There is no single right answer to any non-trivial problem in software engineering. So, if multiple correct solutions exist, how do you decide which is best? It’s difficult to determine which is best because “best” is highly subjective and deeply personal — your opinion is formed from your individual collection of experiences, strengths & weaknesses, and values on related aspects like simplicity, maintainability, & scalability.
It’s these internal values that make all of this so tricky. Imagine a spectrum with engineering excellence at one end and business needs at the other. Both elements are required for a project to be successful, and operating at either extreme can be detrimental to the other. As an example, making too many quick-twitch fixes to address urgent business needs can have significant long-term impact on the quality of the code base or system maintainability; conversely, focusing too deeply on engineering excellence can lead to over-investment in areas or competitive disadvantages from being slow to market.
Understanding this spectrum — and having awareness of where you and your colleagues lie on it — can help your team to be more pragmatic.
Awareness of this spectrum alone isn’t going to do you any favors in resolving conflict from perceived disconnects between you and co-workers, though. I’ve found myself in design/requirements stalemates many times, and I’ve used the spectrum as a way to visualize my frustration.
“You see, I live over here on one end of the spectrum,” I’d say, “and my colleague operates here, at the other end. We can’t agree on scope, and we aren’t getting started or making any progress as a result.”
The problem with the visualization as a tool for conflict resolution is those pesky personal values. Neither of us thinks we’re advocating for a solution that would be in the unhealthy extremes of the spectrum. The person in the engineering excellence camp just believes that business value is generated by following all the best engineering principles and creating scalable, high-performing, resilient applications whereas business needs nation wants quick delivery and maximum responsiveness to meet the ever-changing needs of its customers.
So, how do you find compromise when the source of conflict is so visceral?
Let’s see if we can steal a page from the Goldilocks playbook. She’s got a knack for identifying the undesirable ends of a spectrum before settling into a satisfying sweet spot. If you and your team or colleague(s) can’t agree on the scope of a solution, can you agree on what it shouldn’t be?
What’s a reasonable solution that everybody agrees is over-engineered, and what’s the fastest, but perhaps short-sighted, thing you could do? What’s the effort required for each approach, and what are the risks or consequences?
Just to be clear, I’m not suggesting to simply compare different proposals by plotting them on the spectrum— that probably won’t get you anywhere. Instead, work collaboratively to come up with bad solutions that lean too far in both directions. Find agreement by identifying undesirable characteristics of these options in the unhealthy parts of the spectrum.
Still not able to find compromise? It’s probably time to bring in a 3rd party, preferably a stakeholder. Show them your spectrum and explain the tradeoffs that exist at the opposite ends, then present the “real” options that are on the table and allow the stakeholder to decide.
The whole activity is an exercise in pragmatism. How can two parties with equal but conflicting opinions find common ground? The key is to calibrate and remove as much subjectivity as you can. By acknowledging the necessity of both aspects — engineering excellence and needs of the business — and agreeing on the “bounds” of the spectrum, you create a framework for identifying the region for compromise. That’s your sweet spot. That’s likely where your “best” solution should be.
Originally published at The Startup on September 21, 2020.
I grew up as a computer kid in the 80s and 90s and, consequently, spent a lot of time in video games. Now I manage two software engineering teams, and it’s time to prove to Mom that all those hours spent in Final Fantasy were actually valuable career development.
I’m primarily thinking about two video game genres: party-based RPGs and MMORPGs. The formula for these is pretty simple. You have a cast of characters with various capabilities, and they work together to accomplish amazing things. In order to succeed, you must be aware of your characters’ strengths & weaknesses, understand what skills are required to complete a challenge, and combine characters in a way that allows them to achieve the goal.
Well, that doesn’t sound so different from a software development team, does it? You’ve got a group of people with different abilities and aptitudes; you have a backlog of stories to complete; and the team must collaborate to achieve goals and accomplish amazing things. It’s, like, the same thing!
Given these undeniable parallels, what lessons from RPGs can be applied to software development teams?
It’s Dangerous To Go Alone
When you adventure alone in a video game, bad things don’t happen most of the time — but there’s risk. The same is true in software development, especially if individuals on your team are all working on separate things. The primary risks of soloing a software development project are consistency, quality, and knowledge-sharing.
Working alone on projects offers short-term risk due to the fact that quality and completion time are largely dependent on who does the work. Someone who’s very experienced and understands the subject area well will probably get by fine, but a developer with less familiarity will take longer and be more likely to make mistakes that require re-work. Peer code reviews are slower and less effective because reviewers need to context-switch and focus deeply to understand decision-making and what’s been done. Individual stories are completed less efficiently because everyone has their own assignments and agenda, and other people’s work becomes a secondary objective.
Knowledge-sharing and information silos are the long-term threat. It’s easy for people to become specialized as certain types of work gravitate toward them. If an individual is the only one that worked on a specific project, guess who gets tapped when it needs attention later? Future work can also be bottlenecked when the person with all the knowledge isn’t available, and the situation gets worse when that same person is the bottleneck for multiple workstreams or when personnel changes occur.
In the video game world, the risks and consequences of soloing are typically limited to just you. People choose to solo because it’s convenient — they don’t want to wait or look for other players — or because they enjoy the challenge. Sometimes it feels easier to go it alone in software, too. However, software development is a team game, and you’re not typically looking for extra challenge just for the fun of it.
Ensuring you have multiple people working on a project mitigates the risk. One person doesn’t go down a bad path by themselves, and at least two people should know how things work and why decisions were made. Pair or group programming is inherently review-as-you-go which leads to better initial code quality, and that in-turn helps with completing stories efficiently due to less feedback cycles and re-work.
You don’t want everybody acting alone, but it’s equally important to make sure you don’t have too many resources focusing on a task. In other words…
Quests in RPGs require a certain set of skills to complete. Easy quests are achievable by smaller, less experienced groups, but hard ones require more people, specific skills, or other special assistance. In games, it may seem more efficient to clear easier, low-level content with powerful, advanced characters, but if the strongest characters are focused on easy things, it means they aren’t working on more challenging, higher-reward objectives.
Software development is an exercise in efficiency. You have a backlog filled with user stories. Some are easy and some are difficult. You want to complete them all, though, and the faster you can do it, the better. The trick is for the team to determine the optimal way to complete as much as it can as possible as quickly as possible.
How would you do that in a game? First, you’d decide which quests are most important and what skills they require. Next, you’d look at which characters are available and what skills they possess. Then you can optimize who can do what. Perhaps one small group could tackle three easy quests while another group works on a single complicated one.
This same strategy can be applied to sprint planning. You’ve got stories in a prioritized backlog and a team of people to complete them. Which people have specific skills or knowledge required to complete the highest-value stories? Assign those folks first. Who’s left, and what’s the best way to utilize them? Make sure everybody is assigned, and balance the groups.
You may find that you don’t have enough people with the right skills to succeed with all the most important things. Luckily, RPGs give us a solution for that, too!
Characters in RPGs progress by completing tasks that reward experience. Once you’ve accumulated enough experience, you level-up and become stronger or gain powerful new abilities. In order to have a well-balanced team, you must use all your characters so they all gain experience. If you have a weaker, low-level character, you can grow them most quickly by leaning on them as heavily as possible in content that’s within their reach. Sometimes that can be painful, if you need to return to a low-level area and perform low-value activities, but the investment pays off with time. Once those characters “catch up” they provide valuable versatility to the team.
On a software team, this means ensuring that everybody’s getting reps with the most important skills. People won’t suddenly gain project management skills by not managing projects. Instead, acknowledge it: “I want you to manage this project so you can develop these skills.” You want someone to be a better decision-maker? Ask them to make decisions. Similarly, just because someone can do a thing in 15 minutes doesn’t mean they should do it simply because it would take someone else 2 hours. Instead, invest that 2 hours. The experience is as valuable as the time saved, and it contributes to both the growth of the individual and the strength of the team.
To be a successful team, there are multiple roles that need to be filled. As employees gain experience, they typically focus on a single role. For example, “I’m a developer, and this is what I do as a developer.” Often times, that’s enough to do a good job, but a great team member understands all the roles within the team and has the ability to recognize and step in when a role isn’t being fulfilled.
How do you develop that level of situational awareness?
Often times, games have encounters with mechanics that need to executed in order to succeed. You can win by knowing enough to go through the motions, but the best players don’t just know what to do — the know why they’re doing it. That also means understanding the consequence for not doing a particular mechanic and being able to adjust on the fly when things start to go sideways.
Software teams have a process designed to help the team succeed or — more pessimistically — prevent it from failing. Individual contributors can be successful by adhering to the rules and following the process. Great teammates will understand the underlying reasons for the process, though, and be able to make decisions around when it’s time to deviate.
Much like in games, a great way to become more familiar with the intricacies of your process is to learn it one role at a time from people who are already proficient. How does your team gather requirements and translate them into actionable work? What’s the best way to do a peer code review? What tests need to be run to verify there are no regressions? With better understanding, you can help improve these processes. Don’t just learn how to do things; learn why to do them.
These lessons from RPGs aren’t necessarily special or specific to software development; they’re just general best practices for teamwork and growth.
Recognize the strengths and weaknesses of individuals on the team, and utilize them in a way that makes sense — just like you wouldn’t have your squishy wizards standing in front of heavily-armored knights in a game.
Use the right number of people based on the task. There are both short and long term risks associated with using too few people but possibility for reduced efficiency & throughput with too many. It’s also important that the group have the right set of skills to accomplish the goal. Don’t poke the dragon alone, don’t bring the entire village to feed the horses, and don’t send a small group of adventurers into the dark cave without a torch.
Build your team by ensuring that people have the opportunity to grow. Acknowledge that having somebody less skilled perform a task pays dividends as they gain competency. If you don’t give people the chance to improve, they won’t. Having more people at level cap lets you tackle a wider array of challenges or accomplish more things at once.
Invest in helping team members understand the big pictures. What’s the greater purpose behind your team? Why do your processes exist? Encourage people to learn new roles and step outside the bounds of their specific job. The increased awareness and versatility differentiates good teammates from great ones. In an epic RPG boss fight, you can still win if somebody knows how to kite when the tank dies.
Know the team, grow the team. Use the right skills to get the kills, and collect that sweet, sweet loot.
adamprescott.net 