Saturday, March 22, 2014

Who Puts the Scary in Pearson? Meet Knewton.

Behind the data generating-and-collecting behemoth that is Pearson is a company called Knewton. And here's a video from the November 2012 Education Datapallooza (a name that I did NOT make up, but was officially given the event by the Dept of Education, because they are so hip. I believe they also listen to the rap music).  In just under ten minutes, Jose Ferreira, Knewton CEO, delivers the clearest picture I've ever seen of the intentions of the Acolytes of Data. (H/T to Anne Patrick.)


He opens with the notion that in the next few decades, we will become a totally data mined world. There are plenty of reasons to be concerned about that, but that's another post. He may well be right. He believes that has big implications for education, because while everybody is just collecting data in dribs and drabs, education is the Great River O'Data.

Knewton is now (and remember-- "now" is 2012) collecting millions of data points per day per student. And they can do that because these are students who are plugged into Pearson, and Pearson has tagged every damn thing. And it was this point at which I had my first light bulb moment.

All that aligning we've been doing, all that work to mark our units and assignments and, in some places, every single work sheet and assignment so that we can show at a glance that these five sentences are tied to specific standards-- all those PD afternoons we spent marking Worksheet #3 as Standard LA.12.B.3.17-- that's not, as some of us have assumed, just the government's hamfisted way of making sure we've toed the line.

It's to generate data.

Worksheet #3 is tagged LA.12.B.3.17, so that when Pat does the sheet his score goes into the Big Data Cloud as part of the data picture of Pat's work. (If you'd already figured this out, forgive me-- I was never the fastest kid in class).

Knewton will generate this giant data picture. Ferreira presents this the same way you'd say, "Once we get milk and bread at the store," when I suspect it's really more on the order of "Once we cure cancer by using our anti-gravity skateboards," but never mind. Once the data maps are up and running, Knewton will start operating like a giant educational match.com, connecting Pat with a perfect educational match so that Pat's teacher in Iowa can use the technique that some other teacher used with some other kid in Minnesota. Because students are just data-generating widgets.

Ferreira is also impressed that the data was able to tell him that some students in a class are slow and struggling, while another student could take the final on Day 14 and get an A, and for the five billionth time I want to ask this Purveyor of Educational Revolution, "Just how stupid do you think teachers are?? Do you think we are actually incapable of figuring those sorts of things out on our own?"

But don't be insulted-- it's not just teachers who are stupid, but the students themselves. Knewton imagines a day when they can tell students how they best learn and under what conditions. Will you do best watching videos or reading? "We should be able to tell you what you should have for breakfast [to do well on a test]"

Because human beings are simple linear systems and if you measure all the inputs, you can predict all the outputs? That seems to be our assumption, and even I, a high school English teacher for crying out loud, know enough about chaos theory and the systems of complex systems to know that that is a fool's game. (If you want to read more about it, and you should, I highly recommend Chaos by James Gleick)

Beyond the privacy implications and the human-beings-as-widgets implications and the necessity to tag every damn sentence of every damn assignment so our data overlords may drink their fill-- beyond all that, there are implications for what an education means.

One aspect of becoming an educated person is getting to know yourself, to understand your strengths and weaknesses, your abilities and deficits, defining your own character, and making choices about how to be in the world as a your particular version of a human being.

How, I wonder, do we adjust to software that attempts to do most of that for you? How do you get to know who you are when you've got a software program looking over your shoulder and telling you all about who you are with implacable inhuman data-driven assurance? It's a huge question and one that I feel unsure of how to answer. I wish the guys at Knewton shared a little bit of my fear and unsureness.

UPDATE: Twitter user Barmak Nassirian directed my attention to this article, which provides an even more complete view of exactly how Knewton thinks they can accomplish their goals. It confirms the impression that these are guys who know a lot more about data systems than about carbon based life forms. It's long-- but it's interesting reading.

"The New Intelligence" by Steve Kolowich. Inside Higher Ed.

5 comments:

  1. The Knewton developer's blog gives some insight to the methods of their madness. This particular entry might be of interesting to those familiar with control theory (engineering) and/or Bayesian inference (probability).

    In a nutshell, Knewton plans to make an estimation of a student's knowledge and future performance, in part, by measuring the time interval between answers to questions. Kalman filtering reduces the noise, and a Bayesian inference (e.g, given A, what is the probability of B, refined over time as more data is collected?)

    If you watched the CEO's video above, you begin to see why he thinks as he does with methods described here. I'd love to see anything outside of Knewton that corroborates this application of techniques for estimating someone's future knowledge on a subject.

    "We will use a Kalman filter to infer an approximation to a student’s ability given their response times to a set of questions."

    http://www.knewton.com/tech/blog/2013/11/kalman-filter/

    ReplyDelete
  2. Horrifying and sick, all in one presentation.

    ReplyDelete
  3. When he talks about concepts in the homework a student will fail in, I kept thinking about how someone should have looked at his data points and told him he didn't know how to use that slide clicker.

    ReplyDelete
  4. This is SCARY and EVIL! I think the purpose of data mining students has a lot more to do with Control" and less to do with Predictions."

    According to literature by Silent Sounds, Inc., it is now possible, using supercomputers, to analyze human emotional EEG patterns and replicate them, then store these "emotion signature clusters" on another computer and, at will, "silently induce and change the emotional state in a human being".

    Edward Tilton, President of Silent Sounds, Inc., says this about S-quad in a letter dated December 13, 1996: "All schematics, however, have been classified by the US Government and we are not allowed to reveal the exact details... ... we make tapes and CDs for the German Government, even the former Soviet Union countries! All this with the permission of the US State Department, of course... The system was used throughout Operation Desert Storm (Iraq) quite successfully."

    By using these computer-enhanced EEGs, scientists can identify and isolate the brain's low-amplitude "emotion signature clusters," synthesize them and store them on another computer. In other words, by studying the subtle characteristic brainwave patterns that occur when a subject experiences a particular emotion, scientists have been able to identify the concomitant brainwave pattern and can now duplicate it.

    "These clusters are then placed on the Silent Sound [TM] carrier frequencies and will silently trigger the occurrence of the same basic emotion in another human being!"

    Regarding system delivery and applications, there is a lot more involved here than a simple subliminal sound system. There are numerous patented technologies that can be piggybacked individually or collectively onto a carrier frequency to elicit all kinds of effects.

    There appear to be two methods of delivery with the system. One is direct microwave induction into the brain of the subject, limited to short-range operations. The other, as described above, utilizes ordinary radio and television carrier frequencies.

    Far from necessarily being used as a weapon against a person, the system does have limitless positive applications. However, the fact that the sounds are subliminal makes them virtually undetectable and possibly dangerous to the general public.

    SOURCE: http://www.nwbotanicals.org/oak/newphysics/synthtele/synthtele.html

    ReplyDelete