By contrast, few people -- not even politicians -- believe that having more politicians would be good for the world. They probably believe that different politicians would be better for the world, but not simply more. Most money and effort in politics seems to be attacking politicians that oppose your interests, rather than helping people that support your interests. The zero-sumness of politics and the relative ease of attacking over defending makes that inevitable.
Part of what makes YC work is that most founders are happy to help most other founders, except for a few direct competitors. If you pick 2 YC founders at random, they'd be very likely to want to help and very unlikely to want to harm the other's business. If you pick 2 politicians at random, how will that go? That's the primary value-creating process of an accelerator, so it matters. You only get Metcalfe's Law network effect if most interactions are positive.
You're proposing an EA focus, which is an essentially progressive set of values so you won't have many conservatives. Perhaps more would be willing to help each other than in national politics. Still, even among EAs, my impression is that many of them oppose each other's goals.