I have been sitting on this topic for over a year. I finally am breaking down. After the CES show, I don’t see anything better on the market right now, but this idea for operating system–level notification modeling from Shruti Gandhi has given be enough hope to propose some ideas.
One answer could be consolidation. Snowball is almost headed in the right direction. Snowball consolidates all your alerts in one place.
The challenge is notifications. In theory, we want to be notified when something happened, is happening, or will happen. Sometimes we need to be directly involved (a pending meeting). Sometimes we just want to know (a score in a football game). Sometimes it is critical (a loved one in a car crash). Sometimes it is opportunistic (walking by a store with a product on sale that we have been shopping for). All of these are useful on their own. But when you put them all together, they can be overwhelming.
There are several human factors-related issues that arise. One is in the design of the notifications. Katherine Bierce has some great advice on this. She recommends:
- Sending notifications only when the information is really necessary.
- Say only what you need to say – to reduce the attention and cognition needed to process it.
- Include options of what the user needs to do or what doesn’t need to be done immediately.
- Allow restricting notifications to a daily digest.
- Presenting each notification at a location on the screen that is customized based on what is on-screen at the time.
- Allow users to customize the content and timing, which device each one is sent to, and the salience of each one when it arrives.
- Use context-awareness algorithms using the user’s calendar, GPS location, open apps, and active tasks to determine what and when to present them.
Using another tactic, LEDmeKnow is a rubix cube looking device that allows the user to keep his or her phone in their pocket. It has color coded squares that tell you what notifications are arriving. This seems like a very poor solution to me because it doesn’t have any of the design features recommended by Bierce.
Another human-factors related issue is the problem of ego-depletion. Even when the user can suppress the need to check a notification, it depletes some of his will-power. After a few of these, it is only a matter of time before he succumbs. There are just too many to resist.
A third human-factors issue is one that I have personally been involved in. There are many motivations that compel us to check our notifications. Often it is social. If a notification might be coming from a member of one of our in-groups, we have a tribal need to see it. Perhaps they need us, making us feel powerful. Or they want to work with us, evoking our sense of collaboration. Perhaps they like us, making us feel a stronger sense of tribal belonging. Perhaps it is a challenge, evoking our sense of competition.
Another set of motivators is extrinsic. The notification might be something of tangible value. A sale that will save us money. A work project that will make us money or advance our career. A financial warning that will save our 401(k) or a tip that will build it up.
Each of these motivations has a different kind of pull. A different priority. A different emotional dimension that may impact how, when, or where we want to receive the notification.
So that is why I find Shruti’s idea to be compelling. Having the notification system driven at the operating system level gives it a lot more information. It can add long term personality and behavioral profiling. It can add short and medium term activity tracking. It can endure through device and app upgrades.
But that would also require giving up a lot of privacy. This is manageable if we use a local profiling algorithm. We don’t have to worry about government interference if all of the user data is stored locally and only the algorithms are managed in the cloud. But would the typical user have the confidence that this is the case after the series of Wikileaks, Snowden, and even the Sony hack by North Korea or whoever turns out to be behind it. Most of us, including me, are completely clueless about how much various government agencies know about us. I am a fan of the show “Person of Interest” where they know it all. Who knows how closely that tracks reality?
But for a truly smart notification system, I might make some tradeoffs. But I would rather have a privacy infrastructure that gives me the best of both worlds.
How about you? Are you a notification Zombie? How much of your privacy would you be willing to give up to alleviate it? Would you be willing to allocate some of your tax money or research budget to the development of an autonomous privacy system that combines the local profile with global algorithms?
Image credit: Thrifty Look