David Doria

I am a huge open source advocate, and long time Linux user, but I have to push back on this one a little bit. I will agree fully with "Linux *should* be "as easy as Windows" " these days. However in practice it simply is not the case. Some examples:

1) Fedora's NVIDIA support (or lack of it). I had to learn about kmod, akmod, how to disable the system's attempt at installing the driver to install the one from NVIDIA, install the kernel headers in order to compile NVIDIA's driver, etc. This drove me crazy enough to abandon Fedora and switch to Ubuntu (and I (claim to) know what I'm doing a little bit).

2) Too many obnoxious errors. Clearly Windows has plenty of obnoxious errors, but significantly many fewer. I still can't figure out how to get the default IM client in Ubuntu 11.10 to not make me "click ok to access my password wallet". When I first installed 11.10, the package manager would simply not let me install packages because I was not root. Only after hours of Googling I found out that I had to install some other polkit-like package to allow the package manager to prompt me for a root password like it should. The list goes on and on.

3) Hardware support. I bought a Canon laser printer, and it turns out Canon does not make drivers for Linux AT ALL (and neither does anyone else for this printer), so I cannot print to my printer from my Linux (main) machine. Even though I've been a Linux guy for many years I didn't think to check that before buying it!

4) Software support. There is an open source alternative for almost everything now. Unfortunately, most of them are terrible. I will not touch Libre Office with a 10 foot pole. There is no reasonable 3D CAD (SolidWorks style) software. The development IDE's can't touch Visual Studio, etc.

----
The Litmus test should be "would I recommend this to my grandpa?". Of course if grandpa will never touch anything except a web browser, sure - in fact I'd argue that most people would not even know which operating system they were using as long as you show them where the web browser icon is. However, if you expect grandpa to get frisky and want to upgrade anything besides his RAM, you'd be crazy to recommend Linux.

I'd say it is about 90% there, but the last 10% no one seems to care about fixing, and that is the 10% that people notice.

I'd change "Linux is no longer hard" to "Linux is no longer 5x as hard". In my opinion it is still much much harder than it should be.

This was really a fantastic presentation. It is nice to see people thinking about education from an economic perspective like this. Some of the figures were shocking - by replacing a SINGLE "lower 5%" teacher with an average teacher for 1 year, a $330k total earnings raise will be achieved - amazing!

To answer the questions you posed:

1) What do you think about this research?
- I really liked how they really proved that the Value-Added measure is unbiased. It seems that too often people make a "necessary but not sufficient" type of argument without even knowing it.

- There was clearly much more than an hours worth of presentation that he stripped down for this talk. He touched only briefly at the end on how to turn these statistics into a policy recommendation. I found two things quite interesting - first there are current attempts to retain good teachers by incenting them with a $2k bonus after their 3rd year if they stay on. However, he noted that 90% of these teachers would have stayed on anyway, so the majority of the funds are spent motivating teachers who did not need the motivation! The second interesting thing was the "how long do we gather statistics about a teacher before dismissing them for poor performance?" question. He showed that with only 1 or 2 years data, we could likely increase total earnings by about $100k by replacing a single lower 5% teacher with an average teacher. Of course by gathering statistics for longer (about 10 years or so) you can be much more certain that you have truly selected a lower 5% teacher, but the return seems to be only about $10k/yr. The point is that if you wait two years so that you gain $110k instead of $100k, you have forgone the $100k for 1 year only to gain $10k the next year!

2) What should be included in teacher evaluations?
There is a great book "Measuring Up" (http://www.amazon.com/Measuring-Up-Educational-Testing-Really/dp/0674035216/ref=sr_1_1?ie=UTF8&qid=1328708775&sr=8-1) that discusses in great detail how to properly design, administer, and interpret tests (for students) properly. When done correctly, it seems they could be a great source of information about the achievement of the teacher (though it seems they are almost never used correctly...). Others (http://www.joebower.org/) are passionate that exams/grading should be abolished entirely.

I think it would be very easy for a trained observer to watch a very short presentation (unannounced of course) by a teacher and conduct short interviews with students and come to a conclusion about the quality of their instruction. However, where do the salaries of these reviewers fit into budgets, etc? (Perhaps from the large long term gains predicted by the presentation!)

3) How long should a teacher have to improve an evaluation score?
According to the presentation, less than 3 years. In my opinion, it is extremely hard if not impossible for a teacher to change their ways. So, if they are evaluated with a unbiased, reliable metric that shows them doing extremely poorly, it seems to me that they should be dismissed almost without any time for improvement at all. Surely no teacher in the bottom 5% does not know that they are at or near the bottom! If you believe this, then they have already proven throughout their tenure that they are unwilling to change :)

Thanks for posting this very interesting piece!