Telling Time as an Engineer

Time is the most precious resource. It’s in limited supply, once spent we can’t get it back, and you can’t trade it directly. This might sound a little radical but most global, national, business, and personal problems, seem to me, to boil down to problem of time and who’s time is more important than yours.

Before we can decide how we should spend the time given us, we have to put some thought into the process of analysis of tasks which take time. In software development a considerable amount of thinking has been applied to just this analysis. We usually call it “process management” and “time management”. Many methodologies have been created to solve the problem of time and yet when the rubber hits the road the management of time, which includes task prioritization and effort estimation, is full of errors and random results.

A great example is the Agile Development Processes, which has become the standard as well as declared dead by many of its original creators. Why is this?

Here is a simple example…

A high priority story is pulled from a general backlog and estimated, along with other stories, by an experienced engineering team. A product owner then weighs the cost of the stories based on the effort estimations and value to the business and feed them into a sprint backlog. The engineering team then works on each story, in order of priority, and completes the required stories by the end of the sprint. The work completed is demoed to the stakeholders and everyone is happy as everyone’s time has been well spent.

Well, except this happy plan almost never happens.

Something like 80% of all feature and products are delivered late or not at all. And often when a feature or product is delivered its buggy enough that we regret delivering it on time if at all. I’m sure Samsung engineers are less concerned about deadlines these days and more concerned about taking the time to do their tasks with more quality. Blizzard has made a billion dollar business of never giving dates for games and missing them when they do. Facebook and Spotify just spring new feature on their users without any warning and kill bad ones before they spread beyond a small segment.

It my opinion successful tech companies don’t bother with time management and leave schedules and task estimates to unsuccessful tech companies. I’m not saying successful tech companies don’t do agile or create project plans. I’m saying these are more like historical accounts and data gathered for analysis than pseudo-predictive planning.

Why is task estimation so non-predictive?

The problem is that it’s impossible to know how long a task will take unless you have done exactly that task before. When I worked at Apple Computer (before it was just Apple) we said that in order to to understand how long a project would take you had to build it and then write the schedule.

This is why experienced engineering team is so important in effort estimation. If you get a group of engineers who are a bit long in tooth they can work together to pool estimates on work they have performed previously.

But much of the work of an experience engineering team is work they have never done before. Experience engineers tend to see everything though the lens of previous experience. The result is that effort estimates are inaccurate as they have mistaken a novel task for an nostalgic task. I can’t count the number times I have said, “I thought the problem was X and it would take Y story points, but the problem is really Z and I’m still doing the research so your guess is as good as mine.”

The fact that for novel work your guess is as good is mine is why startups of inexperience engineers succeed with problems that mature companies fail at. The boss says “This problem will take too long to get to market. Let’s just not do it. It’s a waste of time.” The boss also says, “Hey brilliant engineers, you didn’t deliver this product on time! You suck! You’re fired.” Both judgements are typical of mature companies where value has to be delivered every quarter and experimental failure damage previous reputations.

In a typical tech startup, or any kind of new business, if you did the estimates you would have never started down the path. But startups don’t care! They are labors of vision and love usually staffed by people without experience who don’t know better. A good startup certainly doesn’t worry about effort estimates or punish engineers for not being able to tell time.

My advice to any engineering team that needs to worry about time is as follows:

One

You need a mix of experienced and inexperienced engineers on the team. This doesn’t mean old and young as much as people who have done it before and people who have not. Mix your insiders with your outsiders. For example if you’re building a web app bring in a a few mobile devs to the sprint planning. And some interns. And listen to the outsiders. Engage in a real discussion.

Two

If someone in charge wants to know how long a novel task will take from just the title of the task, without any true discussion, walk away. You’re going to give them a wrong answer. By the way good estimates are rarely rewarded–they are expected! But bad estimates are almost always punished. An honest “I don’t know” is always better than “2-3 weeks” or “2-3 story points”.

Three

Remember there is no value in hitting the deadline without quality, performance, or value to the user. In fact I’m always a little suspicious of teams that never miss a deadline. Apps that crash or cutting scope to the point of no visible progress is a hallmark of teams that hit their deadlines. I’m not saying don’t work hard or try to hit your deadline just be tough about the result at the end of the schedule: Give it more time if needed!

Four

The problem with my advice is that everyone wants to know the schedule. So many other functions depend on the schedule. Releasing a product on time is critical to the business. So my final piece of advice, and you’re not going to like it, is let the business set the deadline. Instead of wasting everyone’s time upfront with a broken planning process to arrive at a deadline, get the deadline first and work backward with effort estimates. While time is limited the amount of time we spend on on a task is flexible. We work differently if we have 3 months, 6 months, or 12 months to accomplish a task. Ask any college kid how much time they put into their studies at the end beginning of the semester when time seems unlimited vs the end of the semester when time is in short supply.

Time is always in short supply.

Leave a Reply

Your email address will not be published. Required fields are marked *