Comments on: Puzzler: Be in a usability study

By: Andrew Newell

Andrew Newell — Mon, 27 Oct 2008 13:33:04 +0000

It runs 30x faster still if the brackets are removed:
sum(union(0:3:1000,0:5:1000))

By: Mark

Mark — Thu, 23 Oct 2008 14:15:15 +0000

This solution runs about 20x faster than the “arrayfun” solution: sum(union([0:3:1000],[0:5:1000]))

By: dhull

dhull — Mon, 13 Oct 2008 14:26:28 +0000

Nicolas,

That had to be the fastest solution I have seen to the problem. It was also very readable.

Everyone,

Thank you for the movies. I have gotten them all and watched them. They are being packaged up for the Usability team to watch too.

I will be making a review of the skills and different techniques used for an upcoming video.

The t-shirts will be sent this week.

Thanks for playing!
Doug

By: Daniel Armyr

Daniel Armyr — Mon, 13 Oct 2008 12:01:07 +0000

Oh, and on the topic of one-liners, here is a generic one-line solution that is by far faster than anything proposed. Unfortunately, it is approximate, but the error is less than 1% for values of n > 50.

sum = 70/3*n^2;

This algoritm was found by pure trial-and-error using plot functions and sample datasets.

–DA

By: Daniel Armyr

Daniel Armyr — Mon, 13 Oct 2008 11:39:34 +0000

Oops, seems I misread the question. I calculated the sum of values multiples of 3 AND 5. No wonder I got another answer.

Oh well, video is in. Unfortunately, working while recording was a real drag because my computer became so slow. Now that I redid the exercise without recording I realize that I work slightly different when things flow a bit smoother.

Daniel Armyr

By: Nicolas

Nicolas — Sun, 12 Oct 2008 21:29:46 +0000

I’ve uploaded my video to the ftp site, and sent it as an attachment to your e-mail. Hope it works.

By: Omid

Omid — Fri, 10 Oct 2008 08:31:46 +0000

Here’s the link to the video I made for this usability study
http://www.screencast.com/t/Pxnu4qqDD

By: dhull

dhull — Wed, 08 Oct 2008 14:02:00 +0000

Doug,

You are right, that this is a very short problem. We are trying to keep the barrier to entry very low on this event, and keep things within the time limits set by Jing.

Even though there are a plethora of one-liners that solve this problem, the videos are interesting none the less. We look to see how people vary from developing at the command line vs editor, if they use shortcuts and accelerators to speed their work and so on. We can see how they have their desktop arranged, and what shortcuts they have added to their custom shortcuts bar.

There are very often small productivity gains available to people (like F5 to save and run a file vs. saving, going to the command line and then running) that are available but have remained undiscovered. Knowing what we need to make more Learnable and Discoverable is important to us.

So yes, this is a simple exercise. However, even with the limitations we have doing this remotely, there are things to learn.

Thanks,
Doug

By: Doug Schwarz

Doug Schwarz — Wed, 08 Oct 2008 12:43:53 +0000

I think this problem is too simple to tell you anything about usability. Most people will not even have to consult the docs. For what it’s worth, I simply typed in

a = 3:3:1000;
b = 5:5:1000;
sum(a) + sum(b) - sum(intersect(a,b))

By: Danilo

Danilo — Tue, 07 Oct 2008 23:05:13 +0000

There is no Jing for Linux, and I too far from the US to receive a t-shirt… so here is my solution:

I just typed

a = 1:1000; sum(a(~mod(a,3)|~mod(a,5)))

and was happy (and a bit surprised) to see it worked in the first try (resulted in 234168, just like the first comment).