The myth of the 10x developer

"10x" developers don't exist. Or at least... not in the way that people think (or hope) that they do. 

In theory a 10x developer is a programmer who is considered to be 10 times better than their peers.

The whole "10 times better" part cracks me up - I'm not really sure how you quantify "10x better". Does that mean the developer completes 10 times more unit tests than their peers? 10 times more comments? 10 times more code reviews? Drinking 10 times more coffee than every other game developer working triple overtime on a rushed AAA video-game? And why even stop at "10x"? Why not skip the 10x engineers and go straight for the 100x ones? Quantifying this thing gets absurd quickly.

I'm unsure of the origin of this myth, and I'm not keen on looking into it much anyway - I feel like during my pursuit of tracing this thing back to its roots I would just unearth even more layers of cringe. Unfortunately, people keep pushing the narrative further despite its outlandish nature for a simple reason: it's appealing. What programmer out there wouldn't want to be part of an exclusive "10x club"? Content creators capitalize on this and promise that watching their videos will answer your burning question: "how do I become a 10x developer?". Which is a bit of a shame, because many of these content creators were, or still are, developers themselves, and have to realize the bogus nature of this. 

Youtubers push the narrative by telling stories of devs who've accomplished seemingly superhuman engineering feats and talk as if the developers behind them are imbued with some mystical hard to attain skillset. These stories usually leave out the parts that talk about the countless hours of coding, trial and error, and learning the developers invested into perfecting their craft before they were able to pull off large-scale projects like the ones talked about in the videos.

Years ago, I built a digital signage application for a client that was synchronized across several of their office locations around New England. They wanted something that could be maintained by Java developers. Up until that point, I hadn't written a single line of code in Java. I familiarized myself with Java and had a prototype done in a little less than 2 months. It pulled data from a SQL server, featured images, played videos (utilizing VLC via JNI), and had a JSP-based control panel HR and Marketing leaders used to manage and refresh content. I recall that it was between 20,000 and 30,000 lines of code - not including the code in 3rd party libraries I used like JFreeChart and FTPClient in Apache Commons.

If my story was dramatized into a 10x dev motivational video, the title would be: "Programmer who never coded before learns Java and builds digital signange app from scratch in 8 weeks". The video would embellish and exaggerate every detail about the development process - all the while ignoring important details about my programming journey: like the more than 2 decades of experience I had coding and all the knowledge I was able to recycle and carry over into the digital signage app project. Not only did I have some intermediate experience with C++ experience (so Java syntax was easier to pick up due to similarities between the two languages), but also the development methodologies, tools, and time management habits I had adopted through the years were extremely helpful in forming a strong foundation for me to build off of.

The biggest problem with this 10x mentality? It can lead to a polarizing mindset.

Devs can be led to believe that they either "have it" or they "don't". They may think exceptional programmers have some "X factor" that enables them to pull off the development feats they do. If they think they "don't have it" they might come to the conclusion that they should just give up before they even try. It's beneficial and can even be a good motivating factor to aspire to being an amazing programmer, but you're going to get exhausted if you try to meet that expectation too quickly (especially for devs starting out in programming). 

On the other extreme side of the spectrum, someone might recognize in themselves a natural affinity to programming and assume they possess the magic "10x" trait - and develop an attitude of laziness and complacency: in their mind - there's nothing more they need to work at or improve to get better at. Not only this - but when people make this their entire identify they get frustrated when their "elite status" gets challenged.

I remember growing up there was a kid in my hometown that by all accounts certainly would have been what most people today would consider 10x: even as early as elementary school he was pretty good at C and Python. Once high school rolled around, some other programmers started to surpass him. This frustrated him, and he decided the best course of action was to move to another continent. Not another town or another state, but another continent. He believed (for some reason) that he wouldn't get challenged as much where he was going.

According to people from his inner circle, the relocation apparently didn't mitigate the issue for him. I guess he found himself going up against even greater competition at his new home, and went back to school with the new goal of being a "literary scholar" - a new field he was sure that, this time - he would definitely be better than the majority of his peers at. This might have been a rare coping mechanism... and in a way, sadly also a little comical, but its basically a disaster story about the "10x" mentality running amuck before the 10x term was even a thing.

It's impossible to find... or be, a "10x dev" because it's impossible to exactly measure programming skills in some sort of numerical fashion. There are, however, exceptional developers - devs who are extremely productive and know how to get stuff done.

Exceptional developers are creative and resourceful. They don't let a lack of resources or manpower get in the way: when they don't have what they need - they build what they need. If they're hindered by the tools or processes that were implemented before them, they'll code new scripts and tools themselves to speed those processes up.

It's not a dev who has all the answers - it's someone who knows how to find all the answers. Whenever I've helped to screen developers in hiring processes, I personally never cared about candidates that stumbled a bit remembering the proper syntax during a coding assessment - we all have the occasional mental block. What I did care about was their ability to ask good questions and synthesize creative solutions, even if it resulted in pseudocode that didn't compile.

Extremely productive coders are those who think before diving into coding - they've adopted a "work smarter, not harder" approach to problem solving. They've completed enough projects to know that its better to spend a little more time upfront mapping things out, white-boarding components, and doing their best to foresee and tackle edge cases before they're even encountered in the testing phases. 

Highly productive developers have also perfected a balance between impatience and patience. They'll persevere longer than most to solve a problem - but also know when its time to pivot to a new strategy when one particular path to solving a problem isn't yielding fruitful results.

Stop asking yourself, "How do I become a 10x developer?". Focus on making valuable contributions and get better and better at shipping them in an efficient manner. Learn and diversify your skill set and find new ways to solve problems. Tune out the noise and stop concerning yourself with trying to live up to fairy-tail expectations. Be competitive, but compete against yourself. Set the bar higher for yourself with each project. 

If you're waiting to hire a "10x" developer - you should forget about it because they simply don't exist. Every so often an organization will get lucky and stumble upon someone they deem worthy of the "10x" title and craft an interview template to detect more purple squirrels. Ironically, instead of finding more gifted developers, they'll just find more people who are good at gaming and passing those specific tests.

Hard-working, talented, productive, and resourceful developers, do exist, and they're everywhere - you just have to know the right ways to find them. The majority of modern day interview processes aren't conducted in a way conducive to finding such talented developers. But, well.... that's a topic that deserves a separate post on its own.

Make every project yield value

You can't guarantee that every product will be a major success - but you can guarantee that the development process of each one creates value for your organization every time. 

I've been consulting a lot over the past couple of years. One of the things I do is help leaders within smaller organizations make informed decisions regarding their tech stacks and the necessary tooling that should be in place to ensure their development processes and workflows are in line with their business needs and requirements. 

One trend that I've noticed an uptick in is with companies shifting their tech stack from Java and migrating over to Golang (often times I think this is more in response to evolving trends than actual necessity... but I digress...). Often times, these companies have a lot of legacy code written in Java that needs to be managed by Java devs. 

One of the groups I'm working with recently expressed interest in a new software project. Going forward, they want future features and microservices to be built in Go. There are 7 Java devs, and 2 dedicated Go devs currently on this group.

The initial group consensus was that both of the dedicated Go devs should handle the experimental project while the Java devs would cover the maintenance side of things on the Java side of their code base.

After analyzing their dev allocation, it was clear to me that there was ample Java resources allocated to the maintenance side of things - an area area where, frankly - the day-to-day work centered around bug fixes and occasional "change the color of some label or button" type 2-point sprint work. Feature requests had pretty much stalled, and there really weren't any signs pointing to a reversal in that trend.

I proposed an idea: pull a couple of the Java devs off the maintenance work and re-assign them over to the Go side of things. Now the Java devs will gain exposure to the Go tech stack and who knows - might even end up liking and excelling in that area more and find a fit on that team.

Going even further: The 2 Java devs that are being mentored by the Go programmers are senior developers. The Go developers, on the other hand, are entry to junior level. This dynamic might not seem intuitive at first if you're thinking about things from a hierarchical perspective - but I wasn't thinking strictly in terms of a hierarchical perspective - I was thinking in terms of what each side had to offer the other. The senior and lead developers are senior for a reason: they already have significant experience leading other engineers and working closely with stakeholders to ensure proper alignment of design decisions with business needs - why not offer the chance for the junior Go devs to get some experience in this area as well? 

The process I've outlined here isn't exactly groundbreaking - it's been thought of before: I've seen it proposed many times in the past by not only me, but other devs I've worked with. It is unique though in the sense that in reality, it's rarely implemented. When there's a new project, leaders will often hyper-focus on which developers can get certain parts of the project done the fastest. There should be an emphasis on structuring the project around fostering career development and a climate of talent diversification - building a more flexible team that can support multiple sides of an organization if need be. From my perspective, there's no better time to let other's break into different sides of the product than during a new speculative endeavor.

The project is still in the earliest phases of production and only time will tell how much money it's going to bring in. In a lot of ways though, it's already a success: the organization now has Go devs with heightened leadership abilities, and an added layer of insurance with Java devs that now have enough insight into the Go codebase that they can assist with a hotfix or other high-priority fixes as they arise.

Some products skyrocket, others barely break even, and inevitably, some will even fail without ever yielding a dime; then there are others that provide just enough cash flow to cover opportunity costs of the resources poured into those projects. Regardless - it's entirely possible to derive value from the project regardless of the monetary value it yields. 

This isn't to say, of course, that a company should just leap into a new product idea even if they think it won't make them money. However, once the green light has been given to proceed with a new idea that has a high chance of turning profit and things have progressed to the implementation and design phase, there should be other decisions besides simply determining which engineers can complete which tasks the fastest. Engineering talent and resources should be arranged in a way that not only maximizes development efficiency, but also learning and talent dispersion. This will lead to a more robust, flexible, and agile engineering environment equipped to respond more swiftly to evolving expectations. 

Finding the square root of a number using binary search

The other day I was going through a hard drive full of homework assignments I did years ago in college.

I found one of the assignments interesting enough that I thought I would include the solution on here. The goal of the assignment was to write a function that didn't use any built in math library calls to calculate the square root of a number. 

The instructions stated that 1000 test cases would be performed, and they ALL had to pass in less than a second. In other words: the sum of all the runtimes of each of the test cases had to total less than a second -- which is pretty much to force the coder to get creative and not write a brute force solution (a bruteforce solution.. while easy to code, would perform horrendously). 

For reference, a brute force solution would look something like this:

public static double bruteForceSQRT(double num) {
    final double precision = 0.00000001D;
    double result = num;
    if (num < 1.0D) {
        if (num == 0) {
            return 0;
        } else if (num < 0) {
            // Handle these cases Like Math.sqrt() does it
            return Double.NaN;
        }
        result *= 2;
    } else {
        result /= 2;
    }
    while(result*result > num) {
        result -= precision;
    }
    return result;
}

In the brute force approach we halve the number we are given and then decrement that by a certain precision decimal value until we reach an approximated value that is extremely close to the square root of that number. Unfortunately, the greater the accuracy that you desire, the drastically slower the runtime is going to be. It took more than 2 seconds! for the bruteforce code to calculate the square root of 25 - which of course is ridiculous. 

The approach I went with for the homework assignment utilized binary search instead. Binary search provides superior performance and accuracy, and looks something like this:

public static double bin_sqrt(double num) {

double left = 0.0D, right = num;

if (num < 1.0D) {

if(num == 0) {

return 0;

}

else if (num < 0) {

// Handle these cases Like Math.sqrt() does it

return Double.NaN;

}

// If we got to here then 0 < num < 1

// so we want to actually set LEFT to num, and RIGHT to 1

// The resulting square root value will be GREATER than num

// (i.e: the square root of 0.25 is 0.50)

left = num;

right = 1.0D;

}

double middleNumber = (right + left) / 2;

double lastMiddleNumber = num;

double mSquared = middleNumber * middleNumber;

while (lastMiddleNumber != middleNumber) {

if (mSquared > num) {

right = middleNumber;

} else if (mSquared < num) {

left = middleNumber;

}

lastMiddleNumber = middleNumber;

middleNumber = (right + left) / 2;

mSquared = middleNumber * middleNumber;

}

return middleNumber;

}

The binary search method is clearly faster than the bruteforce approach - but you might be wondering -  how does it stack up against the built in Math.sqrt() function? 

Below is a table I put together which shows the results of runtimes for the binary search version (bin_srt) vs. the standard Math.sqrt();

As you might expect, Math.sqrt() in almost all cases is still significantly faster than the binary search method - sometimes almost 5 times as fast (on modern day systems, Math.sqrt() is likely to execute the square root hardware instructions directly - which is likely going to beat any homegrown software solution any day). 

Nevertheless - I could see the binary search method being useful in some niche cases: like if you can't or don't want to include a standard math library due to size constraints for a game jam, or solving an algorithm question, or, as in my original case - solving a homework problem. In those cases this is a perfectly fine solution that offers a good balance between speed, efficiency, and accuracy. 

Lastly (and arguably most importantly), this example stands as another testament to the power and efficiency of binary search and the many use cases it can be applied to.

A Java version of the binary search square root implementation can also be found on my github here: https://gist.github.com/edev90/bda51627b722a0d825453b3d797b7283