Well, we're all just speculating, so I don't think we can really figure it out, without asking the guy. For what it's worth, though, I actually think the method I describe would be the least time-consuming. I just don't see how a video effect, from within a NLE software would be fast or easy.
On second thought, the method I describe, would not take that long.
1. Export jpegs. That takes about 30-seconds per jpeg. He clearly didn't do this at anything close to 24FPS, so overrall, we're not talking about that many jpegs.
2. Convert jpegs to mosaics. I've not used any mosaic software, but I can't imagine it is any more time-consuming than exporting a jpeg. Press a button. Done.
3. Place jpegs in timeline, in NLE software. If it were me, step 3 would take no longer than 10 minutes, for his entire video. Just put stuff in.
He got people to send him pictures of themselves holding up pre-specified colors of paper. It's rather ingenious, if you ask me.