Raw sample size means very little when you're wanting statistically significant information. A properly prepared poll can get a statistically representative response from as little as 100 responses as long as you're not needing something like 1% margins.

On the other hand, a twitter poll will self-select based on that userbase. It won't be an representative base of the country/world... but it will give a statistically representative response for the group of users still on and actively using the MechaHitler site. MechaHitler social media users giving a 96% response like this is insane. Twitter is nearly as far from a left wing bastion as you can get, other than Truth Social.