If you want a serious answer is because peer to peer webRTC doesn't scale beyond two people.
If you have 4 people every client would need to maintain 3 streams, a total of 6 streams between all participants.
To have any kind of scalability you need a proxy in the middle that can act as a single stream to each participant.
This middlebox can also handle normalisation, interpolation and other useful features you might want to smooth things out when clients have connection difficulties.
Why don't we have a Free Open Source webRTC proxy server implementation? Because these days just publishing a protocol isn't enough for adoption; not to mention that proxying large amounts of data will incur a significant cost.
And that hasn't covered the need for authentication, which is yet another required service.
And if the proxy box is interpolating and gracefully handling frame drops, does that mean it will be handling gasp decrypted video traffic? Yes, it will, unless you want to move all that to the client and then have a key exchange happen not just at the start but a renegotiation every time someone connects or disconnects.
So you see it's not as simple as, "everyone just opens this URL, webRTC is a thing duh".
In theory one could implement the "supernode" model using WebRTC, turning the client with the best connection into a middlebox as well. In practice I suppose many meetings don't have any clients with a connection that could support that kind of bandwidth requirements.
