We Don't Type for a Living: Plural Programming Practices for Productivity and Profit
- 1. we don’t type for a living plural programming practices for productivity and profit @jonfazzaro
- 2. It is counter intuitive, but 2 people working at a single computer will add as much functionality as two working separately except that it will be much higher in quality. With increased quality comes big savings later in the project.
- 5. Pair programming makes the heart-grindingly slow process of software development go faster.
- 8. manual work (time * people = value)
- 9. knowledge work (time * people = maybe)
- 19. for an idea to go from your head into the computer, it MUST go through someone else's hands. Llewelyn Falco Agile Coach Awesome Name-haver
- 24. focus
- 26. The human eye has an almost infinite capacity for not seeing what it does not want to see.... Programmers, if left to their own devices, will ignore the most glaring errors in their output—errors that anyone else can see in an instant. Gerald Weinberg The Psychology of Computer Programming
- 29. cost
- 33. done
- 34. If you want something marked done, give it to a solo developer.
- 35. If you want something to get done, give it to a pair.
- 36. If you want something to stay done, give it to a mob.
- 37. If you want to go fast, don’t go alone.
- 38. we don’t type for a living plural programming practices for productivity and profit @jonfazzaro
Editor's Notes
- #3: They buried the lede here What do you say after reading that? Quality is good. Savings are nice. But what do the people paying for the nerds hear?
- #4: Eat your vegetables This doesn’t connect Here’s the real problem. “What else does it do?”
- #5: Software development is SLOW You will be up against this problem for your whole career as a dev
- #6: Here’s my rewrite. Two nerds, one keyboard How does this make us faster? Another question.
- #7: Is this cat writing code? No, he never stops typing We don’t type for a living
- #8: (story of the chalk x invoice) Great story Illustrates manual work vs knowledge work thinking
- #9: folding laundry, doing dishes, factory line linear and predictable
- #10: This is knowledge work Unpredictable, reacts non-linearly to adjustments of time and resources. Setbacks This is because in knowledge work, we are forever slipping backward each time we move forward. Trial and error A creative process
- #11: So, back to this guy. The slowness of knowledge work. The slowness of knowledge work is an illusion based on our expectations of manual work. We completely discount the cost of the creative, generative work inherent in it, And we expect to be able treat it like linear task work. This inevitably disappoints everyone. So, how do we make this faster? We have to think about going faster differently.
- #12: A Wrinkle in Time? Not the movie Traditionally, we think of the shortest distance between two points as a straight line. To get to point B, we just have to slog it out.
- #13: Tesseract Folds time and space We can speed up the non-linearity of Knowledge work with a non-linear approach to the work.
- #14: But this doesn’t just happen when we put two developers together. There’s are some dos and don’ts here. pairing is not two machines one watches one checks reddit Every team is different, Every pair is different. But here are two essential practices to ensure that you and your team get what you pay for with pairing and mobbing.
- #15: A new way to split the work. We know from Lean Manufacturing that to make any system go faster, we have to first identify its bottleneck. The bottleneck in Knowledge Work is our singular locus of attention in search of the next right right idea. By adding more brains to the same problem at the same time, we are adding resources to the bottleneck. we multiply the points of attention. we multiply the ideas generated, and we multiply our chances of finding the right next thing to try.
- #16: The traditional approach. We add more brains to the problem. But then we give them each their own programming task! division of labor. Frederick Taylor was right about one thing--in order to be efficient in solving a problem, we have to first decompose it. We have to break it down. But there’s a problem here. Do you see it?
- #17: Let’s take a closer look. What is it that we do when we develop software? This is just what happens when coding. Not to mention the other components of software development like requirements, QA, and operations. There’s a lot going on all at once. How much of this resource—this brain—is really able to focus on that bottleneck? Could we break this down further?
- #18: This is where pairing and mobbing come in. We split the job between two or more people. This lightens each person’s cognitive load And the pair can move quickly through a complex task.
- #19: As far as what you do want to do, there is one sentence I keep coming back to to articulate it.
- #20: Driver and navigator
- #22: Attention is the resource to manage for Knowledge Work. We can’t manage attention directly, but can care for it through managing our time. This problem multiplies in importance when programming in a group. One of the most counterproductive states a team can be in is pairing/mobbing while one or more in the group are tired, and their attention is flagging.
- #23: An effective way to prevent this is to Take a break before you need a break. As a group. This sounds easy. But it is not. Use a timer. Preferably one that BREAKS your flow.
- #24: Mobster is my current favorite. Yes, I said to BREAK your flow. But Csikszentmihalyi! Flow state!
- #25: I read that 20 minutes? Don’t we need focus to code? Yes, sometimes
- #26: Don’t we need to focus for long periods of time to do this work? Don’t nerds take up to 20 minutes to recover from an interruption? Don’t we need that DEEP FOCUS to write code? Yes, sometimes. But remember, when we solo, we’re trying to juggle all of the aspects of this job at once. This of course would require intense concentration, and would probably fall to pieces at an interruption. More often than not, what I see happen as a side-effect of this difficult balancing act is that one or more aspects of the work suffer from even trying.
- #27: To this point…
- #28: And what you may call focus, Gerry and I call tunnel vision. So yeah, when you are heads-down and focused, without interruption, you could write the wrong code really fast.
- #29: This results in cleverness complexity bugs technical debt rework Things that hemorrhage your project’s budget. But are they more costly than two people coding together at the same computer?
- #30: Let’s talk about cost. “Getting done sooner doesn’t help me if this breaks my budget in half, we’re paying devs by the hour!” Again, this assumes a linear relationship between fingers on keyboards, hours, and features delivered. Focus on division of labor But we don’t deliver “tasks”. We deliver Features! What about the integration? As it happens, this aspect of managing project work is so often missed, that the missing it itself has a name:
- #31: In other words, we focus and obsess over the division and individual execution of the work While totally blowing off how difficult and expensive it will be to integrate and deliver the valuable whole afterward. So while we are clutching our pearls over the idea of more than one nerd at a keyboard, we turn a blind eye to the time spent on merges, deployments, code reviews, back-and-forth with QA, etc. dribbling every task out to a slow, painful, expensive death.
- #32: DevOps/Continuous practices are an attempt to directly take on coordination neglect On a team of solo developers, this is an uphill battle.
- #33: Pairs, however, have a much easier time. Mobs have it licked. Speaking of finishing things.
- #34: Let’s talk about this word. What does this word mean? Ask everyone on your team to write down an answer. Every one of them will probably be different. But what if you combined those different answers? Together, they are probably right Solo dev: Rework Bugs Side effects Missed scenarios Technical debt SLOW drastically reduced by plural dev Gives us a living definition of done
- #39: Thank you!