Critique an Interleaving Proposal for a Search Team

A search team proposes to use team-draft interleaving for everything: "We will interleave the old and new ranker, count clicks per team, and ship if the new one wins more than 50 percent of queries. This will also tell us the revenue impact and the effect on page load time, and we will use it for the new carousel layout too." The test will run on 1,000 queries per day.

1. Explain what team-draft interleaving is and why it is more sensitive than an A/B test.
2. Which of the team's claims are wrong, and why?
3. How would you test whether win rate is significantly above 50 percent, and what do you do with ties?
4. Give a defensible end-to-end evaluation plan.