This is the profit centre vs cost centre problem. Companies are incentivized to spend more on developers, since developers build features, and features drive sales, and sales drive profit.

SysOps folks keep the site up, but once you start hitting >99% reliability, throwing more money at sysops isn't going to increase revenue.

From a "job security" lens, the sysops crew needed to keep the site up will be the last to go during downsizing.

There are plenty of managers with the "why are we paying them when everything just works?" mentality.

Yeah, firefighting is a more visible show of "value" to those not directly involved than reliability is, because nobody at too many places likes to ask why we're always fighting fires in the first place.

Not always. I was one of the only devops/linux admin at a past job and was in one of the first rounds of layoffs. It's a bit dangerous to think that just because you're critical, you're immune - on paper, you're probably also one of the most costly.

> SysOps folks keep the site up, but once you start hitting >99% reliability, throwing more money at sysops isn't going to increase revenue.

Which is why SRE evolved from ad driven sites like google -- all uptime is worth money.

Yup, it's always a good idea to figure out a way to position yourself as the profit generator, instead of overhead whose costs needs to be minimized.

Funny thing, tech is pretty unique among industry sectors in that engineers are part of the profit centre. Just about everyone else treats engineers as operational costs.

And a Product Designer is $5k less still.

It's not a question of "valuing" employees, it's about what salary you need to offer to be able to hire and retain people at the required skill level. And good SREs are apparently willing to (or plentiful enough to need to) accept slightly less money than equally good developers.

That said, the phenomenon extends well beyond our industry. It's common in all industries that the people who design (as in architecture, not graphic design) and build the product make better salaries than the people who operate the infrastructure, however critically important their role is.

> This is especially vexing in light of the fact I've never met a highly qualified Ops person who wasn't also a skilled developer

Be careful to compare like for like. A highly qualified ops person compares to a highly qualified developer, not merely a "skilled" one. For all the difficulty of objectively measuring skill, I would assume that a highly qualified ops person would make substantially more than a merely skilled developer.

> Be careful to compare like for like. A highly qualified ops person compares to a highly qualified developer, not merely a "skilled" one. For all the difficulty of objectively measuring skill, I would assume that a highly qualified ops person would make substantially more than a merely skilled developer.

I think that I failed to communicate my meaning clearly in that statement, so I will attempt to clarify. What I mean is, that in my experience a highly qualified Ops person will /also/ be a skilled developer, but a highly qualified Developer will /not/ be a skilled Ops person. E.g. based on skill-set alone, highly qualified Ops people bring more skill to the table in union than a highly qualified Developer. Whether that's the skills you value or prioritize as an organization or not is another question entirely.

For the purposes of my salary comparison above I used the same years of experience and skill rating in the calculator for the Site Reliability Engineer and Developer roles. As you say, I would expect a highly qualified Ops person to make more than a "merely skilled" developer. I wasn't really trying to state anything opposite of this.

> that in my experience a highly qualified Ops person will /also/ be a skilled developer, but a highly qualified Developer will /not/ be a skilled Ops person.

I have ~15 years of ops experience, and I'm not a skilled developer. I bill ~$200/hr for infrastructure/AWS architecture/DevOps consulting (when not at devops day job), with limited need to be proficient at software development.

Curious where you're getting that hourly without also doing coding.

I'm a terrible programmer but I have 20 years of massive scale op experience.

> Curious where you're getting that hourly without also doing coding.

Both startups and enterprises.

> Henry Ford once balked at paying $10,000 to General Electric for work done troubleshooting a generator, and asked for an itemized bill. The engineer who performed the work, Charles Steinmetz, sent this: "Making chalk mark on generator, $1. Knowing where to make mark, $9,999." Ford paid the bill.

Yes, I think the subtle distinction between highly-qualified and skilled is what matters here.

$10k is the difference between several years of experience. To switch from SRE to developer, you would have to be willing to give up a few of those years.

I recently wrote a blog post coincidentally using Stack Overflow's own data to find correlations between the self-reported skill level of developers vs. other metrics such as salary: http://minimaxir.com/2016/07/stack-overflow/

The tl;dr is that yes, there is a positive correlation between developer skill and salary (http://minimaxir.com/img/stack-overflow/so-programming-6.png) for example, but it's hard to say which-causes-which.

Your graphs look a lot like the ones I made internally :) Try comparing the `agree_` and `important_` across age ranges / genders, I was able to pull some stats out of those but since the sample sizes are so small I didn't want to release it to the world.

Typical scenario over the last few years at different companies:

I'm ask to help find DevOps SREs.

Find people and interview them

Recommend the few good ones.

They turn the offers down because not enough money is offered.

I'm asked what could be done differently to hire good DevOps/SREs, because it's impossible to get anybody!

Later, once I've moved on, I found out that devs were offered and paid significantly more.

We use various resources (public data, information from VCs) to figure out the base market rate for each position. Even though the base salary is less because our information says the market rate is less, we don't think developers are more important than SREs. The same benefits, equipment, remote work, autonomy, opportunities to speak and contribute to open source, etc are the same.

> We use various resources (public data, information from VCs) to figure out the base market rate for each position.

So you're basically paying what everyone else is paying (on aggregate) rather than figuring out yourself if SREs and developers are worth to you as a company. This seems like a lazy-ish approach to me and negates most of the positivity that salary transparency has.

I realize though that you're never going to find an approach that satisfies everyone, so maybe this one works well for you (StackOverflow is an awesome site!).

> The same benefits, equipment, remote work, autonomy, opportunities to speak and contribute to open source, etc are the same.

That is tautological, "the same benefits are the same". $11k buys a lot of extra "benefits" in the world.

as a business that has to turn a profit, of course you will not pay significantly more than you have to. Base Salaries are demand and supply driven which also means that if i can pay more, i can also get more skilled/experienced people for that money.

Supply and demand?

Even if skilled SREs aren't as common as full-stack developers, it could be that so many more full-stack developers are required industry wide that companies have to compete harder on compensation for the best ones.

In my experience as a hiring manager, it was a lot easier to find qualified developers than it was SREs or similar. Plenty of people doing ops work, but not a lot of them at the senior level. And the senior ops people we find demanded, rightfully, much higher salaries than the developers.

I've found the opposite to be true. There's lots of mid-level developers around who are capable of handling most common stacks. Even mid-level Ops people are few and far between. Truly senior Ops people are a rarity and in very high demand, even if salaries aren't matched to that demand. The difference in responsibility in those roles that you are entrusting an employee with makes the comparison a bit difficult, but generally speaking far more is expected of Ops hires than Developer hires at most companies. When you hold the keys to the kingdom you have an extraordinary ability to cause damage to the company purely through innocent mistakes, much less malicious behavior.

I think what you're explaining is actually reality (not shown on the salary graph).

There's a large donut hole in ops. The truly high demand people make way more than the salary listed there. However, developers are within the normal range.

I.E. - Developers make more on average, top end Devops pays more.

A very good friend and senior, expert, skilled developer just transitioned into a DevOps position. I was very surprised. His motivation was to have a greater impact on the success of his org. Commendable.

I guess I did something similar when I served as a QA Manager, years ago. I loved the work and felt really good about the progress that team made.

Came here to say this. At minimum, an SRE is worth at least as much as a developer - and frankly, are probably worth more.

Could this be market supply/demand? That banks in NYC are paying more for developers and less for site-ops (similar reasons mentioned elsewhere on cost vs profit center) and to hire and retain these folks, they need to pay accordingly. In this sense, it's the career interest in Siteops that lowers their pay, despite the skills.

This happens in the real world a lot. Two people major in Mathematics. One has an additional education certification and goes to teach in a school. The other gets an IT job at the same school. The guy with the IT job gets paid more. (You could insert social work, music, or any of a dozen other underpaid fields)

I'm surprised none of the other respondents have caught this error in Stack Overflow's terminology.

Stack Overflow's salary calculator is lumping two very different roles together and is therefore not useful for true SRE pay calculation. I'll explain my reasoning.

Notice that there is no entry for "System Administrator" or "DevOps Specialist". Neither of these roles call for software engineering ability. Since Google coined the term Site Reliability Engineer and wrote the book on it [1], I'll use their definition as the canonical description: "SREs are engineers. [They] apply the principles of computer science and engineering to the design and development of computing systems: generally, large distributed ones." "Google caps operational work for SREs at 50% of their time. The remaining time should be spent using their coding skills on project work."

SREs use engineering practices to develop durable architecture, automate emergency response (since human response is too slow to preserve the error budget), develop monitoring, profile, and finally, pass operations control back to the developers when this is done.

Real SREs spend at least 50% of their time architecting and coding reliability solutions and automation for their infrastructure. They are software engineers in the most real sense. At Google and other companies that hire real SREs, they are paid more than typical software engineers at the same level of experience, etc. Most companies don't have SREs. SRE is a new occupation, but if you have to give it a Bureau of Labor title, "Software Developer, Systems" is the best fit.

Stack Overflow would have been better off not using the SRE title and sticking with "Sysadmin". Sysadmin is by far the most common title for ops workers. The SRE community is still too small to be a meaningful choice in this menu.

This would be the same as lumping in "HTML programmer" with "software developer". Two very different roles, and to consider them to be the same is serious title inflation.

To approximate SRE salaries, I would choose "Software Engineer" and bump the skill level up by one. This is my experience and matches other SREs I know.

[1] Site Reliability Engineering: How Google Runs Production Systems. J. Petoff, N. Murphy. O'Reilly. 2016.

I misread TFA and thought the calculator was based off of jobs posted in the site, not their own salary calculation. Judging by their comprehensive operations articles (DC migration, etc), it appears their ops staff are also software engineers. In that case, it appears they've misinterpreted the market. At least SREs know they should apply to Stack Overflow only for software engineering roles in order to avoid a $10k penalty for extra skills.

That was pretty much my point. An SRE or DevOps person who's any good is basically a Software Engineer + SysAdmin in one person, vs two people, so paying them $11k/yr less is kind of ridiculous.

StackOverflow is a Windows shop.

That almost entirely explains the pay gap.

and that's one problem with this "salary transparency" stuff, "my work is just as important or even more important than his, why I'm getting paid less?"

I could argue against you too, for example, you say "competent developer in numerous languages and environments", that's really just a small fraction of what computer science is really about.

I don't think it's a problem. I think it's a valuable and fair discussion that is in some way the entire point of salary transparency. Those who bring the most value to the company in theory should be the ones who are compensated the highest. It's always a difficult question on how you determine who brings what value though, which is why compensation is such a contentious subject to discuss.

I should clarify I'm not claiming that an SRE is more valuable than a Developer outright. I'm saying that most highly qualified SREs are also Developers, while highly qualified Developers are not also SREs, so the amount of skill/knowledge required to attain the salary is more burdensome for an SRE vs a Developer. If anything, my statements should be construed as a realization that its a better career progression (at least when focusing on salary) to become a Developer than it is to go into Ops because for whatever reasons society values Developers more than Ops.

Also, separately, you need to consider BATNA [1]

If 10 guys all showed up on your door with a magic box that would give your business $100k in added shareholder value, you'd be rational to pay somewhere between 0 and $100k for it. The top end is determined by how much new value is added, but what you end up paying would have to do with how each person priced it -- if there were more of them, the thing would probably cost less unless the providers of it colluded.

Just keep in mind there's a push and a pull and, even if a factor adds a lot of value, it can be cheaper than a less-value-added one, because factor prices are set by supply and demand, not their relative value-added output (which can also depend on how well they're used, which is why some companies can pay so much better for exactly the same work as others)

(Which is really a broader recipe for success in business generally: figure out how to employ factors of production -- people, capital, etc -- more profitably than they cost to purchase in the relevant markets, and BAM, you've got a profitable company)

[1] https://en.wikipedia.org/wiki/Best_alternative_to_a_negotiat...

"Who brings the most value to the company" is the most utterly loaded question ever.

Companies are complex systems with many moving parts and team dynamics. If a new feature gets released, do you attribute incremental revenue to the dev who coded it, the product manager who specified it, the sales team who sold it, the marketers who described it, the executives who hired the people who did any of the things above, etc?

Bottom line: "value creation" is a totally sociological construct having more to do with the organization's culture (who's most highly valued) than any objective metric of "value add". You need to be in a place where the people empowered to make decisions about pay perceive that the work you're doing is valuable. No more, no less.

If you asked each of those constituents how much value they/their group added and summed those values, I suspect you'd end up with a significant multiple of the total value actually created.

I've done a significant amount of hiring in my career and I've had a more than a few conversations about the "importance of my work" vs "level of my compensation". In the nicest possible way I've had to explain that 'importance' has nothing to do with it.

Hiring is supply and demand. There are more competent accountants vying for the same jobs than there are developers. More developers than there are SRE's.

The rarer the skill set, the more expensive it is to hire.