It makes sense - fewer schedules to coordinate; and as noted in the article, three allows you to cover the three basic areas of competence in many games - hitting, magicing, and sneaking. With fewer players it is easier to give each more screen time and do more in-depth, character-focused arcs.
However, as my Dresden game demonstrates, going this small means that there's a minimum amount of room for schedule snafus. Sunday AM, one player had to cancel due to a small family emergency, and boom, there's a third of the regular folks at my table. Not his fault, but the arc is sort of starring him as well, so if he wasn't there, a lot of the hanging plot is focused around what his character brings to the table, so even MORE troublesome than normal. Then I lost my second player due to other unforeseen circumstance, and I was down to just a 1/3 of my table. So I called game.
If it had been a more stereotypical six player game and I'd lost two players, I probably could have continued with the arc. Plus with a larger group, the likelihood that the story arc would have been focused around a particular player versus a particular issue would have been less.
So what's better? As always, it is "It depends." What are your goals as a group, what stories are you interested in telling? There's no better or worse, just what works for your group.
I'd argue that if you want to tell heavily character driven arcs, where each player has a critical role then you either need to be okay with not playing every time you have something scheduled, or have an inordinately reliable group.
If you are willing to play with a larger table, and know that 1-2 people are going to be missing, then a plot that is more group focused versus character driven is probably wiser. That way if Sir McGuffin's player has to go into work, the game can still go on because the plot is not dependent on his particular relationships. Plus you'll have enough duplication of roles that a single missed person shouldn't provide a critical weakness.