Ideas are not necessarily worthless. Some people think of game ideas as "imagine Mario.. mixed with.. Civilization!" sure those are totally worthless and take 5 seconds to come up with. But there's also "imagine a game where you have chess but pieces can move in 5D, and the piece movements can be generalised in this specific manner with these rules". Those represent a ton of work to discover, and outside of games, research like this is protected by patents.

I think the decision on 'should I share this early' can be broken down like this: Ideas come in two scales, there's 'how well the idea stands on its own' but also theres 'ease to replicate'. If your idea is high on both scales, it is a good idea to keep your cards close to yourself until you're ready to release. A game like Superliminal is OK to reveal early because most of the weight is in the ideas around the core concept. A game like Beat Saber is better kept under wraps because the core concept is where most of the weight is.

Imagine the Beat Saber devs showed a very early prototype, someone like IGG saw it and was like "yes lets put our 1000 employees on making this immediately!". They'd beat Beat Saber to market and everyone would have seen their game as the innovation, and see Beat Saber as a copycat. It wouldn't matter if Beat Saber had much better implementation as long as IGG didn't make any major screwups. It's the same way tons of BS clones came after BS and none did anywhere close to as well. The Beat Saber devs understood all of this and appropriately kept everything secret until it was fully polished and ready to release.