Three knobs that pull against each other
Every oracle design choice covered in this course, how many sources report, how they're aggregated, whether updates push or wait to be pulled, ultimately trades off against three properties that can't all be maximized at once: decentralization, cost, and latency. Adding more independent data sources and more rigorous aggregation across them directly improves trust-minimization, a single bad actor matters less when there are more honest, independent voices diluting its influence. But every additional source that has to fetch, sign, and submit a report adds gas cost for the extra on-chain data, and adds time, since the system has to wait for enough independent reports to arrive and be aggregated before it can finalize an answer.
A single centralized source sits at the opposite corner of this space: it's cheap, since there's only one report to pay for, and it's fast, since there's no coordination or aggregation delay, but it reintroduces exactly the concentrated trust assumption that decentralized systems exist to avoid. Neither corner is simply 'better,' they represent different points along the same tradeoff, and understanding oracle design means recognizing that pushing hard on any one of these three properties costs you along the other two, not finding some configuration that escapes the tradeoff entirely.