Published in


Let’s talk about comp!

Talking about compensation in America continues to be weirdly uncomfortable. This feels like a taboo that is increasingly self-defeating, especially as our culture continues to suffer from worsening levels of economic inequality. In thinking about what kind of compensation framework we wanted to put in place for Proof, we revisited and ultimately rejected the assumptions that are typically used to justify secrecy of corporate compensation structures. We’ll walk through our reasoning in this post.

Unsurprisingly since we are computer scientists, we approach nearly everything as an engineering problem. If we are going to settle on a “good” compensation framework that reflects our values, we first have to define what we want the end effects of our framework to be. And then we will reverse-engineer what kind of structure and tools we need to achieve those desired end effects. So what do we think makes a compensation structure “good”?

Goal 1: Employee Happiness

Sometimes when employers say this, they actually mean “employee productivity” and/or “minimization of unwanted turnover.” Obviously these things are correlated, and we will need to maintain a high level of employee productivity and avoid disruptive turnover in order to be successful. But we are going to take employee happiness itself as the first order goal, because we expect it to be a better driver for the others than the other way around. Also, we expect our company to stay pretty small and mission-driven, so we don’t plan to attract or hire employees who would be very happy being unproductive.

Goal 2: Fairness

Some might question the necessity of this in light of Goal #1. If a system is unfair but everybody is happy, is the lack of fairness really a problem? We would argue yes. Perceived fairness or unfairness may be covered by the employee happiness, since employees are unlikely to be happy in a system they perceive as unfair, but we will go beyond this and make fairness itself (not just perceived fairness!) a first order goal. The caveat here though is that “fairness” is one of those words that people think they have an intuitive sense of, but is actually really difficult to nail down in many particular instances. For me personally, fairness encapsulates diversity, and that means diversity in every dimension — diversity of gender, race, socioeconomic class, age, experience, and ways of thinking. This will be hard to see in the early days while Proof is very small, but in the long term I will consider it a failure if our company is not reliably attracting and retaining a proportionate representation of diversity on any one of these dimensions. I have never seen compelling evidence to suggest that any “fair” system could coincidentally (cough) leave out any substantial category of people on a medium to long term timescale. This is the reason I am not including diversity as a first order goal — when I say fairness is a goal, I mean this to include diversity as a natural consequence. If we don’t see diversity emerge, then we aren’t truly achieving fairness. This is not to say we won’t be proactive on diversity directly — we aren’t so naïve as to think it will emerge from thin air. We will make an effort to identify, recruit, and retain diverse candidates for employment at Proof as an important piece of our efforts to achieve fairness.

Fairness also encapsulates rewarding and incentivizing employees proportionally for the work that they do. This means that employees performing similar roles in similar ways should expect similar compensation. Some systems for compensation (like the academic system I come from) have check point levels (like achieving tenure) that lock in higher levels of pay almost independently of future performance. This has negative effects from an incentive standpoint, and also tends to privilege front-loaded career trajectories at the expense of those who suffer work disruptions due to family responsibilities, changing fields, etc. It also can compound the effects of pay discrimination suffered in the past if new employers lock in salaries based on past compensation that was unfairly set.

That being said, the supposed benefit of a system like tenure is that it enables long term risk taking. Though I agree long term risk taking is a good thing to foster, it hasn’t been my experience that a tenure-style of system accomplishes this goal. The more typical industry approach here is to use equity as an incentive mechanism for risk taking. In our combined salary and equity compensation structure for employees, we will be looking to strike the right balance between incentivizing/rewarding performance in the near term and long term risk taking. We will be particularly mindful to minimize unwanted side effects like compounding past inequities or being unfriendly to a wide variety of career trajectories and work/life balances.

Goal 3: Mission Alignment

This is where we will impose a burden on our compensation structure to help us attract and retain employees who embody the mission and values of our company. We will also make sure that our compensation structure itself embodies the mission and values of our company. Proof is all about earning trust, and using data and transparency to shine a light on unnecessary barriers to entry that plague the financial industry and are detrimental to the interests of the long term investors that the financial system is supposed to serve.

Many employers recognize that trust is a crucial asset in keeping employees happy and in enabling management to make the tough decisions that need to be made. But just because trust is important does not mean it can be demanded — trust is something that must be continuously earned. At Proof we will apply that philosophy when it comes to our employees as well as our clients.

Sub-Goal 1: Employer Accountability

All of this leads us to a natural sub-goal of employer accountability. If we design a compensation system that has no built-in mechanisms for management to be held accountable, it is hard to imagine it will achieve our goals of employee happiness, fairness, and mission alignment over the long term, despite our best intentions.

So what are potential ways of ensuring employer accountability? Well, at the light touch end of the spectrum, we have stated policies and procedures that force management to justify compensation decisions in particular ways. Though I suspect this is likely better than not having any mechanisms of accountability, I personally doubt this method alone is highly effective. Human beings who have already made a decision (either consciously or subconsciously), are quite masterful at rationalizing it after the fact inside whatever system of constraints is imposed upon them. Also, this mechanism would not be very mission-aligned with Proof. Introducing the overhead of extensive policies and procedures as a weak substitute for transparency would be pretty opposite to Proof’s philosophy. Which leads us to:


I tend to believe that the main driver of effectiveness when it comes to accountability is who is ultimately empowered to enforce accountability. One appealing option here is to empower all employees to help enforce accountability. The simplest way to do this is to have all compensation practices and outcomes be transparent to all employees. In order for this to foster full accountability, both salary and equity must be included.

It is clear why compensation transparency is mission-aligned for Proof. It avoids the unnecessary overhead and layers of administration that would be needed to enforce policies in conjunction with information barriers, and it fosters trust between employees and management. It also builds in corrective forces for fairness: pay discrepancies that don’t sit well with employees can be openly discussed and addressed.

Given our goals and the small expected size of our company, this seems like a no-brainer.

And yet, many people have a very negative initial reaction to the idea of full transparency when it comes to compensation. Let’s now examine some of their typical counter-arguments.

Counter-Argument 1: Employees will be disgruntled upon learning they are not all superstars.

It’s not surprising that this counter-argument resonates with many people, because there is something deeply American about most people believing they are above average. And it is doubtless true that if a company that has historically had opaque compensation practices suddenly opens up their books to employees, that some (perhaps many) are going to be upset by what they learn. But in using this as an argument against pay transparency, there are two implicit assumptions that are worthy of scrutiny.

Implicit Assumption 1: The employee who is over-confident about his pay represents the common case.

I’ve used the word “his” here deliberately. A recent report on wage inequality for technology workers from Hired ( states that 54% of their female survey respondents claimed to have known they were paid less than a male colleague in the same role at some point in their careers. A much smaller, but significant 19% of male respondents claimed to have known they were paid less than a female colleague in the same role at some point in their careers. Interestingly, a majority of male respondents (53%) and a strong majority of female respondents (74%) claimed to believe that people are paid more or less as a function of gender identity.

This suggests that the American workforce is not really starting with a lot of trust in employer fairness, and their skepticism is warranted based on personal experiences. It would be surprising if most women, racial minorities, or other marginalized groups blindly believe they are currently fairly paid in a secretive compensation regime, when nearly all data, their own experiences, and the experiences of their communities suggest otherwise. And even a majority of men, it seems, suspects that gender inequality exists. Optimizing the entire compensation structure for the happiness and productivity of employees who currently overestimate their place in the workplace hierarchy and are unconcerned about pay inequities is not going to benefit the “typical” employee of Proof — at least not in our desired future when our employee population matches the general population somewhat proportionately.

Implicit Assumption 2: Transparency is being enacted at a middle point in a company’s existence.

There is also a strong body of psychology research that suggests that differences between realities and expectations drive human happiness/unhappiness. Since Proof is a new company, we can institute full transparency around compensation from the very start, aligning employee expectations with reality from day one, and avoiding the messiness of a transition from opaque pay practices to transparent ones.

There is also psychology research to suggest that people’s current expectations are formed with an over-emphasis on very recent data (this is called “recency bias”). This suggests that one way to foster employee happiness and keep expectations and incentives in sync in a transparent regime is:

Sub-goal 2: Time-proximate rewards

We plan to adjust pay and give award performance-based bonuses over smaller increments of time than the typical annual schedule. This will give us more dynamism in re-balancing our compensation structure, and allow employees to learn about pay changes at a time when they are most likely to clearly remember and appreciate the performances driving them. We suspect it will be considerably easier for a manager to explain to Bob that “Alice is getting a bigger bonus than you this quarter because just last month she built that awesome data visualization tool” when these kind of things are fresh in everyone’s mind. Alice herself might also take more pride in her work when she can see it is appreciated in closer to real time. When employees can see the payoffs of high performance being distributed, they will have clear and constructive incentives to up their own game.

Counter-Argument 2: Transparency will harm recruiting and retention.

Another argument commonly made against transparency is that it will tie the company’s hands when it comes time to recruit or retain a superstar. I do expect this will happen on occasion. But transparency is not just a negative in terms of recruitment and retention: it is a tradeoff. Some superstars believe in pay transparency! Some may even be lured because of it. I suspect that our transparency will be seen as an attraction by talented prospects who have cause to believe they were not compensated fairly at previous jobs. Also, as a mission-driven company, Proof is not likely to attract or retain (or want to attract/retain!) employees who are solely interested in maximizing their compensation. Obviously we want employees to make a good living and be rewarded for their talents, but we are looking for employees who put a strong positive value on values (pun intended).

Counter-Argument 3: Transparency isn’t helpful because employers already do a good job. The pay gap for women and marginalized groups is explained by other factors or too small to be important.

Wage gaps are exactly the kind of statistics that can suffer horribly from p-hacking, both intentionally and unintentionally. There are many confounding factors, many reasonable ways to slice the data, and not enough data points to support the necessary degrees of freedom. Most reasonable attempts to measure a pay gap between men and women in a male-dominated industry like technology find a small but significant gap, with women reliably getting the short end of the stick. The size of the gap varies by study, which tends to foster skepticism in the results. The 2018 Hired study, for instance, examined applicants for the same job at the same company, and found men were offered a higher salary 63% of the time, and women on average were offered 4% less in salary.

This might seem like not such a big deal. After all, the statistics are likely imperfect, and 4% is pretty small, right? And maybe, some argue, controlling for the same role/same company is not enough. After all, what if the female candidates are systematically under-performers even within this conditioning?

To address the maybe-women-just-suck spectre of an argument that always hovers around such discussions, let’s examine one final implicit assumption:

Implicit Assumption 3. A woman who makes it to the same role as a man inside a tech company is expected to be the same as (or worse than) her male colleague.

To see why this might be drastically false, let’s perform a thought experiment. We’ll start from the presumption that ability is distributed equally among men and women as they enter the pipeline of a technical education and career. Suppose at the next milestone, the top half of the men remain, but only the top third of women. At the next milestone after that, the process repeats, culling the remaining men in half, and culling the remaining women even more harshly. Before long, we have reach a rung of the ladder with very few women left, but the women who remain are of a significantly higher average capability than the remaining men.

This is why the “we’ve controlled for everything so maybe women just suck 4% more, and that’s pretty small so let it go already” argument is so unsatisfying. For women and other under-represented groups who have been disproportionately culled at every level of a vast hierarchy, the reality is that a 4% pay gap at a given level may still represent a huge discount compared to what their talent is actually worth. This is even more true for racial minorities and other marginalized groups.

Is this the only interpretation that is consistent with the data? No. Can we draw any firm conclusions from the messy state of compensation data for the technology industry in the US? No.


Though it’s not a panacea, we are convinced that transparency on compensation is a crucial step towards a fair structure that fosters employee happiness and reflects Proof’s values. Though this conclusion is rare among financial companies, the arguments we’ve made above are not in conflict with core market principles. In other aspects of the finance industry, many participants are calling for increased transparency and data-driven decision making: lit quotes are a public service! Market data fees and other trading costs should be more thoroughly disclosed! How can we trust execution quality/access fees/prices/regulation/etc if we don’t have the data? And these are arguments put forward by companies who don’t practice anything close to transparency when it comes to compensation.

Why should our employees deserve less than a fair employment market? And why do we suddenly abandon the concept of price discovery when it comes to people? Are we so afraid of what we might see in the light, that we should prefer to wander indefinitely in the dark? It’s not a sustainable bargain. It’s time to flip the switch.



Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store