Thursday, January 06, 2011

Anticipation vs. Automation Part I: Take a hint

I recently tweeted that thanks to iOS' autocorrect feature, I was going to be renaming all of my usernames from "rstephens" to "estrogens". That's because my "smart" phone tries to change rstephens to estrogens every time I type it. It's a recent, modern problem. Our "smart" phones apparently are smarter than we are. I am constantly correcting my phone, deleting, and re-typing almost everything. In fact, the only way it appears I can efficiently enter information faster, is to intentionally misspell it, so the autocorrect function will suggest the word I really want.

We are all going to look back and realize that autocorrect was the first sign that there was something wrong with HAL.

This is just the latest hint that something is missing, and correction. I don't think one should criticize without suggesting some better alternatives, so this is the first part of how I think this will actually evolve. My hope and assumption is that these ideas might be useful to you today and the systems and processes you all are building and influencing today could be improved.

We have been trained to think that automation is the height of technological advancement. I would propose there are even more advanced levels: anticipation, and prevention.

Anticipation vs. Automation: the future is software that "takes a hint"

The problem with autocorrect is that it merely automates a series of rules. If the designers added anticipation, they should easily be able to detect that I do not want the same suggestion offered each time.

Here is my hierarchy of anticipation

Once is a hint
Twice is a pattern
Three times is a preference
Four times is a habit

For example, every time I start my car, my radio defaults to showing all the radio stations, despite the fact that I have presets programmed. So each time I start the car, I press the preset button to show my favorite stations. If the software designers were smarter, they would track this and realize that the first time I did this was a hint. The second time I did it was a pattern. By the third time a user repeats an action, I consider that a clear preference. If they had tracked my history of button presses, they would have seen that I do this EVERY time I start my car.

This suggests that autocorrect lacks anticipation. If I decline estrogens 2-3 times, it should "learn" that preference, and at least offer another suggestion.

Lastly, how could this be applied to your business processes? Are you "anticipating" your customers using signals they are already giving you? Are you using caller-id to even identify them before connecting them to your operators?

Customers are giving us hints. Repeated hints are patterns. Repeating patterns are preferences. If those preferences are acknowledged, they become habits.

Customers with habits are profitable.


Vivek Venugopalan said...

Robert - it sounds like you are asking for automation with self correction - sort of a learning system.

That does come with its own can of worms though - if you remember Microsoft office 2003 and how they hid menu items that you rarely used - I had to take the pains of turning it off as when I needed that rare formatting feature that I use, I could never find it on the menus!

Mike Hugo said...

Awesome idea - I have run into this same problem several times. Hopefully some day your idea becomes a reality.

In the meantime, you can work around it by adding a contact to your address book - create a company, for example, with the name of "rstephens" - then the system won't autocorrect it when you type it.