Pair Crossword Puzzling

My wife and I have been doing these cryptic crosswords - we usually do one a night as we're trying to coax Sofi asleep.  They're very satisfying once you start figuring out how to do them - the first few we did we couldn't make head or tail of, but after looking at the answers we started figuring out the tricks. 

Interesting thing about them--and I haven't done tests to prove this scientifically, but I'm pretty sure it's the case--it takes two people less than half the time to solve a cryptic crossword than it would take one person.  (This is true for regular crosswords also, I'm pretty sure.)  Each answer gotten lays the groundwork for future answers, so we're accelerating each other through the project.

Which is the opposite of Brooks' law.  So there are situations, such as solving crossword puzzles, where it's not true.  Another example:  an artist and a coder can get more than twice the game made that just an artist or just a coder can.

And sometimes you put two programmers together on a single machine because two together can outperform two by themselves.

Comments

[Paul Sinnett]

What you've found backs up the existing evidence about pair programming. (A closer fit might be slightly more than half the time but with less mistakes along the way; perhaps correcting mistakes is quicker with crosswords.) This doesn't exactly contradict Brooks' law, as I understand it, because that came from large projects with many programmers.

Unfortunately the pair programming experiments I've seen focus on comparing pairs working together vs pairs working apart. Perhaps a more interest test would be teams of 6 pairs vs teams of 12 individuals?

Or maybe you could see how long it would take if you had n people trying to fill the crossword together? Is the relationship linear (t/n) or something else?

Idea for a game experiment: "Collaborword." You play through the web and fill answers against the clock. When all the answers are in it tells you how many words are wrong (but not which ones.) Any size of team can compete on one puzzle.

[Rob Drimmie]

An important element of Brooks' Law is that the additional manpower you're adding are to a late project.

To use the cryptic crossword example, if you had completed 80% of a puzzle and your wife came in, at that point would you finish it any faster than if you'd just kept going yourself? Or would the time it would take to get your wife up to speed on the particulars of that puzzle take more time?

An implicit assumption to Brooks' Law is that the project is properly staffed to begin with. It is ultimately an argument in favour of proper planning and scope assessment, not one for starving a project of necessary resources.

[eh]

You don't seem to understand Brook's law. A crossword puzzle is not a late software project. Brook's law does not say more people always work less efficiently.

[Jamie Fristrom]

Okay, so forget about Brooks' law, which is "Adding programmers to a late software project makes it later" - which isn't even true (which Fred Brooks himself admits) - I've added programmers to projects that seemed late and brought them in on time.

Let's go with another law, which you'll find in most management books, I reckon, but I can't attach a name to it - "Adding more manpower to a team makes it less efficient." I'm just pointing out a couple exceptions to that law - mostly for my own benefit, because I tend to forget there are exceptions.