It's a good question. Though CLC certainly didn't invent the concept, we've probably cut and shipped more of them than anyone. My guess is around 7,000 CLC boats with puzzle joints. That's a vast data set from which to draw conclusions.
Scarf joints are definitely stronger in absolute terms, because there's more surface area for the glue.
However, puzzle joints never let go at the puzzles; just like scarf joints, they break on the wood, alongside the joint.
So one way to look at it is that scarf joints probably give you 600% of the strength you need to hold planks together, and puzzle joints give you maybe 400%. Either way, the wood's breaking before the glue in the puzzle joint lets go.
In thinner plywood, where a puzzle joint offers even less gluing surface, you could start to wonder about it. However, all CLC boats made of thin plywood (3mm or 4mm) are fiberglassed on both sides of the plywood, which multiplies the strength many times. Indeed, you can make plain butt joints work perfectly well with just fiberglass on either side of the joint.
There are still occasions where scarf joints make sense, and we still cut scarf joints on a couple of kits. (WoodenBoat Magazine's Shellback Dinghy and Nutshell Pram, for example.)
One feature of puzzle joints that is an unassailable advantage is the ease of assembling hull panels accurately. Scarf joints or butt joints require a lot of care in alignment when you glue them. Before we switched over to puzzle joints on most of the kits, misaligned panels caused persistent and widespread trouble. Thanks to the forgiveness of the stitch-and-glue process, the boats would still go together okay. But the minor misalignments yielded crummy fits of bulkheads and other hassles.
Since migrating to puzzle joints starting around 2005, accuracy of the hull assemblies have improved massively. A huge reduction in assembly hassles. And more accuracy begets more accuracy; we've been able to add kit features that simply wouldn't have been possible in the scarf-jointed boats.
There are two downsides: First, CNC-machining puzzle joints is HARD. Even very, very expensive computerized machinery has trouble holding tolerance so that the puzzle joints aren't too tight or too loose. You have to slow down the machine to cut the joints, which really adds up when you're cutting hundreds of them per week. (An Annapolis Whery has forty puzzle joints, for example.) It has contributed to longer turnaround times for CLC kits, something we have recently spent a fortune to remedy.
The second downside is that some people just don't like the looks of puzzle joints. Even when they acknowledge the functional benefits, some people simply recoil at the appearance. With the votes running about 100 to 1 in favor of puzzle joints, we're all in. But we still have the scarf-cutting machinery so if it's a show-stopper for you, we can always explore a custom kit...

