To clarify, our video data is all crowdsourced - our community paste in URLs to the videos/slides. We then fetch the URL they give us and parse it to find the embed codes (mostly we just rely on Embedly for this, but we have our own fallback code as well).