This is not how things work on the modern web. Did you just wake up from a 20 year coma?
poVoq
This is actually a feature missing from many browsers. Maybe Discord has a workaround, not sure.
Movim.eu recently added it, but it only works on Chromium based browsers.
Can't hurt I guess?
Long, but worth reading 👍 Also explains a lot of interesting details about the European enlightenment period.
Starting point based on what? Do you plan to use it personally to join many large channels from FOSS projects? If so, you need to plan for a relativrly beefy VPS, like 4 core, 8gb ram, 100gb+ ssd storage.
But for small private chats with no federation a cheap minimum VPS can work.
If you want my personal recommendation I would avoid hosting Matrix. A well federated server is costly to run and not really worth the hassle and a small private chat server with bridges etc can be done equally well or better with XMPP.
You can absolutely share your videos and livestreams with it. Technically it is all working well and also includes nice tools to migrate or mirror your content from Youtube etc.
But there is no real audience and even relatively well known tech vbloggers often get only low double digit views on their videos.
As others have already alluded to, Matrix is a bit odd in that regard as it runs a distributed database and the resource requirements depend on how much of the matrix network is mirrored on it. A single power-user can cause huge resource use just by connecting to a lot of federated active rooms. On the other hand a server that is mostly used as a private family chat can run on a modern RasberryPI without much problems.
Synapse or Conduwinity etc. makes little difference in praxis as both need to do the same database merging operations.
They removed a lot of cruft and low quality code from Pleroma and in general it is developing at a nice and steady pace. In addition the developers are not right-wing nut jobs.
Akkoma is better. Look for those.
This will just make them sound more believable when they hallucinate. LLMs can conceptually not be made to not lie, even if all the info they are trained on is 100% accurate.