Arguably, the complexity of getting the blob of fissile material should count as complexity. I'd argue that the simplest reactor therefore is the kind you can make with natural material, so a beryllium, heavy water, or graphite moderated natural uranium reactor like CP-1. Even then there was much complexity in getting pure enough graphite.
Well, that was 2 billion years ago. Since U-235 has a shorter half-life than U-238, the natural uranium enrichment was far higher back then. And indeed that was a water-moderated reactor. [1]
Some people have postulated that the moon itself was formed in a nuclear fission excursion [2], and/or that there is or was a nuclear reactor in the center of Earth [3]. These are both not commonly believed.
