Is Dev Compensation Bimodal? (2016)
danluu.com
danluu.com
Why are programmers so well paid... in the USA? In the rest of the world that's not the case (programmers are paid a bit higher than other engineers but the difference is not that pronounced as it is in America).
Edit: And I would add: programmers are paid very well in the USA at FAANG companies. I would love to know what's the percentage of programmers that work at American FAANG companies (my bet is less than 1% of the world's population of programmers... but I haven't done the math).
It's higher, but I wouldn't say "a lot higher". I have lived in France, but I think the situation is similar (if not worse) in Spain, Portugal and Italy:
The majority of senior software engineers (more than 5 years of experience) get around 60K euros/year. It's more difficult to get around 80K euros/year but it's possible.
Now, the average non-software senior engineer in those countries gets around 45K euros/year. So, yes, 60K euros/year > 45K euros/year... but it's not such a huge difference.
It's not uncommon for guys from Polytechnique to go straight to SV, sometimes after multiple internships in the Bay.
For Industrial Engineering (the most common) it's €18k-€22k when you finish the 6-year degree (4+2), and if you go so far as to become "project director" with some extra qualifications you get €71k/year (which is literally a "director-level" position). I'd say that these are rare enough to be well matched with the few people making €100k+ working for a company abroad.
One of the problems is that there is a lot of self-taught developers, where they make that same "junior" salary of 18k-22k after 1-2 years of a bootcamp or learning on their own. So this brings the whole average down, but for the same amount of experience (studies+work) I'm pretty confident a software developer makes 1.5x-2x that of an Industrial Engineer for all of their _technical_ careers.
Effectively no one in the US gets a master's in CS, other than immigrants who get higher priority and possibility different visas (think like O-1) with said degree, so it makes sense for them. There are certainly PhD students, but they tend to genuinely want to do research IME rather than just do more school for more money down the line.
It’s also great incentive alignment, the higher you go the more you’re compensated in stock to tie your work to company performance. See: https://www.levels.fyi/company/Google/salaries/Software-Engi...
The same has been true for other trades at certain times, here in Australia truck drivers, welders and other trades could be making an absolute fortune working for the mining companies.
My pet theory that I don’t know how to rigorously test in data is that programmers are very well paid so they don’t go to the competitor or start their own competitor.
Programming software requires very few fixed costs outside of the programmers themselves; electrical engineers aren’t going to be able to make fabs (or get contracts with fabs), file patents and compete with intel the way software engineers can rent out a garage and make a social network with some novel twist and capture enough attention to get significant advertising revenue.
When you select for the tails of the normal distribution, the following distribution will have a pereto-like-shape. Since it's such a high skill field (abstract symbol manipulation to create automatons that carry out work), it selects on the right end of conscientious & intelligence distributions, which are both normal. That has a long tail, and incomes follow that distribution.
The more normal distributions you are cross selecting for, the more extreme the skew becomes. When you begin selecting for intelligence, conscientiousness, privilege of opportunities, sociability, creativeness, negotiation skill ... it gets more and more extreme.
Practice :)
If you look at it from the business perspective, devs are part of company that built the product, which they then go to market and make profits from. Profits that can be in the hundreds of millions. Why shouldn't the devs and the other workers (both engineers and the other company employees that helped make the profit happen), be paid well for it? And in certain companies where profits are in the billions, shouldn't those employees be making a lot more then? A lot more than what they're making now, that is, which means devs are not so well paid relative to the profit margin. The same goes for any profession too, waiter and line cooks at restaurants should be making more; their work directly contributed to the restaurant's profits, and should be better compensated than they are now, instead of this 12$/hr across the board without regard to profits.
If you look at it from the economic perspective, we can go with the claim that programmers usually make more than the average job, sure. But 'so well paid' is kinda relative at that point. We're 'so well paid' relative to professions that are not so well paid, and keep in mind that the other professions have been around a lot longer than software (ie waiters). Wages themselves have not really increased on its own for the working class over the years, a lot of it has been artificially forced in the form of minimum wage by LAW. So if everyone else is making terrible money, and we as devs are considered to be 'so well paid', then maybe we aren't really 'so well paid'? Basically the other professions are artificially being kept lowly paid, so in comparison of course we make a lot. But if we factor in appropriate wage gains over the years for them, devs should be making a lot more than what we have now.
Haven't thought of it much more than that, but the thought always comes when someone mentions that devs make a lot. Anyone else have these kind of thoughts?
We won’t get a bigger share of the value we produce by treating our white collars differently. We’ll get it by recognizing we’re workers (high paid often overworked) and find commonality with our peers, not our bosses.
Your labor is, by definition, worth more than you're getting paid. If it wasn't, your employer would fire you.
From the billionaires, I'm taking all I can get. They're certainly not going to give me any more than they can get away with.
SV also has a load of startups that reckon they are gonna be a FAANG someday, so they figure they'll take on some of the business practices, such as paying devs a lot of money.
The other lump is places that don't see devs as central to the business model, which is a lot of diverse places.
About supply, I'm not sure even with that upward curve of CS grads, that there's enough. A fair few of the people who take CS during a bubble are not in there for the long run. How many people do you know who sat next to you in a coding course and now can't write a Hello World? Strip those out, and you still get an upward curve, but a few tens of thousands of grads doesn't seem like very many.
I'm not so familiar with the other branches of engineering, but my impression is that many of them are in capital-intensive industries. If you need a chemical plant to be able to do your job, that will have an impact when you negotiate to get on the job ladder. You'll get more than your average wage slave, because the owner of the factory wants his capital utilized effectively, but he's also got a lot of leverage over the employees.
So why don't the FAANGS think this? After all, you can't work in ML without a load of machines? Thoughts to ponder.
You won't change the world there, but you will work on interesting projects, have good job security, and you can put your work down at the end of the day.
Source: Acquired by Oracle, left shortly thereafter because I believed I still wanted to change the world.
One is where you are a cost center where you get paid around $50k.
Another is when you aren't as much which is the ~$100k.
I don't disagree with your basic point. I just think there's likely to be enough variance in both categories that they may not resolve themselves into clear modes such as seem to exist in law, for example.
- They vest in 4 years in most companies, so do you do Salary + 1/4 of stock options?
- They are options and not even stock, so you are not getting much _at the moment_ you receive them. Is $1 of stock options === $1 of hard cash?
- Do you even discount the stock options accounting for risk? I know I do in my head when evaluating an offer (0 < multiplier < 0.2 depending how generous I'm feeling based on certain metrics).
- Refreshers make things messier, now you have salary + 1/4th * N_0 * val_0 + 1/4th * N_1 * val_1 + ...
- What about bonuses or incentives for referrals or similar?
- Transportation, food, etc. budgets?
TC = base_salary + bonus + current_year_rsus_granted
If you just want to budget your life, well, IMO just use base salary for that when looking at FAANG scale comp. The rest goes to retirement / brokerage investment / Tesla. If you need to proceed anyways, I recommend:
- if paid in Restricted Stock Units in a publicly traded company, discount the value by 20 percent - if paid in non-public shares or options, mark to whatever you could get for them if you sold today (typically 0 dollars) - consider all shares vesting this year, regardless of when they were granted -- summing your refreshers is not that hard - plan on 2/3'rds of last year's bonus. - don't rely on referral bonuses unless your job is "recruiter" - don't bother calculating fringe benefits unless you know the money would be spent similarly if you spent a few months on unemployment.
Options are harder because you have to predict an expected value and may not be able to actually exercise them early for cash: in that case I usually default to valuing them near 0 for yearly comp.
Options really aren't worth much unless the company becomes many times more valuable after the grant, and that's so unlikely to work out for the average employee that it may as well be $0.
They're clearly not worth anywhere close to $0, but if you believe they are, hey "free $20".
It by no means accounts for everything, but at some level the $100/month I get for transportation and other benefits is a fraction of a percent of my overall compensation, so I don't bother thinking of those.
Regarding options, I value those at $0 unless I'm joining the next hottest thing or IPO-imminent company. In those cases, I'd apply some discount based on probability of the exit, etc. Those are so hard to calculate, though. It feels more like winning a lottery than anything.
I started my life in startups and am three for three in "success" in that two were acquired and one is still chugging and making profit after 15 years. In all cases, net value of my options was $0, even when I owned ~1% of all non-founder options. The money I made in the exit was due to being the head of Engineering, so I got a cash bonus from the investor as part of the executive contract, and I got retention RSUs from the acquiring company.
If curious see also
2018 https://news.ycombinator.com/item?id=16337434
Discussed at the time: https://news.ycombinator.com/item?id=12588202
I submitted it and saw it sitting on 1 point for a while, then saw my karma jump up overnight and was pleasantly surprised to find it like this.
I'm disappointed at this section, it is missing fairly obvious things.
- Because programmers are disproportionately also living in the most expensive regions, so they may be paid well, but not as well once you consider the CoL. Imagine if you told 50% of plumbers/pilots/etc they would have to move to Palo Alto or SF to practice their career, those salaries would also jump.
- Because volatility. Anyone who's lived thru 2001 or 2008 knows that feast-and-famine jobs have to pay well to compensate for the famine periods, otherwise people wise up and dont take the deal.
(EDIT) - Because unlike with law/medicine/plumbing/piloting and practically most other fields, there are constantly new things introduced that dont really push the field forward much. Consider the 200 web frameworks. Some are great. Many are just rehashes of older stuff with slight tweaks. However, this has the effect of making experience moot -- because newcomers always have an incentive to create something new and level the playing field. Look at the Cravath salary scale on the site, and imagine if law vastly completely every 5 or 7 years. The salary curve wouldnt be as steep.
- Because they are like athletes and many have a shelf life. There is ageism for many, and for those w/o that problem, there is often a shelf-life for their technology skills. You are being compensated for continuous improvement and the risk that your skills become stale.
- Because scale. Like investment firms, one person in tech can have massive impact and collect some of the premium as salary.
I wouldnt know the real US wide situation, but I ve worked in continental Europe and Hong Kong, and we have nice salary but well far from 300k a year (I reached 100k this year with 7 years of XP, in an investment bank and I think Im paid at the average).
Programmers, in my experience, are still paid less than profit-incentivized workers like sales traders or trading floor quants. But you could say a quant is a programmer and skew the pay up I guess. Just dont let them near your source code (as a Quant you dont get the variable pay part of your big salary if you do enough quality to have version control :D)
I would rather trust someone who's been through a few paradigm, languages and ecosystems as compared to someone two simply memorized the invocations of whatever framework from the bootcamp-du-jour.
As usual it's important to keep in mind that this is a bubble. This is the SV bubble, where programmers get paid out of wheelbarrows filled with money.
Elsewhere programmers get paid about as much as other kinds of engineers, maybe a little bit less on average.
Maybe that's because developers outside the SV are garbage, maybe that's because there's a set of fairly unique and hard to replicate conditions in the SV that resulted in companies having to pay these sums.
That's not true. I'm in the UK, and while software dev salaries are considerably lower than in SV, they're still quite a bit higher than is typical is other engineering disciplines.
Makes sense though - as all of you and not just programmers are rather well paid compared to people in continental EU, the difference is not that huge. Where I am the average programmer has 5-10x more than any other person, and 2x more than directors - and the outlier programmers can make even 20x more than other people, and that's still ignoring the millionaires (and few $ billionaires) that sold their startup stakes.
In the past decade, only programmers became rich from 0 here - all other people inherited their wealth.
Shoot me an email at emteycz@gmail.com, I might be able to help you.
The EU is a big place and everyone's experience is different, but I think you're overstating matters hugely.
Or as a counterexample, there are plenty of industries that do not exist in SF because they don't have large enough profit margins to support it. There are no factories in SF. CoL comes from the high salaries, which comes from the high profits. Not the other way around.
I don't have a dog in that hunt. I have my land on a somewhat sane cost basis. My interest is academic.
The odd thing about software engineering is the number of these jobs that remain unfilled. As an example, Microsoft has 3,746 engineering jobs available plus probably another 1,000 technical roles in sales, service, and support. You'll find pretty similar numbers across the FAANGs and other major tech companies and VC start ups. So you've got lots of open roles and lots of developers that could double their salary but the connection is not being made.
Contrast with something like law. Law Firms will fill the roles they have with the best lawyers they can find. There's not a bunch of open slots at top firms that aren't getting filled because they can't find associates that will raise the bar.
At a public company the stock based compensation is nearly equivalent financially (ignoring taxes) to getting 4yrs of comp up front that for a period of time can only be used to invest in your employer. That exposes you to a lot of risk, but on average it results in higher pay since on average the markets are going up (not just in tech) and you're able to invest money you don't yet have (contrasted with, e.g., if stocks were awarded according to current market rate). That risk profile won't suit everyone, but it's not inherently bad.
> you're rewarded with a shitty 1br apartment because that is all your cash comp can buy
Rent isn't thaat high. A nice oceanfront 2br goes for $3k/mo including utilities. Plus, if you really wanted to dip into stock-based comp for rent, your signing bonus will more than cover the difference for the several months till your first grant.
> because you need to write and debug code on a whiteboard within 30 minutes.
Yeah, that can be pretty shitty. It works out for a lot of people, but I'm certain that process overlooks good engineers.
Sending a resume is free. Takes a few minutes to apply.
It's not like applying to college where, to be competitive you have to write a personalized essay (and pay the application fees).
"Johnny is soo good with computers, always repairing the wifi, he should go work at Microsoft.".
Relevant: https://blog.codinghorror.com/why-cant-programmers-program/
At a startup it means running out of runway. In professional teams, especially SaaS, it's bugs in production that starts taking more and more time that can't be spent on features.
>> Contrast with something like law. Law Firms will fill the roles they have with the best lawyers they can find. There's not a bunch of open slots at top firms that aren't getting filled because they can't find associates that will raise the bar.
Microsoft often calls me also. And i ignore the recruiter calls. The reason they have "3,746 engineering jobs available" is probably because it is a raw deal compared to the alternatives -- especially once you account for moving to Seattle and paying Seattle rents. They would have to increase salaries, and poof, all the openings would disappear. Alternatively, they can make the jobs WFH and they would disappear, because i'd be very interested to apply to these jobs if it didnt involve relocating to Seattle, I could live somewhere with sensible housing costs.
Anyone can create underpaid openings and they will go unfilled. I want a personal trainer for $5/hr. Except it doesnt make sense for a personal trainer after accounting for costs, especially given alternatives.
>> Contrast with something like law. Law Firms will fill the roles they have with the best lawyers they can find. There's not a bunch of open slots at top firms that aren't getting filled because they can't find associates that will raise the bar.
Look at the Cravath pay schedule. If you paid these sums to SWEs, those positions would also get filled really fast.
Sure, but these positions are not underpaid. They are often 2-3 times what other software engineers make. Location can explain some of it, but in other industries that stuff tends to equalize.
>Look at the Cravath pay schedule. If you paid these sums to SWEs, those positions would also get filled really fast.
That's pretty close to what the FAANGs pay and they're sitting on thousands of open positions. I doubt Cravath even has one open slot.
The positions are obviously not underpaid at face value nor are they underpaid on an absolute scale -- but they ARE underpaid relative to what is expected (experience, skills), and what has to be given up (cost of living.), and what you actually get (cash comp vs stock w/ vesting periods)
Cravath pays cash salaries, not step-vested stock that can fall long before it vests. There is no comparison. If there were a comparison, people would be flocking to these FAANG jobs. The fact that there openings again suggests there is something amiss.
As far as I can tell engineers are falling over themselves to take these jobs. It's the FAANGs et. al. that are rejecting them for, as you note in your other post, rather arbitrary reasons. It's an extremely unusual situation.
Remember that some of that top talent are well into their career, they have families, kids with health costs, school districts, college funds, etc. So its a raw deal to offer them just enough to jam them into a 1br or 2br apartment. Why would someone with years of experience and family obligations take such a terrible deal?
The salaries are great for entry level graduates, but in many cases the requirements (either explicit or via tests) are not selecting for entry level graduates.
W/r/t the requirements, I'm not even sure being able to rebalance a B-Tree on pencil and paper -- without a computer or debugging code -- is reflective of any real situation i've ever encountered in my career. It is a fair test if you just want really sharp people...but then you better be offerring really sharp salaries.
A lot of people don't know anything about FAANG, think it's unreachable or hold untrue beliefs about working for a FAANG company.
People are not rational, and engineers are still people.
Another question is if you joined a FAANG in 2013 and stuck around until now (or hopped between FAANGs), would your pay have kept up? Trivially speaking, yes. A stock refresher you got 4 years ago would have gone up: 2x for FB 2.5x for Google 4x for Amazon
You would easily be making 450-500k as a senior engineer with appreciated equity (or, you know, making that in cash at Netflix).
If you joined Facebook in 2012/2013 and stuck around there's a very decent chance you'd be a Staff engineer now, not Senior, and making closer to 600-700k/yr.
This is all ignoring the fact that you didn't need to spend 3 years and tens or hundreds of thousands of dollars on law school, which actually understates the opportunity cost, since an engineer joining FAANG will be at least 3 years ahead of a corresponding lawyer on their career track (including raises/promos/appreciated equity), so the more fair comparison would be to an engineer who joined in 2009/2010, which tilts it even further in favor of the engineer.
Entirely apart from all that, I find the question of whether engineers are "appropriately paid" a bit confused. There may simply not be enough engineers with the correct skills/talent/etc (whatever FAANG is looking for) at any price to fill all their open headcount, so increasing compensation would be a zero-sum game (for the companies, obviously).
I think the more boring truth is a combination of a few factors: 1) engineering at that level is actually sufficiently difficult that most people (when considering the entire population) can't do it 2) most people don't know what engineering pays - even outside of FAANG, hitting 6-figures in most major metro areas is pretty trivial 3) most people don't optimize their life/career to any appreciable degree, so even if they did know, they still wouldn't be particularly motivated to get into software (you'd see a lot of people saying that they're not smart/talented/whatever enough, which would be true for some of them, but the claim would mostly be driven by rationalization, not from any actual evidence they have)
To use myself as an example, I started out as an engineer making under 50k/yr (in a minor tech hub in the US, not the Bay Area/NYC/Seattle). Outside of talking to my parents, both of whom are in the field (though both slightly adjacent to the central conception of "software engineering" on HN), I had no idea what the market looked like. Maybe a year or two later I met someone who worked at a reasonably well-paying tech company (though not quite at FAANG levels) making ~200k/yr, which I hadn't even realized was possible. They told me I could be making that kind of money in a few years if I studied & interviewed at the right places. I didn't really believe them. Fast-forward a couple more years: I'm wasting my time in a (different) dead-end software job, making not-quite 6-figures, doing maybe 5-10 hours of work a week. I stumble across Dan Luu's blog and patio11's article on negotiation in quick succession, which light a fire under me. Six months later I'm working at a startup for a bit under 150k/yr. Six months after that (it was not the best place to work) I'm working at a (non-FAANG) tech company making a bit under 300k/yr. I've been there for a bit over two years now and my current annualized comp is over 500k at current stock prices. None of this is terribly unusual; anyone who joined FAANG a couple years ago has seen nearly as much stock growth (and better secondary benefits). If you joined Snapchat a couple years ago as a Senior eng, well, you'll probably be set for retirement by the time your 4-year grant runs out.
Something similar is happening at Microsoft. It is not like no one is applying to Microsoft. There are likely hundred times more applicants than there are employees. Good engineers too. But if they can't invert a binary tree on a whiteboard they can fuck off. And then Microsoft will claim they can't fill the position.
For law, it's pretty obvious. There's the partners at white shoe law firms and then there's everyone else--many of whom are doing personal injury law or scraping by doing routine wills, real estate, etc. Sure you have in-house corporate counsel and the like who are somewhere in the middle, but it's not hard to imagine why it's generally bimodal.
With programming/software engineering OTOH, yes you have outsized salaries at some big firms mostly in some high CoL areas, but it's a generally pretty well-paid field even if it does tail down for developers at non-tech companies in low CoL areas.
Software scales massively. Other industries do not.
You are simply paying for the cost of innovation. Whole categories of products created overnight. If you can afford it, it's best to get the best possible engineers onboard. Else your competitor will.
Anyone remembers BlackBerry?
That's not what bimodal means, it just means there are two modes (i.e. peaks). It could equally well mean that the general case is pretty well paid, but there is a distinct peak of high earners (e.g. FAANG etc.)
I can foresee a talent grab happening in the next years where the best devs in KC all start working remotely at 2x what they are currently making.
Again, this isn't what bimodal (or multimodal in general) means. There can be lots of middle ground, it just doesn't end up with a single mode.
To turn this around the other way, is there a reason that we would imagine a unimodal or even normal distribution for salaries here? If so, why?
In tech there are far more factors:
1. What industry you work in
2. How big your company is
3. Geography (this plays a much bigger role than in Law in my experience)
4. How good you are at negotiating
5. Equity luck
6. What type of tech you do (front end vs backend vs ML etc)
I wonder how engineers would do with the same sort of structure? If tech firms had to be owned by people who actually participate in creating tech, how would that change us? And how would it change what we build?
A difference between programming and engineering is that you can't really use a team of 50000 electrical engineers. Also, the number of EE's in an organization can't grow to the point of controlling the business.
There may be some kind of positive feedback cycle where programming takes on a life of its own within a given business, and demand for programmers expands exponentially.
Also, I think the author underestimates how hard it is to learn programming. It seems easy to those of us who program, because we somehow got past the "wall" at some point. But for most people on the other side of the wall, it's prohibitively hard.
(The other answer to the problem is that looking outside of the upper-tech bubble programmers aren't particularly well paid at all. But the linked article does make a decent case that there's a real mystery here.)
I feel sites like levels.fyi and Blind have given us a better look into the salaries of top (and not so top) paying companies. However they are missing the vast majority of companies paying modest salaries. I suspect these make up the bulk of the lower peak of a theoretical bimodal distribution.
Because for now, we are necessary tools in the enterprise of automating wealth concentration at scales never previously achieviable.
I may be able to tell myself; Oh I don't do any of that directly, but I benefit from those that do none the less, through pay inflation of those who are bent the same way I happen to be.
You can have a dictatorship, but you will still need people / your crownies. The day the drones are smart enough that they don't need anyone to maintain them and adjust them, maybe they start getting ideas about that one meatbag in charge of them.
Risk labour is simply a convenient myth.
Unlike the wage earner who loses his house and health insurance the moment he is fired ....
Suggested reading: "Capital in the 21st century" by Piketty. He presents data and arguments for why this is a major trend in capitalism, and suggests a progressive wealth tax to ameliorate this.
Note that I'm not advocating for or against it.
As a framework, it's stretching its usefulness when applied to inequality in labor markets, and that would include bimodal compensation of developers.
https://en.wikipedia.org/wiki/Capital_in_the_Twenty-First_Ce...
I've had the book recommended to me several times by "I'm happy to pay more taxes"-types as a one-stop shop for societal ills, but it's not that simple.
The seeming lack of attention on land and other rentseeking also puzzles me.
Wages don’t keep up with productivity.
Software development is a field where supply and demand apply without debate. Supply and demand are only ineffective when there is a form of monopoly / regulation in place. Which is not the case here:
1) It's very easy to make a startup software company (in comparison to others types of companies), and there is a healthy pool of existing company
2) CS education is very popular and we have a flow of new graduate every year, on top of that self-teaching is quite accessible.
Now, that means that the market is leaning toward demand (for programmers). Being "necessary" have never being enough to have a high salary, there are countless jobs badly paid at very wealthly companies, you must be necessary (demand) AND difficult to replace (supply too low) to have a high salary.
Some reason of the current trend:
- Very low supply of very high quality that a lot of projects requires: there is a huge difference between "programming" and actually doing software engineering at scale. And if you add some specialities like ML or security, the pool of candidate is getting very low.
- Rush for leading certain fields like AI, Autonomous vehicule, blockchain, etc... Those rushes are so fast paced that it drives the demand up.
- There are so many new startups, as I said earlier it's one of the field where making a startup is the easiest, and a good amount of them have strong ambition and investments (again in comparison to other fields)
High performance software engineers are rare and valuable.
That is the reason why some software engineers make tons more money than others.
Because the companies paying have silly amounts of money and can afford to pay them silly amounts of money. See also: bankers.
- the cost to start delivering value is effectively nothing (a computer and access to the internet)
- the cost of delivering value is small (you need to provision hardware your stuff runs on, sometimes) which gives you high margins
- you are not limited to one geographical area nearly as much as other engineers are (since you don't make anything physical, and legislation is relatively light)
- you don't need to wait a lot of time for dependencies like you do in manufacturing
You can go on and on.
The dynamics in software are different obviously but the core truth is the same: good engineers make a difference and the return on investment justifies the cost.
Just think about Fritz Haber who (with Carl Bosch) created nitrogen fixation, which completely changed our agricultural landscape. That is actual, real value (=increased food production) we all benefit to this day. Or look at Alexander Fleming, Edison, Tesla, Nobel, etc. this list can go on forever.
Yes, Linus did good on Git and Linux, but I think with RCS/CVS/SVN and Windows/macOS the world would not have lost that much value.
Bankers see powerful staff as risks, they much prefer having subpar staff that are guarded by rigorous process and management, as that is viewed as less risky. A single staff member who knows a lot and controls a lot of systems means they have dangerous leverage over the business.
Tech companies don’t view labour as a risk, they are part of the product.