How do interruptions impact different software engineering activities
rdel.substack.com
rdel.substack.com
I worked at a small company where a significant portion of my effort went toward shielding my team from the distractions created by a CEO who couldn’t seem to help meddling in every aspect of the business. I think it’s because he started out doing, or at least being involved with, many of the functions of the company and had a hard time letting go as we grew. Even after the organization grew to 50+ people he couldn’t keep himself out of the nitty gritty details, but the format of the distraction changed over time. Instead of walking up to people and interrupting them in person (a double whammy according to this study, including both an “important” person and the in-person aspect), he would send what we called “Slack attacks” throughout the day. These were paragraphs-long Slack messages without any semblance of organization, punctuation, or line breaks. Fortunately, many of these messages were sent during the very early hours of the morning so they could be dealt with first thing in the AM, but that wasn’t always the case.
In the first phase I literally moved my team location and reorganized the desk arrangement so it was harder for him to get in and bug everyone. I had to “guard” the area and try to stop him from physically entering the space, which was always a strange dance. I couldn’t control his Slack messaging behavior so I worked with people to understand that while yes “the CEO is asking you for urgent work in Slack” seems like a valid reason to switch gears, but instead let me work to figure out what actually needs to be done and we’ll catch up later about what to do.
It was a weird dynamic but there was no doubt the distractions were a drag on performance. Every time he went on vacation we saw a marked increase in productivity, and more creative solutions seemed to come up as well. I don’t wish this type of environment on anyone but in a way I’m glad to have gone through it and learned some lessons about interruptions and how to avoid them.
When I was a middle manager at a past job, my team kept getting bombarded with meeting requests (it was a "fire drill all the time" kind of place).
I ended up making a 4 hour "team meeting" every Friday morning to give them focus time.
I mentioned to them that they could have done the same on their own if they wanted. My team lead pointed out that since our calendars were visible in the department, having me as the organized gave the meeting more "weight" and so line employees were less likely to push for time in that slot.
But I wonder what would be the CEO's side of the story. I'm sure his behavior was bad, I'm not pretending the opposite. But maybe he also had reasons (misguided reasons, sure, but still reasons) to act as he did. Or maybe he did not have any reasons, it's also possible.
I've observed cases where indeed people were disturbing too much software developers.
But I've also observed cases where software developers were not enough aligned with the business side, despite them being 100% sure they were.
It's a tricky situation: being in the team means that you are not impartial, you don't have a remote view, you are only seeing one side of the story. I had situations where I've observed some non-dev team explaining their needs, then the software team went away, and then came back with something that was not what the non-dev team asked. Not only the non-dev team was indeed not satisfied, but I was agreeing to them: it was not what they explained, I was there, I understood what they said at the time. Worst, in the majority of the cases, the non-devs don't just say "no, it's incorrect", they try to find a compromise. Usually, it comes from the fact they have no idea what is possible or not, and just assume that if the devs did not do what they were expected, it means there are good reasons for that (either it is not possible, or that there were others things they did not know about, such as other requests from other part of the business). As someone with a lot of developing experience, I was able to see that the problem was that the devs just underestimated the need to fully understand the business side.
It's very tricky, because for a dev (or a dev-side person), it is very easy to just ignore that. If they ask for adjustment, it's "they don't know what they want". If they point at some requirements and underline that it does not mean what the devs thought it meant, it's "these requirements were badly written". And in the majority of the case, the business-side just makes do with the sub-optimal solution and the dev-side is considering that they successfully delivered. And similarly, I've also seen some devs being happy to be very productive, going very fast in developing something ... that the business did not need at all. When not adopted, it was again blamed on the business-side for not using the solution they've developed, rather than to wonder if what they have done was indeed productive or not.
Why is it hard to accept that maybe being a CEO doesn't mean you're good at your job?
The fact that one is bad at their job does not imply that someone else is good at their job. Why is it hard to accept that maybe being in a dev team doesn't mean you're good at your job?
The world has changed significantly since it was written, so you have to translate their measurements to modern contexts, but it's easier to translate real data on human behavior than to guess based on guesses.
Breaking the Flow: A Study of Interruptions During Software Engineering Activities
Yimeng Ma, Yu Huang, Kevin Leach
(2024) icsi Portugal
it's not properly quoted, but it's directly linked, prominently.
yes, I agree, peopleware is on the must have reading list, but this article certainly is not guesses on guesses
If you have an average of 4/day or more, it will probably consume all of your working time. But one of them isn't a problem at all.
And there's the problem, because it's both not a problem in small numbers, and a DoS attack still in small numbers.
So yeah - have 4 of these a day and you've wasted an hour. Have this every working day and you've already wasted half a day per week.
It gives my mind a few seconds or a minute or two to do background processing and potentially come up with a better way of doing something.
Or to realize that I’m not working on the most important thing in the first place.
These breaks were taken at a "natural" time, when you felt one was needed. By definition, an interruption is a "break" when you don't want one.
When I’m deep in the weeds on a task, stepping away for a walk in the park, a workout, or to prepare a meal or some coffee, affords me the opportunity to clear the micro details of the task from my mind while retaining the macro at the tip of my attention. This state is very frequently where the best insights on the task emerge, whereas an interruption resets both micro and macro attention entirely to something else.
For me, it depends. When working at home with kids around, I actually dread the restroom break or making myself a cup of tea, because in those two minutes I'm out of the home office, it's virtually guaranteed they'll drag me into whatever is happening at home at the given moment, wiping my focus entirely.
Sometimes this does break me out of being fixated on doing the wrong thing, but more often than not, it just plain prevents me from doing any thing at all.
Picking what to work on for [time] without external additions is a pretty good improvement over getting new assignments daily.
Or maybe you're just doing it wrong, I'm not here to defend anything codified in Agile, I'm just saying that elements of what most people call scrum has been the best I've seen in over 20 years. Maybe I just missed the days of locking yourself in your single office.
The basic building blocks can actually work great, it's often how they are executed and how many places go very much micromanagement wrapped in agile's skin while also instituting lots and lots of meetings - total opposite of what both Agile and XP advocated.
Some of the best agile/scrum I ever worked under came from "Scrum master" who explicitly stated that one of his jobs was to see what approaches work and change what doesn't work for the team.
unfortunately from my experience, "what works" can vary wildly from person to person. One person wants to be given a task and left alone to understand it and write some throwaway code as part of the understanding process. another person wants to bikeshed it to death first before letting you touch any code. so the first person has to hide the work on the prototype and pretend to come up with the insights for the first time during the bikeshedding session.
Sometimes you can compromise, sometimes you can figure a proper allocation of resources, sometimes you can part amicably instead of murderously.
Ironically, if I were to give one-sentence summary of what the idea behind Agile is, it would be exactly that: "see what approaches work and change what doesn't work for the team". Running through that feedback loop rapidly is the whole point of Agile, and the one thing that actually distinguishes it from typical management methods at the time (as opposed to silly comparisons with "waterfall" boogeyman that never even existed).
Everything else is process - the thing that's supposed to be sculpted. Focusing on that is committing the sin explicitly called out in the Manifesto - putting process over people.
If you're phrasing it like that I've been doing agile since way before anyone invented the term.
But that's only 10% of cult agile, which is what is practiced generally.
In the end, very simple Kanban is the way to go if your org can handle it.
> Individuals and interactions over processes and tools
> Working software over comprehensive documentation
> Customer collaboration over contract negotiation
> Responding to change over following a plan
> That is, while there is value in the items on the right, we value the items on the left more.
The thing I think everyone has grown to universally despise is the cargo cult of “Scrum” that got bolted onto that.
The problem is all the "Agile methodologies". Every single one.
(Also, notice that "Agile methodology" as an idea is already against the first principle there.)
The DOD agile BS PDF covers most of it.
https://media.defense.gov/2018/oct/09/2002049591/-1/-1/0/dib...
Once the consultant forms start to monetize the GAO's agile assessment guide things should get better IMHO.
Unless that gets co-oped somehow, which is possible.
How GM screwed up on the NUMMI plant shows how it is a Taylorism problem if you want something outside of tech.
But even TOGAF and ITIL are adjusting because the feds will require it.
Taylor measured people loading pig iron into train cars, faked a lot of things and the BCG/McKinsey types packaged and solid it.
It has always been BS, is part of why the US manufacturing sector failed as well as why the USSR failed.
Pretty hard to kill but it needs to die.
I'd argue that "agile" where I have done it and it worked well is at its core about tightening up OODA loops - with code (refactoring+tests+iterative improvements), with customer interaction (frequent interaction/experiment driven tests), with team organization (retros, adapting team processes).
Feedback loops are, however, not mentioned once in the philosophy and neither are any concrete examples of "agile" or "not agile".
Agile as a concept will stop being broken when people stop saying "to me, agile means X". Which will probably never happen - I suspect people will just stop talking about it one day and start using different terms for all the concrete steps that are sometimes filed under "agile" but which actually work.
(Also worth noting that scrum and most of the best known agile methodologies predate the manifesto; the manifesto was formed from guiding principles of those methodologies, not the other way around: https://agilemanifesto.org/history.html)
The only problem is that all the problem people never take the hint, and the well-behaved people tend to take it to an extreme. So on practice it both doesn't get rid of the interruptions, and delay everybody's work due to the latency you need for avoiding interruptions.
More than once I've been asked discretely by my PM if everything is OK with me, because half an hour down a team meeting whose contents is "99% of noise entirely irrelevant to me", I'm too exhausted trying to control my movements to keep my body from telegraphing the fact I'm bored out of my mind. Turns out, to an outside party, I start looking like I haven't slept the night, or am about to get a stroke, or something.
It's not at all a mental illness. That's ignorant.
Some of the most technically adept people I know are on the spectrum and they are incredibly valuable. We're talking Chief architects, lead/senior engineers, Network engineers, and the list goes on.
People like that probably have divergent neurochemistry compared to you.
Learn to leverage that asset and stop being a Karen.
Feigned wisdom isn't.
I often find this with people who think they are good at "multitasking".
On one hand: well, when you have to keep big mental models in your head you don't want to be interrupted. The big headphones serve to indicate you are in the middle of work.
On the other hand: have you ever been on a construction site? You'll hear some radio usually.