Stupidity scales

I’m fed up with conversations that end something like this.

Yes, that would be the smart thing to do, but it won’t scale. The stupid approach is better because it scales.

We can’t use common sense because it doesn’t fit on a form.

We can’t treat people like people because that doesn’t scale well.

We can’t use a simple approach to solve the problem in front of us unless the same approach would also work on a problem 100x larger that we may never have.

If the smart thing to do doesn’t scale, maybe we shouldn’t scale.

Related posts:

Organizational scar tissue
Rewarding complexity
What’s wrong with paper?

15 thoughts on “Stupidity scales

  1. I agree, John. Of course, there is nothing wrong with scaling per se. If you can find a solution that scales well, that’s great. The problem, as you point out, is avoiding solutions for the sole reason that they don’t scale well. I haven’t given this a lot of thought before. In education, we often set as a criterion from the beginning, before a solution is found, that it must be able to scale well. By doing so, are we overlooking more obvious solutions? I’m going to give this some more thought. Maybe it would be helpful to put more concrete examples out there.

    I know you have mentioned standardized testing in another post. Scalability is a terrible excuse for the poor quality high stakes testing we use in the U.S. Other countries rely on richer sources of testing, like essays, which we seem to think in the U.S. can’t be scaled up objectively. And yet other countries manage to do it just fine. My wife is French and she never saw a multiple choice test in her life until she came to North America.

  2. Thanks, Scott.

    We could just weigh students. That would be very objective. And it scales! (Pun intended.)

  3. Having once worked for a giant government department, I’d have to disagree… the reason that common sense and humanity doesn’t ‘scale’ is because the larger the organization, the more idiots and a**holes one tends to employ… and while the number of such individuals will increase linearly with the total number of employees, the damage that they do (legally, financially, and to reputation) increases exponentially.

    The best way to deal with this is to improve the selection process and employee training, and to be able to swiftly deal with transgressors. On the other hand, its far cheaper to replace thought with rule books and common sense with forms.

  4. Scott, I would say that education has already made this choice. Haven’t they basically said that finding good, motivated teachers who may not teach the same content, but engage the students does not scale, so we will take a fixed curiculum and ensure everyone teaches to it, without asking whether the kids are engaged?

    I know I’ve read recently a number of articles that said “a group of motivated parents and teachers may be able to improve a school, but how can we expect there to be a suitable group at every school?” – in other words they think that grassroots doesn’t scale. I think there is a difference between being able to control something and it scaling.

  5. Who is “they”, Walt? I am guessing politicians. All educators I know lament the fact that there is so little trust of teachers in the U.S. Other countries treat teachers as professionals. In Asia they give teachers enormous respect even though they do not earn high salaries. But here in the U.S. teachers are not trusted to do a good job, so they are controlled by increasingly restrictive curricula and standardized testing.

    But you are right. Engaging the students does not scale, and yet we know local solutions are often the best. What is amazing is that the U.S. has always been a grassroots country. Other nations control education from the top down, usually starting with an all-powerful Ministry of Education. And yet they manage to respect and empower their teachers to do their best. I don’t get it. Like you said, there must be a difference between scaling and controlling. There must also be different ways of controlling, and I think we’re picking all the wrong ways.

  6. Here’s an essay by Paul Graham I wanted to throw into the conversation. He argues that working in large organizations is unnatural in much the same way that eating highly processed food is unnatural. We work for large organizations for short-term benefits much like the reasons we eat junk food.

  7. Scott, you’re correct that education policy is not made by educators – it is mostly made by state education departments, some of who are former educators, but they must answer to the elected bodies.

    When I wrote the first comment above, I was thinking about grassroots vs. controlled having the same relation as emergent properties vs. reductionist properties. The former exist even when we don’t know how they come about or how to guarantee that they will occur.

  8. That is what the so called hip of scientific parallel computing is all about: Reject any clever idea because we *need* to run it in 1000 processors. But, what if the clever idea can solve the same problem in 10 processors (but not more)? Still, It doesn’t matter, because it *must* be run in 1000 processors, so we can’t use it.
    Scaling in this context means be as dumb as you can because dumb things scale better. I would put it the other way around: a dumb approach can and must use a lots of processors exactly because it is a dumb idea.
    … still, we *need* to fill up the $1000000 cluster!! it is empty and it is a shame!

  9. By all means, let us not scale, as long as we can only figure out how to do it stupidly.

    Several of the earlier comments have to do with Coase’s ceiling, which is vastly lower than is commonly thought.

  10. Nice. However, I wonder: in a world that is “small world”, even “local” properties do scale. Beggining to treat people like people would be a nice move, and I reckon I could very well scale. Any virus (biological, ideological, computational,…) ends up infecting large parts of a population even though it’s contagion (sorry if this is not the right english word…, let’s say: transfer) mechanism is just local “from person to person”. You need a rather small probability of transfer for percolation to happen in a small world.

  11. Most people here touch on the big problem with common sense : it’s not egalitarian. It would never fly politically because it will amplify differences between schools/businesses/departments/…

    The politician that approves grassroots education gets to answer the question “I want my kids to do better” with “that’s impossible” or “do it yourself”.

    The only way to make people equal is lowest-common-denominator performance. Until we are ready to answer “why do black schools do worse ?”, “why are muslims so violent at school ?”, “why does X do worse than Y”, until these questions can reliably be answered with “it’s their own fault” (since they mostly control the school districts responsible for their own fate). Until we get to that point, forget about maximizing kids’ performance.

  12. @tomcpp Right if extreme in some ways. I don’t know violent muslim kids but they probably exist. Egalitarian is a really stupid idea. I’m about 2.5 standard deviations above an average IQ why should my results be comparable to the norm? Why should I be “progressively” taxed more for being lucky? Are we so jealous of people that do well that the only way we can deal with it is to find a way to take from them and give it to others?

    That is precisely why responsibility and resources should be pushed down to the local community. If the collection of inner city parents, churches, schools etc cannot find a way to teach their kids how should some rich white guy in the capital come up with a way to “relate” with them any better?

Comments are closed.