Give me a solar panel and battery any day. No moving parts to fail. Small, compact, and a single 1.3AH, 3.2V LiFePO4 stores 14000 joules that I can recover 100% of or about 40 pulls of a 25kg weight. I guarantee that battery will likely need to be replaced less often than these units.
I'm not so sure about the service life being so short. There are all kinds of sculptures with motor-driven joints that move far more often than twelve times per day for decades. And properly-made plastic-housed device can be quite durable. Just look at many modern firearms with "polymer" bodies. At reasonable loads, even nylon bearings work for plenty of service hours. I have to admit I haven't tried nylon bearings with 25kg weight - Only 5kg. The gears concern me though, as I've had (poorly specced) gears fail in "polymer" (And metal) drivetrains.
I'd like to see a version made of scraps. The benefit here (And in the lights) is that you make it out of things that are laying around. Unless you're in a semiconductor lab, you don't have solar panels and LiFePO4 cells sitting around, nor do you want to wait to see if your village is one of ten thousand that gets a light.
[There are economic considerations for and against shipping in electronics, but that's neither here nor there]
The plan of using some know-how and scrap parts (small motors from broken things?) to create lights is a huge benefit to areas that may have cell coverage and little other infrastructure. And in the middle of unlit areas, 0.2W of light is quite a lot, especially if it's built of scraps and an imported LED. Having them produced and imported won't serve as many people, but it sure makes good press.
Solar panels have some good selling points, but battery maintenance AND low system cost rarely come together. The cycle limit of the cell can be helped at the expense of system cost. The charging capacity can be cut to reduce cost, at expense of cell health. Mechanical to electric could be easy to debug and fix, but not built the way their prototypes show.
