The Invisible Hand Needs a Hand: Two Short Vignettes on Market Design
Top Choice
LinkedIn recently rolled out a Premium feature: users can mark up to three jobs per month as a “Top Choice,” making them 43% more likely to hear back. It’s a minor UI change but likely consequential. The feature draws from ideas in market design, including work by Alvin Roth, Atila Abdulkadiroğlu, and Muriel Niederle, among others, on congestion in markets.
The basic idea is that when everyone applies to everything, employers are flooded. Employers have an especially tough time separating desirable and interested applicants from desirable but not particularly interested applicants. (This is often why many employers have a recruiter screen.) Limiting applicants to just three expressions of interest means that the signal is costly, forcing applicants to choose carefully and helping employers triage more effectively.
Google's restrictions on applications are another example of reducing congestion. They force applicants to think carefully about which jobs to apply for.
The downside of these ad hoc solutions is that applicants are making the decision with limited data, and considerations like risk aversion, etc., may mean that the final matches (after accounting for assessment costs) are worse than the status quo. (In Google's case, an applicant may not even know about the clause till they hit it.) Also, note that signals from lower-ranked applicants are useless. The company can't distinguish desirability from interest. The same applies in dating markets. Now that means that the signals are useful if you have too many desirable suitors.
Batch Feedback
Airbnb’s mutual blind review system was meant to encourage honest feedback: neither guest nor host sees the other's review until both are submitted. In theory, this deters retaliation. In practice, it leads to soft coordination and inflated ratings.
Over time, users adapt. Guests and hosts coordinate via direct messages. What emerges isn’t truth—it’s a 4.8-star détente. Even when retaliation is off the table, social costs remain. Giving a negative review feels personal. Without anonymity, honesty declines. The result is reputation inflation.
A better alternative is batch-aggregated reviews, where individual feedback is anonymized and combined into rolling averages and thematic summaries. This design lowers the emotional cost of honesty, discourages strategic behavior, and surfaces more accurate signals about quality. It shifts the review process from a personal exchange to a shared informational asset—harder to game, easier to trust.
Interested in Learning More
Read my notes on Alvin Roth's dissatisfying book, Who Gets What —and Why: The New Economics of Matchmaking and Market Design, on market design (he considers only a small set of issues).