> The person doing orders of magnitude more work is probably only getting paid 10% more.

You've answered your own question. Transparency is undesirable from an employer's perspective because it encourages a closer alignment between compensation and productivity, and the mismatch between these currently serves as a significant profit advantage to the employer.

If I can get 1000% more work from someone, and pay them 10% or 100% more, I am highly incentivized to prevent them from becoming aware of the value of their efforts, because they might realize their (relative) exploitation.

At the same time, the less-productive worker also benefits from this lack of transparency because they get ~90% of the pay for 10% of the work. Disrupting the status quo would not be economically rational for them either.

Thus, only the small fraction of underpaid, overworked people benefit from increased transparency, and it is not surprising to find that the majority would resist its implementation.

I’m not sure I agree with the basic premise that the majority of people are either paid fairly or overpaid.

I suspect the vast majority of people are overworked and underpaid. The lack of real world wage increases, especially when compared to every increasing productivity, strongly suggests that being underpaid is far more likely than being overpaid.

According to your own logic, only a much smaller fraction of business get any significant advantage.

What entitles business owners to all the benefits of highly productive workers?

This is exactly why many of those workers go on to start their own businesses, to redirect those benefits to themselves.

Eventually you’d have to leave or the boss would have to ditch pay transparency.

There is no world where 10X people will agree to a 50-75% pay cut, so the only people that will be left in these transparent shops will be the bottom half of the talent pool.

The most normal "outrage" response I've seen is "Bob makes more than Alice even though Bob is less productive".

Usually with pay secrecy the employer is covering for the fact that Bob had more leverage when negotiating salary and even though Bob is a profitable employee that they want to keep, Alice is a very profitable employee because she's cheap.

If Alice is no longer kept in the dark, she has both additional leverage and an incentive to leave in a fit of pique, neither of which you want for your most profitable employees.

How exactly? I don’t see any obvious path to this.

Imagine you're a business and your employees bake pies.

Imagine the costs are fixed beside the pie. Imagine the pie costs $1 in ingredients.

Imagine the average worker bakes 10 pies per hour, and your marginal profit per pie is $9, and you pay workers $10 per hour.

That is the labor cost and supplies for a pie costs $8.

Each worker, generates $80 in revenue per hour after inputs and wages.

However, you have one worker that bakes 30 pies per hour for $20 per hour. This worker is happy to earn double the average worker. And you, as a business owner, are very happy to make $300-30-20 = $250 per hour in revenue after inputs and wages (3x).

However, If the worker is aware of how much more s/he makes - it is possible to imagine the worker becomes even more productive.

Maybe the worker can bake 45 pies for $30 per hour (3x). You're paying more per hour per pie, but you're making much more profit per hour = $450 - 45 - 30 = $375 (4.5x).

Put another way, it will be very ... very difficult to reward good people with higher pay, without consequences. The onus would be on the business to articulate in legal terms, why X should get more than Y. This would open businesses up to lawsuits.

I suspect that the net effect of this will be that businesses are going to work much harder at vetting people, and more rapidly letting go staff members who are on the wrong side of the bell curve (say 1 or more sigma below peak) in productivity. I'm not sure this is a good thing, for workers, or for businesses.

As they say, the road to hell is paved with good intentions.

The other reason is that pay transparency is often anchored to specific credentials and characteristics, and it's often the case that credentials can be orthogonal to work performance. You might have a PhD who can't code who demands higher pay than a HS dropout who designed your entire infrastructure, and the HR department (or lawsuits, or etc) would force you to comply.

Such concepts punish high achievers.

The reason the vast majority of jobs don't work this way is because productivity is rarely: A) so easily measured B) so easily mapped to revenue. A 0.1x person whose maintenance work generates real revenue contributes more to the bottom line than a 100x worker in the same position whose project is scrapped for whatever reason.

You could also pay a salary and adjust it based on productivity. In theory, that's something like how raises and bonuses are supposed to work.

If you think your rating doesn't make sense, or your rating does make sense, but your pay is lower than people with much lower ratings - then you have a case to argue.

Of course - this could be just as nepotistic as the current system. But I think transparency is bad for nepotism long-term. Usually the most corrupt governments and organizations are the most opaque.

Developers aren't really that different from bakers.

You could bake 45 pies in 1 hour, and they could all be terrible pies that can't be sold, and you could burn down half the bakery in the process.

In the same way, a developer could ship a bunch of features, but they could all be "half-baked" and a nightmare to maintain.

You could also try to define metrics for what quality and productivity are. But people will just game that. Ratings can also easily be gamed or nepotistic.

In the same way, a developer could ship a bunch of features, but they could all be "half-baked" and a nightmare to maintain.

You can inspect the pies made by your staff. Terrible pies that can’t be sold either look bad (misshapen, burnt, leaking) or taste bad (too sweet, not sweet enough, salty, over/undercooked). It’s not difficult to figure out.

On the other hand, evaluating a developer’s productivity and quality of work would require a lot of code review which takes time and skill and also may require the developer to be present, wasting their time and lowering productivity. Push the review and eval process far enough and you really risk bringing down that developer to an unacceptably low level of productivity.

> Developers aren’t really that different from bakers.

I suspect very few people would agree with this.

Only with an answer to this question does what you have argued so far apply to knowledge work.

In this way - you can't just simply count how many pies a SWE coded.

But you still get paid to produce "something". Is it even relevant what that is?

Should you get paid less at Google if you work on a new product and you absolutely crush your job vs you work on optimizing ads and aren't very good at your job - but regardless are able to generate a lot more profits?

Maybe?

I don't have a good answer for this.

I'm assuming "productivity" is somehow tied to "profit". No one cares if you work really hard and get nothing done. But if you work really hard and do your job well, but your boss told you to work on something unproductive - what should happen? You and your boss are considered unproductive? Just you? Just your boss? I don't know.

I think the easiest fairest system would be a mix of levels like FAANG has along with ratings (which FAANG also has) along with transparent pay (which almost no one has).

From a business perspective - you're currently paid the least possible amount it costs to employ you or replace you. If you have a valuable and rare skill set, you'll get paid a lot more, even if you aren't particularly good at your job.

I don't see this ever going away.

That’s a categorical difference. Not some minor detail.

It’s also not the only difference. Knowledge workers contributions are necessarily much more interdependent than the contributions of people who make commodities.

But see that if you think of knowledge work as little different from manufacturing commodities, your arguments and conclusions makes sense.

I just think your assumptions about the similarities are not correct.

They don't need to. As long as the system is transparent and everyone agrees to it, what it is doesn't matter.

Performance based on lines of code? Based on peer evaluations? Based on users acquired? Based on meeting arbitrary goals your director sets that year? It doesn't matter as long as the evaluation criteria are predefined and available. Companies don't even need to use the same criteria, you can choose a company whose process you vibe with.

If you use lines of code, you incentivize inefficient boilerplate.

If you use peer evaluations, you incentivize positive social behaviors and favors, but not necessarily productivity or good engineering.

What you measure becomes the goal.

Being predefined and available only amplifies the effect.

If you think about this over a long enough timescale, and believe in something like the efficient market hypothesis, companies will converge on near-optimal structures that pay for the performance that is valued in their niche, and individual will be able to make informed decisions to work with companies that will pay them based on their strengths.

Note that I'm not saying that a company can't ever change its compensation and performance evaluation, so if LoC is a bad metric, over time I'd expect companies to move away from it as they find it to be gamed. But like, I'm not going to stop a company from picking a bad metric, or claim that transparency is bad because companies can make bad decisions. Companies can make equally bad decisions today, and you won't know.

I’d be very surprised if the game like companies you describe can outperform a well led organization.

> What if the better criteria for picking a good place to work is good leadership, and not a rule set?

I think an attribute of good leadership is being able to clearly communicate goals and priorities. If you can clearly communicate those goals and priorities in advance, you can evaluate people based on how well they perform against those goals and priorities.

If I'm forced to "anticipate what leadership needs", without any guidance on how or what leadership values, I'm doing the same job as "leadership", in that I'm defining organizational values, goals, and priorities. That's bad, because it dilutes leadership.

> You are making the assumption that an easily written set of fixed rules will outperform the more fluid and less easily quantified but nonetheless real heuristic of “anticipate what leadership needs”.

> I’d be very surprised if the game like companies you describe can outperform a well led organization.

I don't think I am. I don't think such heuristics are particularly easy to quantify or easy to write down, nor do I think they're particularly fixed. I do however think that most successful organizations do something like what I'm doing. As far as I know, Google, Microsoft, Amazon, and Facebook all provide more concrete performance expectations than "anticipate leadership needs". It would be hard to describe all of those as not "well led" or underperforming. In fact, most modern companies with more than a few hundred engineers develop some form of leveling system and rubric.

The challenges with these are that they often remain vague, since they need to apply reasonably well to every new-grad engineer working at the company, from someone working on client features in javascript, to someone optimizing the C++ compiler, to everything in between. But even still, you can usually get better, pre-registered answers from your manager or director that are more relevant and local: "If you complete this project on time, that should prove performance at rating X or at level +". How exactly this works depends a bit on the company (e.x. at Google, since peer reviews are important, having a senior peer or two whoa aren't your manager clearly state that something is L+1 work is probably more valuable than your manager stating it, and good managers communicate that).

This of course doesn't mean that you have no autonomy. Goals can be vague and high level. "Ship Frobulators by the end of the year" is a goal, and "we value end-user experience over implementation cost" is a value. If you know your company (or perhaps at a large company, your organization) has those goals and values, and you know the method by which they convert achievement of goals and values into compensation, you can better do what they want.

The alternatives are that:

1. Leadership is unable to express their goals and values in a way that clearly communicate what they want from employees. This appears to be what you're getting at with the example of LoC.

2. Leadership is unable to maintain consistent goals and values

3. Leadership doesn't want to reward people based on their stated goals and values

etc. etc. etc.

In the first case, they're incompetent. In the second case, they're unreliable. In the third, they're duplicitous and trying to compensate you based on some secret, second set of values, as opposed to the ones that they're claiming to share. None of those alternative are things that inspire confidence in leadership.

> 1. Leadership is unable to express their goals and values in a way that clearly communicate what they want from employees. This appears to be what you're getting at with the example of LoC.

An alternative you haven’t considered is that leadership is excellent at communicating what they want, but it’s not reducible to a simple set of incentive foals.

2. Leadership is unable to maintain consistent goals and values

Another alternative that you are not considering is that the world is dynamic and excellent leadership involves adapting goals as the world changes, not maintaining consistency for the sake of easy comparison with other organizations.

3. Leadership doesn't want to reward people based on their stated goals and values

This is a non-sequitur if we consider 1 & 2.

None of this implies duplicity or secrets.

Perhaps people can judge leaders based on results and culture without needing to see it reduced to a rule set.

Can you give an example?

> Another alternative that you are not considering is that the world is dynamic and excellent leadership involves adapting goals as the world changes, not maintaining consistency for the sake of easy comparison with other organizations.

I did consider this. I don't think maintaining consistency for the sake of easy comparison is a good reason to maintain consistency, but consistency in values is good because it changing values is damaging to an organization. It requires a cultural shift that usually causes deep pain and often requires firing members of leadership (I suggest reading https://apenwarr.ca/log/20190926 for a good post that influences my thoughts here and matches my experience).

If leadership is unable to maintain consistent values, that is a failure of leadership in my eyes. They are unreliable. Values can change, but they should do so slowly and rarely (and with good, and preferably transparent, reason).

If a leader is unable to maintain cohesive goals over a reasonable timescale, that's also an indictment of their abilities to plan within and understand the space they are working in.

Or to put another way, your organization will not develop a cohesive identity and culture without clear goals and values. Now oftentimes, you do have some clear values even if you don't state them, but then the question for leadership becomes "are those really the values you want"? And if you don't know them well enough to be able to state them, I'd suggest that the answer is probably not.

> Perhaps people can judge leaders based on results and culture

I'll reiterate: you cannot judge a leader based on culture if they don't declare what that culture is, or if their values and goals change too often. Values and goals set culture, and if values and goals change, culture changes. If all you know about a leader is that their values and goals change often, you can't judge them and their organization by how it looks today, because you can have high confidence that it will look fundamentally different soon. That's a sign of dysfunction. So I do judge them based on that: I judge them to be bad leaders. You appear to disagree, but I'm not sure how you are judging them if you can't even be sure of what their values will be tomorrow.

I don’t really know why you trust organizations that can write a statement of values down not to change.

Frequently these statements are written by consultants anyway, and organizations that do have good written principles that they live up to, do so because they have leaders of integrity, not the other way around.

I accept that you can’t see any other way to judge leaders than by examining their written statements.

I don't! But I do trust that it will be moderately more difficult (and far more transparent) for them to do so than an organization which chooses not to. If you have your values declared, people can hold you (or the org) accountable to them. If you then change them, you have to announce it, or people will continue to hold you accountable to the old one.

> and organizations that do have good written principles that they live up to, do so because they have leaders of integrity, not the other way around.

I agree! But good leaders also make it easy for their reports to hold them accountable, and it's far more difficult to hold someone accountable to a set of values if those values are unstated. That's why the values need to be stated. Like, please explain to me how you live up to the organizations principles if no one knows what they are! (and if everyone agrees on what they are, why can't you write them down?)

And I should clarify if it wasn't already clear, I'm not talking about values like "integrity" or "respect the user", but things like "bias toward action" and "move fast and break things", that give you insight into how decisions should be made in the org.

> I accept that you can’t see any other way to judge leaders than by examining their written statements.

I judge leaders by examining how their stated values correspond with their actions. Their stated values give me a framework for judging them.

How do are these qualitatively different kinds of value?

>> I accept that you can’t see any other way to judge leaders than by examining their written statements.

> I judge leaders by examining how their stated values correspond with their actions. Their stated values give me a framework for judging them.

An alternative is to simply judge their actions, how they treat people, and the results their organization produces by your own values.

The second set are tradeoffs, they give you a framework for decision making: "integrity" is obviously a thing you want, but everyone wants it, it doesn't really tell you anything about the values that a particular organization has in comparison to another. The second set are implicitly about tradeoffs: when on the fence, prefer doing something to not doing something, or when you're uncertain, make a change and deploy a feature even at the cost of potentially breaking things. This gives you information on how decision should be made and what the organization values (features > reliability, at the margin). When that value changes ("move fast with stable infra") it tells you that while the organization still values velocity over reliability in some places, when dealing with core infrastructure, you should work the other way, preferring stability over feature velocity.

This is explained better in the blog post I linked, I'm mostly restating what it says, but worse.

> An alternative is to simply judge their actions, how they treat people, and the results their organization produces by your own values.

But, and here's an important thing: a leaders' values may be different in different circumstances! The same person, when running a 5 person startup, and a 50,000 person company, will land on different sides of the same tradeoffs! Organizational priorities change. The values I'm describing are how you communicate those organizational priorities.

If you just try to judge a leader or organization based on how they acted in the past, you will be misled if circumstances have changed, unless you can read the leaders mind. That's why stating values is important. Again: they're how you communicate priorities.

How the leader acts and how they treat people don't fundamentally matter when you're trying to figure out if the work you're doing will be considered valuable by the organization. Like yes, you should work for leadership who you enjoy working for on a personal level. But that's totally independent from how good a leader they are. There are very nice people who are incompetent leaders, and there are people I consider assholes and wouldn't want to work for who do a good job of managing their organizations.

It seems like you're talking about personal values (and ethics) which are important yes, but not at all what I was talking about. I'm talking about organizational values as an aspect of leadership. These are in many ways totally distinct from the personal ethics you hold.

And importantly, I think with very few exceptions, trying to glean anything about organizational ethics from stated values is a fool's errand. But they are still important for stating organizational priorities. Don't confuse the two!

Only if you don’t take into account the changed circumstances. And if you’re going to ignore circumstances, clearly no set of written priorities will save you.

How someone actually makes decisions is obviously more discernible from the decisions they have made in practice than from a set of statements they make about how they do it.

> But, and here's an important thing: a leaders' values may be different in different circumstances! The same person, when running a 5 person startup, and a 50,000 person company, will land on different sides of the same tradeoffs! Organizational priorities change.

This is a point I explained to you earlier. I’m glad you now agree with it.

How can you take those circumstances into account without knowing what leadership values in those circumstances? And anyway, how do you even know you have full information? Like, the point of leadership, in a very generic sense, is to distill the context that I don't need to worry about down into instructions at an appropriate level for me to act on. If they aren't doing that, then what are they doing?

> This is a point I explained to you earlier.

You may have attempted to communicate this. You did not succeed, as I still don't know where or when you tried to make this point.

You might be. But I don’t really see how you know that about anyone but yourself.

How someone actually makes decisions is obviously more discernible from the decisions they have made in practice than from a set of statements they make about how they do it.

> Based on peer evaluations?

Based on their interpretation of how you hit metrics? Why not just look at the metrics directly?

I think it's okay if ratings are subjective based on loosely defined job responsibilities. That's how it works at FAANG.

Nothing is going to be fair. But I think it's better than just counting how many bugs you fix or how many lines of code you review or how many users your code acquires (which should be mostly determined by your product and design team's decisions).

There doesn't need to an objective, or even a best subjective, measure of productivity, as long as the company's chosen measure is transparent and preregistered.

It's just that with a transparent evaluation process, you know why that is and can make more informed decisions than having your manager say "no don't worry, you're on track for a promotion next year" every year.

To put it a different way: how do you know that you aren't being compensated based on LoC output today, if you don't have some amount of transparency in evaluation criteria (and, importantly, some amount of transparency in pay, so you can verify what the company says)?

In any case, this piecework example doesn’t seem to be relevant to this forum.

Can you describe an example that would apply to programmers or other knowledge workers? What about people who work in teams?

If one of your developers is consistently able to finish projects substantially faster than other developers - you as a company are supposed to accurately estimate project difficulty and thus pricing (of course this is very hard to do accurately).

Your input costs are just wages, and your revenues are how much you charge for projects.

A fast worker will slow down when measured to game the system - it's one of the reasons why those Amazon workers piss in bottles.

There are a lot of "industrial practices" that most HN commentators will not have come across.

Transparency in pay or transparency in value to the company?

The issue GP had was with disparities in pay and value. If you start being transparent on value, that gets into privacy issues.