Is top talent in software overrated?

The problem is far more nuanced.

The top individuals can be extremely productive and worth every penny, but most companies don't need top talent (they write mundane software that requires plugging some custom code to a framework) and can't even identify top talent, because it takes top-talent to spot top-talent. The best they can hope is just avoiding hiring the worst developers.

> But the top 20% of engineers is not that much better than the middle 60%.

Wrong. But since companies wouldn't be able to tell a difference, you might as well think it's true and be better off.

> An average engineer can do almost anything a top engineer can, but for a much lower price.

I've seen things that the company struggles for years being solved in a couple of weeks by a dedicated top-talent individual. But they were hard things - involving distributed systems, hard performance and data requirements, complex architectures etc. Things that "medium 60%" simply can't solve. But 90% of work in a typical company is actually nothing like that. And if you give top-talent a task to write some mundane business logic they might produce somewhat better code, somewhat faster, but it will be 2x tops, and possibly even less than 1x, because they will just not be motivated.

So you have to know what kind of engineering your company is doing and hire matching developers.

One of the things I like about getting older is that you're old enough to see industries forget lessons they've already learned. A few years back I invested some time reading the entirety of Joel Spolsky's blog, and because he writes so well, a lot of it stuck to memory.

Back in 2006 he wrote "Finding Great Developers", and in it he says: "The great software developers, indeed, the best people in every field, are quite simply never on the market."

This is as true as it gets. So, for most companies, they're not getting top talent. The best next thing is to get VERY, VERY CLEAR on the hiring end on what the job actually demands - and hire for that.

Separately, and you can see it a lot on HN, there's a competitive element around being "talented" or "good". As in, "Oh, she's super talented", or "I'd like to be as talented as they are". To be fair, it's not unique to those who code, you can find it in every STEM department in every college, but it's nonsense all the same.

Companies prize people who can get things done. And that ability is multi-faceted: it's not just about how good a developer someone is. How much they understand the domain, how much relevant experience they have, how good they are at mobilizing other people to work together on a solution, how capable they are at getting others around them unstuck, how well they can explain their ideas and discuss others' in a productive way...So much goes into this bucket of "getting things done".

In my experience, top talent is most valuable in the early stages of a company. That is when it can have the greatest impact. Communication overhead is low, smart people can make fast decisions, and just get things done quickly.

By the time you have 100's of engineers, everything will be quite watered down. Anyone exceptional will get dragged down in bureaucracy, red tape, processes, procedures, enormous amount of useless meetings, etc. Maybe I've just been in the wrong companies. Having top talent is always better than not having it, but often it is wasted. You get promised the world and wind up spending half your days in Zoom meetings and waiting for PRs to be approved.

No it’s severely underrated. 10x is the normal range of eng talent. Truly elite talent can deliver 10000x. Companies never pay these guys enough so they just leave and make their own companies.

In my experience, the ability of managers to recognize top talent when they have it is faulty and takes a long time. They tend to get false positives as they incorrectly value delivery time over things they can't measure, like quality. I can't count the number of times I've seen someone be praised in a group email or meeting for claiming to have fixed some urgent issue (with duct tape and bubblegum), and every single time without fail it was not worth the price and I at least cursed that person after in private. "Top talent" is also generally kind of a bin for a wide spectrum of people from ridiculous coding savants beyond compare to people who just think deeply and have been around a bit.

Also in my experience, what seems to matter a lot is experience. People who have failed and learned and who know good practices when they see them will just make more good decisions which won't come back to bite later.

I've found having some people with good experience and communication skills around can be great if they can be positioned to support less talented or experienced people. Finding a good balance and people with reasonable dispositions is important.

The most relevant thing I can see is that managers can't recognize top talent, especially at larger companies. At a small startup, everyone has a sense of who's involved with what in what capacity so top talent can be recognized sooner.

Small companies fear losing their top talent. Large companies prefer to have more easily interchangeable/replaced engineers for ease of project planning/staffing and often don't distinguish top talent from a typical senior level. Titles don't correlate well with top talent, rather tends to be a level of seniority 'how long have they been here and competent' label.

The biggest of companies like to hire top talent but don't make good use of them, often seeming like a game of 'keep away' so that they don't work at a competitor or start their own competing company.

My own experience is that there are many different forms of top talent and the only way to get efficient utilization is for the employee to themself actively put them on the projects or roles that best suit their skills, whether the manager or company recognizes this or not.

Thats bonkers. Top talent is capable of creating 100M companies. It is easier to keep these people employed for ridiculously low salaries compared to what they are really worth on the market.

>An average engineer can do almost anything a top engineer can, but for a much lower price. They cost-to-value ratio favors average engineers.

I think there's a fallacy here. An 'average engineer' hasn't got a much lower salary than a 'top engineer'.

My take is that it's worth paying above average in order to hire and retain above average engineers and above average technical managers.

Benefits can include smaller teams and better quality code, which leads to cost savings that compound over time.

A great developer can make the difference between your company getting that $ million contract or not. So the value of a great developer can be extremely high compared with the average developer. Also, great developers will often inspire the rest to do better, and remove any “we can’t do that” thinking from the organisation. Managers who doesn’t understand this simple fact should be fired for incompetency.

Probably made up statistics, but 80% of drivers think they're better than average.

Question is, how do you know? It's super subjective. A lot of "top-talent" probably thinks that they're better than average.

I would imagine the actual top-talent doesn't apply for jobs, I would kmagine that they are poached.

99% of pertinent programming skills are high-school grade.

I was programming real-life assembly, I was in my early teen, and many other teens were doing the same.

But the second you have to do accurate and serious floating point computations, you need at least a college/uni master degree in applied math.

In fairness: 1) there are probably a lot more people who think they are "top talent" than there actually is top talent. And the "think they are" group is probably louder about being top talent, and louder about insisting on being treated like top talent. And so 2) this manager has probably been exposed to more think-they-are than actual top talent. So deciding that they can do just as well without the think-they-are group (and their headaches) is defensible.

What's more, not everyone actually needs top talent - any top talent. There's a place for shops that just employ average engineers, because they don't really need anything more.

All that said... this manager sounds clueless as to what actual top talent can be and do. They also sound like someone who is going to blindly steer their organization into corporate collapse, caused by lack of actual engineering vision.

I think this really comes down to what your company needs to do.

If you are doing things that the middle 60% are capable of, top talent might not be worth it to you.

If you are doing things that only top talent is capable of, then top talent is worth any price you can bear.

> But the top 20% of engineers is not that much better than the middle 60%.

Have them watch some of for example Jonathan Blow's talks or coding streams and then ask if they still believe that.