an application that learns how you are interacting with it and dynamically changing the interface, showing and hiding features to optimize your interactions with it. Allowing the user to provide feedback and request less or more detailed information. For example, a confident driver in their hometown might only require the cross streets of the address and a POI nearby only getting the full turn by turn if requested.

Conclusions

As a society, we recognize the need for our children to be good at math and go into STEM, but we are failing ourselves with even more basic mental skills. Why practice critical thinking or analysis when you can “just Google it”.

We need to retool our technology to make it more human friendly. Machine teaching and finding non ad driven business models while be two huge factors in this needed transition.

[1]http://www.strongtowns.org/journal/2017/1/9/the-real-reason-your-city-has-no-money

[2]http://pages.cs.wisc.edu/~jerryzhu/machineteaching/pub/MachineTeachingAAAI15.pdf