This reductio ad absurdum would be appropriate if they argued generally that reducing the work week would be good. They don’t: they’ve tried an experiment and they’re reporting that they’re pleased with the results.

A good criticism would be that the improvements are just the Hawthorne Effect at work, that the improvement is due to the attention of the experimentation rather than a 4-day week being objectively better. 37signals didn’t control for it, but there’s no reason to because they’re not trying for empirical results, they’re trying for improvements.

If you want improvements, you can just use the Hawthorne Effect directly. Ask people their thoughts (a good idea anyways), tinker with something visible to them, and look to see an improvement. You can see this in online communities, if you start a discussion about the community, make sure everyone feels heard, and implement any change, overall morale will go up. It doesn’t matter if a web forum increases or decreases the number of subforums, changes the color scheme, or adds or removes minor features: the forum will be happier and busier afterwards.

The caveats: This can’t be used too frequently, people just get fatigued of changes. It hinges on giving people positive attention, if they feel unimportant it’ll only hurt things. And it’s not a substitute for the real, deliberate changes that give you large and long-term improvements.