Scalable Trust: A Remedy for the Selector’s Curse
I live in New York City and almost never help strangers who approach me. However, I will sometimes stop to help an elderly person who looks lost, assist someone with a wheelchair, or guide a confused tourist studying their phone. But the moment someone directly asks me for help, my guard goes up. I’ll either ignore them or decline politely. Selection effects will follow you everywhere you go.
If too many people read this blog, I’d expect more strangers to start acting helpless to take advantage of people like me. Similarly, when you make your selection criteria explicit, whether for dating apps, job interviews, or standardized tests, you invite adverse selection. Moreover, people respond by modifying their behavior to meet your criteria, even if it’s temporary or deceptive.
You like tall guys? Great. I’m now 5’11.
Want to hire someone who can write SQL? Good. My resume says “SQL” five times. Let’s see if you can find the fifth one hiding in white font.
Think the GRE reflects academic potential? I’ll spend every waking hour gaming the test to make sure it doesn’t reflect mine.
Consider the dating app example. If the height discrepancy is barely noticeable in person, misrepresenting one's height might be an adaptive strategy. The person might correctly intuit that even women who screen for height aren't likely to be height supremacists who will bring a measuring tape to a speakeasy. If he charms her—which he's confident he'll do—then when the measuring tape does come out, it won't hold a candle to his cheeky sense of humor, which by then she has hopefully come to thoroughly adore. While overstating height significantly might not increase odds by much, this isn't necessarily a deterrent. But if most people think this way, the screening criteria—which was presumably a way to cull an unmanageably long list of suitors—becomes that much less effective at culling.
Any time you set up a process and state the criteria, you have now frozen your degrees of freedom while dealing with adversaries/counterparties who can react dynamically to you. Part of what makes this problematic is that parties don’t have perfectly aligned incentives across time even if one generally conceives of dating or employment as positive sum games. For example, interviewees might tell you what you want to hear on cultural fit if they can just trick you into paying them for three months and then leaving you for what they really wanted to do. Relative power always changes with time.
The party with selection power almost invariably sees their relative power diminish over time. This dynamic actually explains why they need to be selective in the first place. Consider the dating market: women generally hold more selection power as a group, but this selectivity exists precisely because they face greater potential costs and risks after romantic engagement. Similarly, employers exercise careful selection because removing an employee becomes increasingly costly over time. If companies could terminate employees as easily as they can reject applicants, they would logically invest far less in their selection processes.
But I have another intuition which I’m far less confident of: setting up a process with increasingly selective stages ie gamifying something also elicits interactions that are good for neither party. For example, a candidate might misrepresent themselves or apply to 150 jobs without looking carefully even if they’re wasting everyone’s time and they are likely to get eliminated later on in the process. Why? because there is psychological payoff in doing something concrete -like applying for a job, and even more in advancing to the next stage of something. You get to mark your achievements to market. People love to make their wins legible and keep their losses illegible. This is partly why I think people prefer spraying and praying on job boards and dating apps instead of networking and meeting people.
At the heart of these formal selection processes lies a fundamental tension between trust and scale. To understand this trade-off better, consider a contrasting example: how I found my co-founder.
A couple of months ago, I met my now co-founder at an EA/rationalist event. I wasn't necessarily looking for a co-founder at the time since things were very much in the idea stage. But we seemed to both have thought about the problem in similar ways and generally got along quite well. I also gathered from a few other things he shared that he was entrepreneurial and is the kind of guy who does tech side projects on the side. Then when I'd made a bit more progress and decided to spend more time exploring it, I sent him an email asking if he wanted to test out working together on the idea.
Contrast this with a selection process you would run to hiring your 15th engineer or product manager. You certainly can’t justify testing fit by working together or even having long informal conversations with more than a couple of people. Even if you can justify testing fit in some intensive way, you’ll have to figure out how to shortlist those two people. How do you do that?
In my co-founder example, the event at which I met Will was an event I chose to attend because it was interesting and I was likely to meet interesting people who like discussing ideas. So the pool was already selected appropriately for serendipity to do the rest of the job. (As an aside, I saw this tweet the other day that claimed that the best people are SF coded people who live in New York. Similarly, I’ve noticed I get along best with people who are at the periphery of EA/rationalist movements but not card carrying members themselves). Also, organic interactions have lots of signal, more so because they aren’t designed to facilitate a match, which tends to introduce distortions.
Admittedly, there are ways in which finding a co-founder is different from hiring your 15th employee. First, the relative power of each party remains constant. The exchange of costs and benefits are not dispersed across time in a way that creates temptations to misrepresent. Even if the 15th employee isn’t as critical as picking the right co-founder, each employee can still have a significant impact on output and probability of success. So how can one guard against these problems of selection and get a great cultural fit?
We can segment the selection process into implicit and explicit. The people you evaluate explicitly have been implicitly selected for through whatever channel you’ve used. If you could at no cost and through some magical technology, obtain the information that is the equivalent of spending 10 full working days with all the candidates in your application pool, you would obviously do it, and maybe even pay good money to do so, since translating this information into legible forms like text or even video results in significant loss of dimensionality. (I wrote about this loss in one of my previous posts)
Meeting someone in person is a high bandwidth information channel - you’re able to gather multi-dimensional data about them, even if most of this is intuited by you. The thing with high bandwidth channels is that they can probably be deceptive if you only interact with a person for a few minutes. But over time, your ability to predict their behavior, understand their decision-making patterns, and gauge their reliability becomes increasingly accurate.
This seems intuitively related to what economists and political scientists mean by "trust" at a societal level. A high-trust society is one in which members can engage in complex transactions and cooperative behaviors without prior personal interactions or formal enforcement. This is possible because shared norms and expectations function as a compression of what we'd learn through extended high-bandwidth interactions. In essence, trust helps us circumvent two problems: the need to force legibility (which often reduces the dimensionality of information and invites adverse selection) and the cost of lengthy high-bandwidth interactions.
When you're hiring someone, you can't have high-bandwidth interactions with everyone, but you can leverage people who have—if you trust them. This trust manifests in shared understanding: when I tell a friend I'm looking for someone "driven and focused on output," they intuitively grasp the full meaning. They know I want someone who can handle ambiguity and act without perfect certainty, not someone obsessed with irrelevant details. This implicit understanding exists because we naturally surround ourselves with people whose norms and values align with ours.
This points to two practical constraints in leveraging your network for hiring. First, you quickly exhaust your immediate circle - there are only so many qualified candidates that any person knows directly. Second, even your closest contacts grow fatigued from repeated requests - there's a limit to how many favors you can ask, regardless of how strong the relationship is.
This is fundamentally what we're trying to solve with Clout—the challenge of scaling trust beyond your immediate network. While your direct connections might be limited, each of them has their own network of trusted relationships. If we can create the right incentives to activate these extended networks while preserving the quality of judgment that comes from trust, we could significantly expand the pool of well-vetted candidates.
If you're hiring and want to explore if we can help, you can email me at vaishnav@cloutcareers.com. We're happy to compete with your existing channels - at worst, you'll get an additional source of candidates, for no additional cost.