Monday, December 24, 2012

People don't need experts; they just need someone who knows

This is a really good analysis. I think you correctly identify the missing bit – helping students over those rough patches.

Interestingly, to my mind, although the problem of understanding and responding to a student question is an almost intractable problem for machines, it is generally pretty straightforward for humans. So what we have tried to do with cMOOCs is connect people with the humans they need to connect with to get over the rough patches.

You don’t need an expert for this – you just needs someone who knows the answer to the problem. So we have attempted to scale by connecting people with many other students. Instructors are still there, for the tough and difficult problems. But students can help each other out, and are expected to do so.

An example of what I mean: I just purchased a new xBox and a copy of MLB 2K12, which is a baseball simulator. My first effort to puitch saw me walk most of the batters, throw numerous wild pitches, and finally get out of the inning only after giving up 14 runs. The problem was, I didn’t know what to do; the MLB 2K12 instructions are far too vague, and if there’s in-game help,. I haven’t found it.

I don’t need an expert in MLB 2K12 to show me how to pitch. I just need someone who knows what to do. Someone who can say “Well you move this control here then here and you’re trying to line this up with that.” Million s of people know the answer to this question. but I’m connected to none of them.

Indeed, I don’t even need then to do the actual explaining. They simply need to recognize what my problem is, then point me to a video or instructions that outline the solution.

Machines will eventually be able to do this, but they will first need to master natural language processing. This is going to take a while. In the meantime, if we want massive learning, we need o structure learning in such a way as to make asking questions easier, and as necessary, to provide more incentives to people to answer them.

I don’t think the xMOOCs are ever going to do this, because their focus is on placing all the emphasis on the expertise of the instructor. To the extent that they respond to this need, they will become cMOOCs. But to the extent that cMOOCs become viable, the value proposition behind the elite universities is weakened. People don’t need experts; they just need someone who knows.

1 comment:

  1. I believe your issue with pitching in MLK2K12 is that you're not taking the flight of the ball into account. You need to know what type of spin or ratation each pitch has, so that when you aim the cursor in the strike zone, you're accounting for how the ball will move when aimed at that position. My hunch is that you can throw strikes with the 4-seam fastball, (4FB), because that's a straight pitch. But, here's the breakdown of the other pitches:

    Slider - away from the trowing hand with slight downward movement.
    Curveball - away from the trowing hand with exagerrated downward movement. (12/6 is straight down)
    Slurve - mix between slider and curve.
    2-seam fastball - starts away from the throwing hand and moves inward on a level plane.
    Sinker - like a 2-seam fastball with downward movement.
    Knuckleball - nobody knows where this will end up.

    So, for instance, if you want to throw a strike to a right-handed batter as a right-handed pitcher, and you want to do so with a Slider, then:

    Aim your cursor inside, off the plate, near the batters thigh or knee. The ball will start out as though it's going to hit him in the leg, but will curve to the left and end up as a low strike.

    Now you know. And now you see the beautiful chess match that is pitching. Trying to get the batter to think the ball is doing one thing when it's actually doing another.


I welcome your comments - I'm really sorry about the moderation, but Google's filters are basically ineffective.