How to fail with data

Sometimes the most effective way of communicating the right way to do something is by highlighting the consequences of doing the opposite.  It’s how sitcoms work.  By creating humorous situations that highlight the consequences of breeching social norms, those same norms are reinforced.

At the 2017 Blackboard Analytics Symposium, A. Michael Berman, ‎VP for Technology & Innovation at CSU Channel Islands and Chief Innovation Officer for California State University, harnessed his inner George Costanza to deliver an ironic, hilarious, and informative talk about strategies for failing with data.

What does this self-proclaimed ‘Tony Robbins of project failure’ suggest?

  1. Set unclear goalssetting unclear goals takes a lot of hard work and may require compromise. It’s way more democratic to let everyone set their own goals.  That way, everyone can have their own criteria for success, which guarantees that whatever you do almost everyone is going to think of you as a failure.
  2. Avoid Executive SupportGoing out and getting executive support is also a lot of work. It means going to busy executives, getting time of their calendar, and speaking to them in terms they understand.  It also means taking the time to listen and understand what is important to them.  Why not go it alone?  Sure, it’s unlikely that you will achieve very much, but it’ll be a whole lot of fun.
  3. Emphasize the Tech Make the project all about technology. And make sure to use as many acronyms as possible.  Larger outcomes don’t matter.  They are not your problem.  Focus on what you do best: processing the data and making sure it flows through your institution’s systems.
  4. Minimize Communication Why even bother to make people’s eyes glaze over when talking about technology when you can avoid talking to anyone at all? Instead of having a poor communication strategy, it’s better to have no communication strategy at all.  You’ll save the time and inconvenience of dealing with people questioning what you do, because they won’t know what you’re doing.
  5. Don’t Celebrate SuccessIf you have done everything to fail, but still succeed despite yourself, it’s very important not to celebrate. Why bother having a party when people are already getting paid?  Why take time out of the work day to reward people for doing their jobs?  Isn’t it smarter to just tell everyone to get back to work?  Seems like a far more efficient use of institutional resources.

Speaking from personal experience, Michael Berman insists that following these five strategies will virtually guarantee that you drive your data project into the ground. If failing isn’t your thing, and you’d rather succeed in your analytics projects, do the opposite of these five things and you should be just fine.

Analytics isn’t a thing…it’s a relation

In response to the 2017 NMC Horizon report, Mike Sharkey recently observed that analytics had disappeared from the educational technology landscape. After being on the horizon for many years, it seems to have vanished from the report without pomp or lamentation.

For those of us tracking the state of analytics according to the New Media Consortium, we have eagerly awaited analytics’ arrival. In 2011, the time to wide-scale adoption was expected to be four to five years. In 2016, time to adoption was a year or less. In 2017, I would have expected one of two things from the Horizon Report: either (a) great celebration as the age of analytics had finally arrived, or (b) acknowledgment that analytics had not arrived on time.

But we saw neither.

Upon first inspection, analytics seems to have vanished into thin air. But, as Sharkey observes, this was not actually the case. Instead, analytics’ absence from the report was itself a kind of acknowledgement that analytics is not actually ‘a thing’ that can be bought and sold. It is not something that can be ‘adopted.’ Instead, analytics is simply an approach that can be taken in response to particular institutional problems. In other words, to call out analytics as ‘a thing,’ is to establish a solution in search of a problem, as if ‘not having analytics’ was a problem itself that needed to be solved. Analytics never arrived because it was never on its way. The absence of analytics from the horizon report, then, points to the fact that we now understand analytics far better than we did in 2011. If we knew then what we know now, analytics would not have been featured in the horizon report in the first place. We would have put understanding ahead of tools, and bypassed the kind of hype out of which we are only now beginning to emerge.

I agree with Mike. But I want to go a step further. I have always been fascinated by ontologies, and the ways in which the assumptions we make about ‘thingness’ affect our behavior. I have a book in press about the emergence of the modern conception of society. I have written about love (Is it a thing? Is it an activity? Is it a relation? Is it something else?). And I have written about dirt. Mike’s post has served as a catalyst for the convergence of some of my thinking about analytics and ‘thingness.’

Analytics is not a thing. I can produce a dashboard, but I can’t point to that dashboard and say “there is analytics.” There is a important sense in which analytics involves the rhetorical act of translating information in such a way as to render it meaningful. In this, a dashboard only becomes ‘analytics’ when embedded within the act of meaning-making. That’s why a lot of ‘analytics’ products are so terrible. They assume that analytics is the same as data science with a visualization layer. They don’t acknowledge that analytics only happens when someone ‘makes sense’ out of what is presented.

Analytics is like language. Just like language is not the same as what is represented in the dictionary, analytics is not the same as what is represented in charts and graphs. Sure, words and visualizations are important vehicles for meaning. But just as language goes beyond words (or may not involve words at all), so too does analytics.

It is a mistake to confuse analytics with data science. An it is a mistake to confuse it with visualization. If analytics is about meaning-making, then we are working toward a functional definition rather than a structural one. This shift away from structure to function opens up some really exciting possibilities. For example, SAS is doing some incredible work on the sonic representation of data.

As soon as we begin to think analytics beyond ‘thingness,’ and adopt a more functional definition, its contours dissolve really quickly. If what we are talking about is a rhetorical activity according to which data is rendered meaningful, then we are no longer talking about visualization. We are talking about representation. In a recent talk, I suggested that, to the extend that analytics is detached from a particular mode of representation, and what we are talking about is intentional meaning-making — meaning making intended to solve a particular problem — then a conversation can easily become ‘analytics.’

So analytics is not a ‘thing.’ It is not something that we can point to. Is it an activity? Do we ‘do analytics’? No, analytics isn’t an activity either. Why? Because it is communicative, and so requires the complicity of at least one other. Analytics is not something that we do. It is something we do together. But it is not something that we do together in the same way that we might build a robot together, or watch television together, where what we are talking about is the aggregation of activities. What we are engaged in is something more akin to communication, or love.

Analytics is not a thing. Analytics is not an activity. Analytics is a relation.