Stay at the forefront of data & information management. Join to comment, vote, contribute and learn. Learn About Member Benefits & Join Now

Blog

  1. No Data Science Without Simulation

    This morning I awake in Hamburg, Germany, after a long sleepless flight in economy and a catch-up day in bed. The sun is shining and the city is alive with rhododendren, bright clothing, and sunny smiles. Hamburg is an overcast city most of the year. Like Brussels, it is addicted to low pressure systems and periodic light rain. But in late May, as the winter melts into summer, something magical grips this river town and the life within it opens up like flower petals.

    I am up early and down for breakfast before the crowd awakes. The hostess tells me to take any table so I find a booth near the window. Laughing African musicians sit two tables behind me, and just in front of me sit a mid-50′s German couple dressed in black and grey light summer shirts and pants, the wife in a matching dress. Both are roundish, rolly-polly, chatting quietly. I have gotten up to get food from the buffet and returned to my table as the waitress brings me a fresh pot of coffee in a chrome thermus. Just then a mid-30′s couple sit down in the booth to my left. Both are short, fit, well-groomed, sporting polo shirts, shorts, sneakers. The man is wearing a round gold watch with a black leather band, white dial, black numbers. Lawyer. The wife sits quietly whille the husband fetches two glasses of German Champaign and matching glasses of orange juice. She then waits in her seat while he fills two plates with food and returns. She then takes her turn at the buffet while he waits.

    While she is up getting food, I notice that the wife of the grey couple is also up at the counter. What unfolds next is the kind of quiet breakfast drama one can only appreciate when one travels alone. The grey man is butttering his bread. His body twitches slightly and his hands shake. He has Parkinson’s Disease and he is suddenly insecure now that his spousal shield has left the table. It is obvious that he is trying very hard not to shake, as if the public display of his ailment is causing him additional grief and shame. Next to me, the Lawyer has a plate of fruit. He has his knife and fork out and he is carefully cutting up his watermellon into 1cm squares. He traces his knife along the rind and carves out every pink millimeter of flesh. His plate is now full of 25 neatly cut squares and he moves on to the Pineapple.

    Meanwhile, Mr. Grey is eating his toast with his hands. No utensils, and his hands shake slightly as he brings the brotchen to his lips. With each bite he looks around the room to check if anyone is watching. The contrast in age, appearance, demenor, and behavior between these two men, sitting just two meters apart, is astonishing. In 5 quiet minutes of observation, I am reminded of the enormous variation in human behavior. Two men with breakfast on their plates behave entirely differently according to their conditions, upbringing, and expectations.

    We cannot remove these behaviors from people. They are ingrained by experience and habit and they shape the smallest and largest decisions.

    Given two plates full of Data, these two men would also, no doubt, reach entirely different conclusions. They have different educations, socio-economic positions, and their lives have been shaped by different attitudes towards risk, social acceptance, human interaction. Which would be right? I wouldn’t know where to estimate that, but I would not want to bet my business on any human judgement without the ability to simulate the situation and evaluate the outcomes BEFORE the decisions are made.

    With all the mountains of Data we are throwing at each other today, now more than ever we must admit that we humans are the most inconsistent components in our Data Science Belief Systems. I really fear how we worship Data, dub it a Science to read it, and expect imperical answers from our largest, fastest, and most complete analyses. We should be a bit more humble. We should recognize the power we posess to persuade ourselves of what we want to read, the impact of self-interest on decision-making, and the enormous odds that given the most obvious intelligence we may still get it wrong more often than right.

    This is why I believe so passionately that we need better Systemic Simulations to test our hypothesis and decisions in safe test environments before committing decisions to actions. Let people make lots of mistakes on the Holodeck before the get to the Bridge and press a button.

    There are as many potential outcomes as there are people to invent them. Lets be humble enough add Simulation to Data Science.

  2. A Big Data Business Model for Facebook and Twitter

    This morning, General Motors announced that it would no longer advertise its cars on Facebook. This announcement comes a day before the Facebook IPO, and casts a shadow on the business model of Facebook. GM said that they will continue to support their page and user community on Facebook, but that ads just weren’t effective in helping consumers to make car buying decisions. Ford jumped on this announcement to say they would continue to buy ads on Facebook and that Social Media requires a consistent commitment to innovation and community development.

    Maybe. But I think GM’s decisions does illustrate a key problem for Facebook and Twitter – the revenue model. Social Media grew up without dependencies on ad-based revenue. On Facebook, you aren’t a customer. You are a product, and its your likes, dislikes, friends, photos, videos, and content that generate value. Selling products to products via advertising is hard. Members don’t use Social Media to go shopping. There’s no commerce platform there. They use it to be social. There are so many other outlets that are more effective for advertising than Social Media.

    So how should Facebook and Twitter make money? My idea: make it collective. The value is in the data.

    1. Make terms and conditions explicit that every member owns their own data via copyright. This does two positive things.

    A. It indemnifies Facebook and Twitter for the crazy, infringing, and potentially libelous posts of their members by allowing them to claim that they are conduits of content rather than publishers or distributors.

    B. Copyright establishes the rights to royalties for content created and posted on their networks, which enables the next step.

    2. Allow members to opt-in to Big Data analysis by Social Media partners and intermediaries.

    3. Charge Social Media for Big Data Searches by data volume.

    4. Pay members royalties every time their data is used in Big Data Searches.

    This simple model creates powerful incentives that transform user members from products into mutual social network content providers with an economic interest in posting content that will be used in Big Data searches. It establishes data property rights that insulate Facebook and Twitter from vouching for the content on their networks. Members will also discover that providing high quality data that companies want to search for means more royalties and so the system will produce better behaviors. And it creates a 2-tier royalty distribution model that will also pay Facebook and Twitter handsome revenue that will change online advertising and make every other content aggregater change too.

    Of course, Facebook and Twitter will have to sort our who’s a person and who’s a bot, and will have to provide content creation tutorials to help users/customers create content that has value by sharing the top 100 Big Data queries and sample results.

    But this Business Model has something for everyone and is a true win:win. It benefits customers by establishing data property rights and royalties for content. It benefits organizations who want to do Big Data searches by providing ever richer data streams of high quality and availability. And it benefits Facebook, Twitter, and their investors by providing an enormous profit making engine selling Data.

    The Data is the Value. The more there is, the more valuable it becomes. Pay your customers to create higher quality data and charge your partners to use it. Its a simple Business Model.

    Dick Costolo – @dickc – and Mark Zuckerberg – @finkd – are you listening?

ChatClick here to chat!+