Douglas Merrill, Big Data Demystified for Learning: What’s Important, What’s Not, and What’s Next #LSCon

These are my live blogged notes from the Thursday morning keynote at the eLearning Guild’s Learning Solutions 2014, happening this week in Orlando, Florida. Forgive me any typos or incoherence.

Douglas Merrill http://www.douglascmerrill.com

• Co-Founder and CEO of ZestFinance.com; Former Chief Innovation Officer at Google

• Author of Getting Organized in the Google Era

• Has a PhD in Cognitive Science from Princeton and a very nice peacock tattoo sleeve on his left arm

In the last ten years, we’ve created a LOT of information.

3% of all employees are working at 2am local time.

We’re killing our employees, we’re killing ourselves.

On average, the American adult worker works 9 hours a day at their office, but only 2 hours a day doing primary child care.

Big Data -- we think it’s magic. We are so in love with the notion that data can reveal hidden things like magic. 

Big Data is like religion. It is believed without being understood.

Math anxiety causes use to believe without understanding.

We create made up numbers all the time (like that American men are taller than women.) There are outliers in the data than can blow your average.

Listen to your customers, but not too carefully….“If I had asked my customers what they wanted, they’d have asked for a faster horse.” ~Henry Ford

The downside of focus groups – you ask customers questions and they want to please you – they want to give you answers they think you want. They can’t envisage answers that they don’t know.

Google results – no one clicks NEXT to look at the second results page – you’re in the top three to four results or you don’t matter.  Google customers asked for 20 results on a page, so Google listened to customers from their focus groups and went from 10 to 20 results. The problem – the search results went from .25 milliseconds to half a second and search completion results went down! People really wanted speed.

Crowdsourcing is a precision tool. It gives you an amazing amount of information. He shares a story of how pilots came up with the best routes and discovered the Jet Stream – by sharing their flight logs.

Go Do Something:

We are part of the problem…statistically, if we hire people -- we're going to hire people who are just like us. White men hire white men. White women hire white women. etc. 

Force diversity into your plan. This creates a broader decision tree. When you're hiring people, hire for diversity. Diversity matters. We all think about the world in the context of the company that we live with. 

You're promoting wrong. You're more likely to hire people who look like you; you're more likely to promote people who look like you. Blow up the reviewing process. At Google, all reviews are public. Every quarter was a full 360 review. All those reviews were entirely public in name - you see what I thought your strengths were; your opportunities for improvements (aka weaknesses); a field for anon feedback; list three employees who are worse than this person, three who are better. Crowdsourcing the likelihood of how well you were doing. If people who are worse than you are higher than you in rank, you should probably be promoted.

Develop people, not just apps. If people talk like you, you're more likely to understand them and then promote them, etc. Development process don't capture what people actually do.

People do what you measure. People do what we tell them to do, even if it's dumb.

We can do better. We can use data that create value for our companies.