Wow. I guess I'm completely different from Jeff.<p>I have no need whatsoever for a "coding buddy." I'm considerably more comfortable working on a project solo. Even with a good team, I hate the coordination overhead.<p>I <i>can</i>, of course, and good team skills are important. I'm not saying that. But the coordination costs of software development increase super-linearly, so I tend to prefer using the minimum number of people that can possibly do a job. The break-even point is around 4-5 - after that, adding more people makes things <i>slower</i>, unless you can split the project into autonomous chunks.
<i>One guy working all by yourself alone. This didn't work at all for me.</i><p>Funny, I have trouble working any other way.<p><i>I was unmoored, directionless, suffering from analysis paralysis, and barely able to get motivated enough to write even a few lines of code.</i><p>I am totally grounded with tons of work and a million ideas of what to do and how to do it. I'm having so much fun, my biggest problem is finding time to triage all the possibilities in my head.<p>I understand OP, but that's just not my experience. Programming can be a lonely lifestyle. I have my code and my cats to keep me company. I love them both.<p>[EDIT: I don't want to leave the wrong impression; I am very social and used to find it challenging being alone. Not anymore. I just bury myself in my work, which is something that has to be done anyway. Now I don't want to be bothered when "I'm busy", but I <i>do</i> want to be bothered at most other times. Dinner with SO, time with friends & family, and visits to the hn virtual water cooler fit in perfectly. Now back to work - see you in a few hours.]
"The minimum remote team size is two."<p>By definition, I suppose, since a one-man shop has nobody to be remote <i>from</i>.<p>I disagree though. I'm self motivated enough to run with stuff on my own. I've built a lot of big successful things with a team size of one, and I find it the perfect team size during the prototyping and buildout phases of a project.<p>Sure, once you hit maintenance mode it's nice to have another person along to share the pain and help generate the discipline needed for months of refactoring and slow iteration. But when you're actually building something, I'm all about small teams.<p>So now we have two datapoints. If I were the author, I would have gathered a few thousand more before publishing this article.
I disagree somewhat. I dont feel that I "Bleed ones and zeros" but find the solitude in my home office far less distracting than an office cubical, and tend to be way more productive.<p>I think you really just have 2 basic types of people: those who want to get their work done(and do it well) and those who arent really committed.(for whatever reason, incompetent, lazy, bad work ethic, not interested in the work, etc...) If you have a team of the first type, you will get things done, if not, you wont. A less experienced developer with motivation to get things finished will not let distance get in the way.
I'm working remote and love it. But I'm a very introvert guy.
I don't like smalltalk, I don't like talking to a group of people, it exhausts me. But I love to philosophize with one person, when one thought generates another one, when both appreciate the thoughts of the other and go together beyond their own border.<p>Most of the communication with my teammates is done by phone
and a request tracker. If someone develops a new feature or
a customer has a demand, everything goes over the rt. It works pretty well, you can always see what your teamates are working on, and how they solved a problem.
Good article.<p>The part that was missing was: your mileage may vary.<p>I love hearing about how teams make it work. But over and over again I see a team do things one way that worked wonderfully well for them and then everybody else just copying that. Then it doesn't work.<p>Every project is a new game.<p>It's a well-worn anti-pattern
Remote work and distributed teams provides one heck of a challenge for sure. We've been finding our way by trying different things and succeeding or failing - but I certainly agree with what Jeff Atwood said about keeping up with chat and mail lists... We're very cross-platform, so video chat doesn't work all that well for us across Linux, Windows, and Mac, etc... But I think the weekly status reports to the team has been forgotten, so we'll have to start that up again. :)<p>I did a presentation for EclipseCon 2010 about our challenges that you can find here (with links to a slide): <a href="http://www.eclipsecon.org/2010/sessions/sessions?id=1156" rel="nofollow">http://www.eclipsecon.org/2010/sessions/sessions?id=1156</a><p>Great article though. Today it's nearly impossible to avoid a distributed team to get stuff done!
I can't stress enough the <i>persistent</i> part of the "persistent mailing list."<p>I worked as a remote employee in a group where we had half the team (couple dozen folks) in an office in Toronto, and the remainder of us scattered about in remote sites across the US.<p>We had a mailing list for questions along the line of "Hey, this question is important but not enough so to bug anyone in chat." It wasn't persistent in any usable manner; it archived to an Exchange shared folder hosted in the main office, and I learned that searching these folders over VPN was an exercise in futility.<p>Basically then you'd have new people hiring on and not having the organizational knowledge pent up from all the decisions made in the past. They'd make bad decisions in their work because either
a) they couldn't search the list for help and just decided to cowboy it, or
b) they mailed the list, got blown off for asking a question that has been asked dozens of times before they showed up, and will just cowboy it the next time.
It seems a bit odd to me that he talks about working remotely and how awesome it was, but then all the coding jobs for his site (which he was somewhat selling in this entry) are for people "in an office in NY".<p>But I agree, if you're working by yourself, you definitely need a very structured workplan, and it's very good to have someone to "report" to, even if they're not coders, but just as a grounding point for feedback and a passive nag to not slack off and get things done.
It's much better working on a truly distributed team than being the one remote worker on a mostly-centralized team.<p>My group at Mozilla is distributed across seven cities (I think) on two continents. We use all of the techniques Atwood listed. At Mozilla, where a large portion of the staff is in Mountain View, it helps that my manager is working remotely just like me - so I'm not more remote <i>from him</i> than anyone else is.
Some good tips in here, we run a fully distributed team and had overcome similar challenges.<p>One tip I would add is pair-programming on complex parts or sticking points through screen and vim, we have an ec2 instance dedicated for pairing which everyone has access to.
Mumble is a great project that works well for teams. Low bandwidth, high quality voice chat.<p><a href="http://mumble.sourceforge.net/" rel="nofollow">http://mumble.sourceforge.net/</a>
At the risk of plugging my own thing, this situation is the reason we built Twiddla, so I'd recommend checking it out if you're trying to build stuff with a distributed team.<p>Mostly it was to deal with communicating little graphical changes to remote designers without resorting to MS Paint and email, but it does chat and voice too, so it pretty much wraps up the whole "tools you need" section in one thing.
I've worked on a distributed team for three years now. We've found that regular phone conversations and weekly "status" emails (nothing formal) really help the process and keep everyone connected. Typically we use email and Skype to communicate, but if an email conversation begins to get too unwieldy, we simply get on the phone.
I wonder how the projects like freebsd, webkit, chromium, and thousands and thousands more are successfully going on without such pathos and attempt to impress the world.<p>Shared culture and vision plus trac with several plugins (or code.google.com) are enough.