I've been spoilt by pairing, I no longer prefer to work solo.
I've also seen organisations trying to substitute asynchronous code reviews for pairing. They aren't really substitutable. Just as pairing has antipatterns, so do code review systems.
As an analogy, I've coached olympic-style weightlifting. The difference between coaching in person and coaching online is night and day. And coaching beginners pretty much demands fast feedback to be most effective.
Performing the task of weightlifting is incredibly different than performing the task of programming. Much evidence suggests that programmers perform better solo because they can more easily reach states of psychological flow, and because they are less inhibited by social interactions which modulate certain parts of our thinking and can inhibit creativity, etc. At the very least, this is more highly true of introverted personality styles, and creating mandates for policies like pair programming is extremely insensitive (almost to a point of blatant discrimination) towards people with differing personality and learning styles.
For me personally, I can't learn anything from other people saying stuff about it audibly. I have to read documentation or code. I am the weirdo who actually reads my car's maintenance manual to figure out how to change spark plugs or use the jack to lift the car when replacing a tire. Someone can tell me how to change a tire all day long and it just doesn't help at all. I even played quarterback on my high school football team and even at that I had to actually read books and biomechanics papers about the throwing motion before I really improved. I went to football camps and everything and it just did no good. For me, I really only absorb information when I literally tune out all audio (especially the sound of people talking) and read.
I suppose I serve as something of a counter-example, then. Just because beginners in weightlifting (or any given activity) benefit most from fast, in-person feedback does not at all mean that beginners in some other activity (like programming) will also benefit from that kind of feedback.
I've also worked in high pair-programming environments, in one case with a group of other experienced, extremely talented engineers, and in another case with a larger number of average to mediocre engineers.
In both cases, across the board, pairing slowed us down and led to worker dissatisfaction. The few times when pairing came in handy were those cases when a person needed to begin working on a section of code or a subsystem that someone else had explicit experience with. Then, pairing them briefly (for, say, half a day) to force the novice to get up to speed quickly and have an "oracle" nearby was really helpful, but after about the 4 hour mark it just started slowing both people down, caused the novice to learn too slowly, and focused hyper amounts of attention on issues that wouldn't have needed to be brought up if the novice instead focused on solving their own problems and boiling down their issue list into a smaller, asynchronous list of just the most urgent needs.
And, truly, the only reason that those first four or so hours of pairing worked at all is that the documentation for the code was poor. If the exact questions and answers that were discussed during pairing had instead been turned into an internal StackOverflow and/or an internal quick start guide, it would have helped far more people get up to speed far more quickly than pairing, while only costing the senior dev a single up front cost.
Pairing is the definition of an activity that doesn't scale, and is arguably less beneficial than well written user guides + StackOverflow-like tools that allow for documenting the learning process.
>
Performing the task of weightlifting is incredibly different than performing the task of programming.
My point was that coaching is coaching. Fast feedback is more effective on great many learning tasks than slow feedback. It's why we obsess over closing loops.
> At the very least, this is more highly true of introverted personality styles, and creating mandates for policies like pair programming is extremely insensitive (almost to a point of blatant discrimination) towards people with differing personality and learning styles.
We discuss this question a lot internally. Speaking for myself, I don't participate in most of the social activities happening in and around the office -- I want alone time to recharge.
Sensitivity to others -- empathy -- is one of our core values. It has to be. We're collaborators. We're learners and teachers. We have to think about our pair, we have to be sensitive to their rhythms and limits.
That said, there's a self-selection process too. Lots of people either read the job description and nope their way out, or go through a pairing interview and nope out.
> I have to read documentation or code.
We discuss this too. It's perfectly acceptable for a pair to agree to split up and read docs.
> I suppose I serve as something of a counter-example, then. Just because beginners in weightlifting (or any given activity) benefit most from fast, in-person feedback does not at all mean that beginners in some other activity (like programming) will also benefit from that kind of feedback.
Some feedback is fast. Some is slow. Pairing allows for both. Asynchronous code reviews do not, especially since the incentives are whacky. I've seen a few such systems at different companies now and there are common antipatterns: for example, tiny patches getting the nod quickly -- but anything more than a bit tricky languishing for days or weeks, becoming stale. Codebases dispersed across forks and branches and patches; enormous fleets of patches in flight and no way to test the permutations.
> In both cases, across the board, pairing slowed us down and led to worker dissatisfaction.
Since we're trading anecdotes: I've never seen this. Ever.
I've been working this way continuously for two years now, with developers from multiple companies in multiple industries with varying levels of experience, wildly varying cultures and power structures and cultures and procedures and pain points. The uniform win has been pair programming -- even with developers who came in promising that they'd hate it.
> If the exact questions and answers that were discussed during pairing had instead been turned into an internal StackOverflow and/or an internal quick start guide, it would have helped far more people get up to speed far more quickly than pairing
There's nothing about pairing that precludes internal Q&A (we have that, informally and formally) or writing documentation (a good README is gold).
> Pairing is the definition of an activity that doesn't scale
There are more than 120 engineers pairing on Cloud Foundry now, spread across multiple offices on multiple continents in multiple teams. And we're hiring. And so are other Cloud Foundry Foundation members.
We scaled it. There were definite growing pains and we've had to eat humble pie garnished with crow a few times. But it scales in the same way you scale any other project. Better, in some ways, because engineers rotate between teams and carry deep insight of distant codebases to their new teams.
The thing is that two years ago, I would've agreed with you. I took my job in Pivotal Labs because I wanted to learn and I was prepared to hold my nose.
Now I'm an extremely annoying proselyte. Converts often are.
I understand that you don't feel the same and that you probably never will. That's OK; it's a big industry and different people can find the way to work that suits them best. I just wanted to put the case that pair programming is, on my fulltime, professional experience of the past two years, fucking awesome.
I also just wanted to clarify about scaling: taking some (semi-large) number like 120 and claiming this means "pairing does scale" seems totally wrong to me.
"Scaling" does not refer to taking some large number and dividing it by 2 (120 engineers -> 60 pairs; 1200 engineers -> 600 pairs). That is emphatically not scaling.
Scaling is about leveraging the expertise of one of your most dramatically productive engineers by dispersing that person's help, wisdom, advice, etc., to more than one other person at a time.
Pairing doesn't scale precisely because the best person you have can only assist the exact same number of people that the worst person you have can also assist (namely, 1 in each case, since they are both part of a pair).
And, as Brooks noted long ago in The Mythical Man-Month, adding more people to a given task often just slows it down. Communication overhead tends to grow as O(n^2) or worse. So if you generalize a "pair" into a "triple" or a n-tuple, the process of pairing suddenly no longer helps and often dramatically hurts.
So when I say, "pairing doesn't scale," what I mean is that the best it can ever provide you is a 2x improvement in information transmission (and often it's far worse than that, and in all the cases I've seen it used first-hand it has actually been less than 1x).
I think it's a little disingenuous to just say, "Hey, we did it with 120 people! So it scales!" That's just not what scaling is about.
By way of comparison, one single, well-written quick-start guide can have almost an arbitrarily large improvement factor in terms of information transmission. You can hire one engineer who is good enough to write that document, and then send it to all of the other 119 engineers.
In my experience, even if that document is just so-so, it's still miles better than having to mentally context switch out of coding mode and into social conversation mode every few seconds to elicit the same info verbally out of a pairing partner sitting with me.
As I said further up, nothing about pairing precludes any other channel of communication.
When I need to learn about a tool or system built by anyone else, the docs are the first place I look. If that doesn't help, I reach out for help.
> So if you generalize a "pair" into a "triple" or a n-tuple, the process of pairing suddenly no longer helps and often dramatically hurts.
Nothing about pairing changes the number of people in a project, or the number of potential pathways between them (some might argue that it halves the paths, but that would be twisting the meaning of pathways).
As I said above, we use pretty ordinary tactics to manage building large distributed systems, such as breaking them down into human-sized chunks of functionality that interact in limited ways.
If you like, I can ask about having you visit an office. It probably won't change your mind, but if you see something we don't, that's useful and valuable feedback.
In pairing, a bi-directional pathway is established between Person A and Person B. This means that, during the activity of pairing, Person A's influence is exclusively dedicated to Person B, and vice versa. If everyone working solo is the baseline, meaning everyone only has the "1x influence" of their own work, then switching to pairing (if it works) increases this to 2x -- yourself and the pair partner.
In a traditional meeting where Person A is the speaker and there are N-1 audience members, Person A's influence is equally effective (ideally) across all N-1 unidirectional communication channels. If Person A is a really great engineer, from whom the other N-1 can all learn a lot, then leveraging just Person A at a factor of N-1 could be vastly more powerful than leveraging everyone but only at a factor of 2.
When you say things like:
> Nothing about pairing changes the number of people in a project, or the number of potential pathways between them (some might argue that it halves the paths, but that would be twisting the meaning of pathways).
it makes me think you misunderstood me, because I never claimed that pairing changes the number in a project or anything else.
However, you can't deny that in an environment where pairing is the expected modality for collaborative work, there truly is an emphasis on precisely the N=2 case, and other forms of communication necessarily have to be reduced to allow for pairing time and consistency in the act of pairing. It's not possible to do all other modes of communication and also pairing, because when you're pairing, you're necessarily not doing the other modes of communication.
My statements are about inherent limits to pairing. You can't extend the concept of pairing (based on 2 people) to triples, 4-tuples, and so on, without it leading to O(n^2) communication problems (e.g. if you tried to put 25 people together in one "mega pair-like group" it would be a big mess).
This is a bad property of pairing; exactly equivalent to saying that it does not scale.
I think the better way of phrasing it from my earlier remark was this:
When Person A is pairing with Person B, Person A is necessarily limited to only helping 1 person. If Person A is your best engineer, this is really bad. Telling Person A to stop doing pairing and to instead do all the same advice/Q&A tasks that would have been done during pairing and do them in some other way that scales (like user guides and SO) would immediately be better.
You can't have it both ways. At a certain moment in time when your best person is engaged in pairing, it necessarily means your best person is being wasted by having merely a 2x impact, instead of being pulled out of pairing and placed into some other activity that has a much higher factor of impact.
In real life, you'll never do any of these things perfectly, pairing or not pairing. But the point is that pairing is inherently not scalable, not even in principle.
It has nothing at all to do with whether your team also writes user guides or also answers questions on an internal SO site when they are not in the act of pairing. It's about the opportunity cost when they are in the act of pairing.
I've been spoilt by pairing, I no longer prefer to work solo.
I've also seen organisations trying to substitute asynchronous code reviews for pairing. They aren't really substitutable. Just as pairing has antipatterns, so do code review systems.
As an analogy, I've coached olympic-style weightlifting. The difference between coaching in person and coaching online is night and day. And coaching beginners pretty much demands fast feedback to be most effective.