Pawel Brodzinski on Leadership in Technology

Whatever it takes to lead a team, build a product, and run a business

Category: project management

  • Why We Don’t Write User Stories Anymore

    There was a time when we were writing user stories to describe requirements. I’d say they worked fairly well for us. But we don’t do this anymore.

    We were using user stories as a technique which allowed us to describe bigger chunks of functionality. There was one bigger sub-project or module and it had more than 10 user stories attached (usually closer to 20) and a handful of non-functional requirements. During development we were often going through several stories at once as technical design didn’t map directly to the stories. The stories were more of input to design session and a base for test cases than stand-alone bites of functionality.

    Then we switched to Kanban. One of consequences was we reduced the size of average feature which was going to development. It no longer had 15 stories attached, but it wasn’t a single-story task either. If we were still writing user stories to each Minimal Marketable Feature we would probably have few of them. My guess is 2 or 3 most of the time.

    At this level stories become pretty artificial. I mean if you think about 2 stories connected with one feature, i.e. administrator can configure this magic widget and user can use this magic widget to do, well, the magic, you can pretty much tell these stories intuitively in your head. Writing them down becomes overkill.

    Besides that I think the often cited role of user stories which make usage scenarios completely clear is overrated. If you can talk with developers in language closer to the code, and functionality description is much closer to the code than telling user story, you’ll be better understood. The standard problem here was that functionality description wasn’t precise and it often became disconnected with usage scenarios.

    The answer for this problem is: make features as small as possible (but only as small as they still does any difference). Small features are easy to define, easy to understand (even for a developer) and easy to chew. It is pretty hard to screw them.

    There’s one more reason why I don’t consider user stories as a must-have. If you happened to create software which will be used by other developers or administrators at best, like some magic box with a lot of APIs and command line interface as the only UI, you should know what I’m talking about. If you write stories for this kind of software you end up with a bunch of “as a developer I want to call the magic function which does the magic” stories. Doesn’t API specification make more sense?

    I don’t say user stories are bad. They aren’t. But they don’t add value all the time and in every situation. This is just another practice which should be used only as long as it does make sense in your specific situation.

  • Manual Testing Is a Must

    The other day we were discussing different techniques aimed at improving code quality. Continuous integration, static code analysis, unit testing with or without applying test-driven development, code review – we’ve went through them all. At some point I sensed someone could feel that once they employ all this fine practices their code will be ready to ship as soon as it is all green after build on a build server. I just had to counter strike:

    “I’m not going to ship any code, no matter how many quality-improving techniques we use, unless some human tester puts their hands on it and blesses it as good. Not even a chance.”

    I’ve seen the situation quite a number of times: a team of developers who are all into the best engineering practices, but unfortunately at the same time sharing a belief this is enough to build top-notch applications. I have a bad news for you: if the answer for a question about quality assurance team is “we don’t need one” I scream and run away. And don’t buy any of your apps.

    It just isn’t possible to build quality code with no manual testing at all. You may build something and throw it against your client hoping they will do the testing for you. Sometimes you can even get away with this approach. Be warned though – you’re going to fail hard way soon. The next customer may not like the idea of doing beta-tests for you, actually most of them won’t, and then you’re screwed.

    Manual and automatic testing is fundamentally different. With automated testing, no matter if we’re talking about unit, stress, regression, integration or end-to-end tests, you go through predefined tracks. You had to create the test in the first place so it does exactly what you told it to do.

    On the other hand humans pretty often had their own minds which they happen to use in unpredictable ways. They may for example make up some test scenarios on the fly or they may sense a subtle scent of a big scary bug hiding around and follow their instinct to hunt it down. Every now and then they will abandon beaten tracks to do something unexpected, i.e. find a bug which would be missed otherwise.

    Note: I don’t say what kind of procedure you should follow in manual testing. Personally I strongly believe in value of exploratory testing, but any good tester will do the job, no matter what your manual testing procedure look like.

    Actually this post was triggered by recent discussion between a few bloggers including Lisa Crispin, James Shore, Ron Jeffries, George Dinwiddie and Gojko Adzic. The discussion, while interesting and definitely worth to read, is a bit too academic for me. My approach is that there is no universal set of techniques which yields indisputably best results. While I like James arguments why they dropped acceptance testing and support his affection for exploratory testing I share Lisa point that even in stable environment our “best” approach will evolve and so would techniques we employ.

    You may exchange acceptance testing with exploratory testing and that’s perfectly fine. As long as some fellow human put their hand on the code you want your users to touch with mine that I’m good with that. I don’t care much which approach you choose or how you call it. It is way more important who does the job. Experienced quality engineer with no plan at all will do way better job than poor or inexperienced tester with the best plan.

    Just remember if you happen to skip the manual testing at all I’m afraid your app may be good only for machines as only machines tested it.

  • Random Thoughts on Estimation

    A lot of discussion on estimation recently. A lot of great arguments but a lot of good old mistakes we’ve already went through. This brought me to a few random thoughts on estimating techniques.

    1. Estimation technique which involves discussion between different people is better than one which just simply uses their estimates as input.
    2. Using factual recorded historical data is better than basing just on experience.
    3. Smaller tasks can be estimated way better than big chunks of work.
    4. Every estimate starts with some kind of guess at the very beginning.

    I know these should be obvious yet I’m surprised how often people:
    – forget about them
    – deny these are true

    Then they head towards wild guesses with some magic number applied, which may have some sense but not when used instead of real estimation.

  • Is It Possible to Over-Communicate In Project?

    While explaining another thing which I thought was obvious for everyone in the team but appeared as not clearly communicated the question came back to me: is it possible to over-communicate in project? I dropped the question on Twitter and expected answers like “Hell no!” Or “Maybe it is possible but no one seen that yet.”

    Responses surprised me though. Author of Projects with People found problems of being too detailed for the audience or revealing facts too early. Well, what exactly does “too early” mean? When people already chatter on the subject at the water cooler is it too early? When managers finally become aware of chatter is it still too early? Do we have to wait until management is ready to communicate the fact (which is always too late)?

    Actually gossips are powerful and spread fast. The only way to cut them is bring official communication on the subject as soon as possible. Hopefully before gossiping is started. Which does mean early. Earlier than you’d think.

    Another thing is being too detailed. This can be considered as unnecessary or even clutter. Clutter is an issue raised by Danie Vermeulen. If something doesn’t bring added value it shouldn’t be communicated. If we kept this strict we could never post any technical message on project forum since there always would be someone who isn’t really interested which framework we’re going to use for dependency injection or how we prevent SQL injection and what the heck is the difference between these two. And how do you know what is a clutter for whom anyway.

    John Moore looks at the problem from different perspective – over-communication can be bad when it hurts morale. I must say I agree with the argument to some point. Some bad news isn’t necessarily related with people’s work (e.g. ongoing changes in business team on customer side) and can be due to change. Then keeping information for you may be a good idea. However if bad news is going to strike us either way the earlier means the better. One has to judge individually on each case.

    Although I don’t see easy way to deal with above issues they remain valid. Actually I can agree it is possible to over-communicate yet there’s no concrete border or clearly definable warning which yells “This email is too much! You’re over-communicating!” at you whenever you’re going to send unnecessary message.

    The best summary came from Lech who pointed that risk of over-communicating is lower than risk of under-communicating. I’d even say that much, much lower. How many projects with too extensive communication have you seen? One? Two? Personally I’ve seen none. On the other hand how many projects suffered because of insufficient communication? I’ve seen dozens of them.

    On general we still communicate too little. Yes, we can over-communicate from time to time but I accept the risk just for the sake of dealing a bit better with insufficient communication which is a real problem in our projects.

    How does it look like in your teams?

  • Sure Shot Method to Improve Quality of Your Estimates

    That’s the old story: we suck at estimating. Our schedules tend to be wrong not only because there are unexpected issues but mostly because we underestimate effort needed to complete tasks. There’s one simple trick which allows improving quality of estimates. It’s simple. It’s reliable. It’s good. On the minus side you need some time to execute it.

    Split your tasks to small chunks.

    If a task lasts more than a couple of days it’s too big – split it. If it’s still too big – do it once again. Repeat. By the way that’s what agilists came with over years – initial size of user stories was intended to be a few weeks long, now it’s usually a few days at most.

    Why it works?

    • Smaller tasks are easier to process in our minds. As a result we understand better what we’re going to do. As a result we judge better what means are needed. As a result our estimates are better.

    • Smaller tasks limit uncertainty. In smaller room there are fewer places to hide. The fewer places to hide the fewer unpredicted issues there are. The fewer unpredicted issues the more exact are estimates.

    • Small tasks mean small estimates. A month-long task should be read as “pretty long but it’s hard to say exactly how long” task. Bigger numbers are just less precise. Splitting tasks ends up with smaller chunks of work. Small chunks of work end up with small-sized estimates. Small-sized estimates mean more precise estimates.

    As a bonus, during task-splitting exercise, you get better understanding what should be done and you early catch some problems which otherwise would appear much later and would be more costly to fix.

    And yes, the only thing you need is some time.

  • Evidence Based Scheduling – One More Way to Improve Your Estimates

    Somehow my recent discussions are mostly about estimating. Last discussion on PM Clinic happens to be on the very same subject. It reminded me about great technique which can improve your software estimates. It is called Evidence Based Scheduling and I’ve learned this from Joel Spolsky’s article.

    The basic concept is pretty simple. Jack the Developer gets the task. Actually a bunch of them. He delivers his estimates basing on his guts. Estimates are most likely crappy but at the moment it doesn’t matter much. As Jack does his work he tracks how much time he really spends on the task. It may happen it takes two days instead of planned 4-hour effort. Velocity for this task is 0,25 and it goes to the history of Jack. When Jack’s history is long enough you can use it to judge how crappy his estimates are.

    Evidence Based Scheduling uses Monte Carlo simulation which takes some effort to calculate results. However Joel brings some good news here:

    “Most estimators get the scale wrong but the relative estimates right. Everything takes longer than expected. (…) This common estimator has very consistent velocities, but they’re below 1.0. For example, {0.6, 0.5, 0.6, 0.6, 0.5, 0.6, 0.7, 0.6}”

    After completing a dozen of tasks you can get pretty much insight which number you should use to multiply Jack’s estimates to get something close to reality. Of course full-blown simulation will probably lead to better results but you get the idea.

    There’s one more gem in Joel’s posting which is worth stressing:

    “Fix bugs as you find them, and charge the time back to the original task. You can’t schedule a single bug fix in advance, because you don’t know what bugs you’re going to have. When bugs are found in new code, charge the time to the original task that you implemented incorrectly. This will help EBS predict the time it takes to get fully debugged code, not just working code.”

    I know it brings quite a lot of effort for Jack the Developer, since sometimes he fixes bugs really fast and accounting time spent on each bug fix to proper original task adds enough hassle to be reluctant to do so. However the ends are worth means this time.

    I strongly recommend you reading whole article as it delivers all the details about Evidence Based Scheduling.