How Paul Graham Is Wrong

I love Paul Graham’s essays and his latest is no exception: Let the Other 95% of Great Programmers In. I agree that the US deserves dramatically better immigration policies, but in the meantime I’m confused with the head-in-the-sand approach most tech companies are taking simultaneously complaining that there are lots of great people they can’t bring into the US, but being stubborn on keeping a company culture that requires people to be physically co-located.

In a region that prides itself on disruption and working from first principles, San Francisco’s scaling problem is pretty humorous if you look at it from the outside: otherwise smart and inventive founders continue to set up offices and try to hire or move people in the most overheated environment since there were carphones in Cadillac Allantes. This is where I feel like Paul Graham misses the most obvious solution to the problem.

If 95% of great programmers aren’t in the US, and an even higher percentage not in the Bay Area, set up your company to take advantage of that fact as a strength, not a weakness. Use WordPress and P2, use Slack, use G+ Hangouts, use Skype, use any of the amazing technology that allows us to collaborate as effectively online as previous generations of company did offline. Let people live someplace remarkable instead of paying $2,800 a month for a mediocre one bedroom rental in San Francisco. Or don’t, and let companies like Automattic and Github hire the best and brightest and let them live and work wherever they like.

Update: There is a vigorous discussion also happening on Hacker News.

51 thoughts on “How Paul Graham Is Wrong

    1. You don’t have to worry about most of that to start, and when you’re big enough that it matters you’ll have a HR/etc infrastructure in place to make it smooth.

  1. I think some people are feeling the pain enough to look for other places to run their businesses from. There’s definitely a migration (more like a trickle, but growing) of people and tech businesses moving from SF to Portland right now.

  2. I’d imagine the U.S. would have no shortage of engineering material if classical education – of which the formal study of logic is a core subject – was restored.

  3. My sentiments exactly. I also disagree that the talent lies outside the U.S. It’s here, it just can’t pick up and move to Silicon Valley. People have families, and other things they love about the particular place they’ve chosen to live.

    1. Even those that are willing to move, do NOT want to move to Silicon Valley.

      In my current location, I can get a 2BR apartment with a riverfront balcony for $510/mo. SV – good luck finding anything remotely close to that for less than $3000+, likely with a longer commute than my current one. (I spend more time walking in from the parking lot than I do driving to work…)

      I admit, where I live is pretty extreme in the “low” end of cost-of-living, but… Even if they DID want co-location, simply opening up a satellite office in a location that wasn’t as insanely expensive as SV would be beneficial for many companies.

      In a related topic – some people just don’t want to live in urban areas regardless of cost of living. For example, even if you don’t live in Silicon Valley, as far as I can tell, unless you want to work in one of Google’s datacenters, you can’t work for Google without working in a major city.

  4. I agree with that, but with reserve. I endorse remote, and actually that’s what I’ve been doing in the latest 11 years, living in Europe. But there’s a lot of developers, talented and not, who would love moving to the US, because that’s where all the fun :). If I would be given an opportunity, I would say yes, and regardless of whether I’d have to work remotely or on site, and, with a few exceptions, regardless of the location.

    But I totally agree that remote is the future – the technology is mature enough, there are several tools and services at our disposal, and they keep evolving. When I started freelancing, the primary tool was email, now we have skype, hangouts, VoIP, github etc. Collaboration is much easier and reliable, and distance is not as important as a good and reliable internet connection.

  5. Also, large companies want foreign workers because they will work for less money. If they paid a higher salary, more U.S. workers would relocate, and more U.S. students would study tech. Many of my friends and I moved out of tech to get larger salaries. We were all previously solid engineers and coders (not superstars like though).

  6. Sure tech companies can get some of the work done remotely. But that is no reason to keep people out of the USA. the best programmers (and other highly skilled people) are going to be paying taxes as well as paying people to mow their lawns, build their houses, and educate their children. It is a win for everyone. As far as training people in the US, of course we should be doing that, but just because we open the doors to some skilled people, doesn’t mean we can’t train at home at the same time. Besides, getting work done remotely doesn’t provide for any more local training than allowing immigration.

  7. I was working for a consulting company 25+ yrs ago where all work was “on location” in temporary project offices. 4 times a year we would have a company wide meet-up and sometimes project meets as well. What we didn’t have then was the internet or even mobile phones but we did have collaboration and project management tools. So distributed workforces are not new.

    But now it is very much easier to work remotely with all the technology. The irony for many people is they would get more “work” done if they were not in a centralised office with all of the “meetings” and interruptions.

    A distributed workforce works well when there is a strong shared culture, excellent collaboration tools, a project focus and a high level of trust. Open source projects already have the culture which helps. Many companies are better organised into smaller project teams and real economies of scale only apply in say a manufacturing situation.

    I would argue that a distributed environment with project teams in close collaboration is just the post industrial revolution model becoming the new norm. What has held it back before is better collaboration tools. For knowledge based work – especially it will become the norm in my view.

  8. Remote working is a niche manufacturing process that either goes really well but doesn’t last, or slowly works to the detriment of the worker. Here’s why…

    We don’t trust people with whom we haven’t shared a meal

    Mutual trust is crucial to shared culture

    Intellectual products require a higher degree of integrated culture, philosophy, virtues, and trust. The organization’s competitive advantage is made of these connections.

    I can purchase an SAE standard #10-24 x 1″ SS pan head, phillips-drive screw from 1,000 places and I don’t care much about the philosophy or principles of the people that made it. Because it’s standardized, I know it’s going to fit.

    You don’t want this to be true of you, your work, or your company. Collectively your intellectual products require the shared tribal knowledge that comes from working together.

    If you are working remotely, it means the type of work you do doesn’t require as much organizational insight and is subject to being commoditized.

    You are one step away from making screws and competing on a commodity basis.

    Either that, or you are so specialized that the company fears it can’t survive without you.

    This works really well in the short run, but it ignores how companies respond to fear and risk.

    As long as intellectual products are made by human beings, these people will need to be together in thought and location.

    1. This talk by Coby Chapple from GitHub, should offer you some insight: http://devslovebacon.com/conferences/bacon-2014/talks/remote-by-default-how-github-makes-working-remotely-not-suck

      Read Remote by Jason Fried and David Heinemeier Hansson: http://37signals.com/remote/

      I wrote a short post on changing our business from being 100% on premise to being mostly remote: http://nathanjeffery.co/2014/12/06/remote-work/

      I’m not saying you should throw having a central office out of the picture, but rather have it to compliment a primarily remote work force. Why should someone need to travel for any amount of time each day, to do something that can done with less interruption and more comfort from home.

  9. The other practice PG misses is building your organization’s talent through coaching, in-house training, education reimbursements, and other tactics that make people more competent as individuals and more capable in teams.

  10. I guess you can bring that down to a (financial) mathematical formula. It’s the cost of relocating a talented designer vs. the cost of making distributed teams really work. I have people working for me in The Netherlands, Bulgaria, The Philippines, and in several places in Germany. We use the *usual* cloud mix (Basecamp, Highrise, Zendesk, Dropbox, GMail, Google Docs, HipChat, Skype, Evernote, WordPress, Wufoo) to get things done. However it has taken quite a while to figure out how to do this. Not just in sense of “what tools to use”, but rather “how to use which tool for what”.

    Just like the atoms of the flower are not beautiful in itself – it is the structure / the relationship between them that makes for the beauty, the “social architecture” of that system (until it was really productive) has taken quite some time.

    Some of the time you can kickstart by organizational experience and giving the individual member a head start, but there is some cost / time involved to familiarize yourself with this style of work. And in my experience about 50% of the people do need a (at least weekly) direct physical / social interaction with colleagues in the long term.

    This being said it makes sense for new organizations without experience in this type of socially-distributed work to just pay the premium of having everyone in one room.

    Being an entrepreneur there is a chance I see: Find out what works and doesn’t work for companies like Automattic, GitHub, 37Signals, etc… and teach this knowledge to new startups (and also “normal” businesses).

  11. You’re pointing out the difference between the 1990s and now.

    Working and producing product should take advantage of today’s infrastructure. I think there’s a fear from some people about a distributed company.

    Thanks for taking a position and saying it quickly. Btw, I love Paul Graham’s talks also.

  12. I know a great deal about US immigration law (strangely– but life sometimes teaches you things you wouldn’t expect!), and it is very restrictive. I certainly agree that we need to take an honest look at how unworkable the system is for everyone and develop new policies.

    However, it’s definitely true that it has never been easier to work remotely, and some companies are beginning to prefer that in lieu of paying expenses to fly out consultants, etc.

    I think the trick is to be on the cutting edge of what you or your company can offer from a remote location!

  13. Companies mostly do not do this anymore because once they train the people, they are worth more to another company who has no problems paying a new hire that has XXX skills, whereas a company that already has that person hired and is paying them 20% less than what it would cost for someone who has XXX skills and trained them, will usually balk at a raise of more than 5-10% because no one gets raises that high.

  14. Matt—Take a look at Steven Berlin Johnson’s Where Good Ideas Come From and Edward Glaeser’s The Triumph of the City. Both describe in detail how ideas tend to emerge from serendipitous, unplanned encounters of the sort online tools are presently not good at facilitating, and perhaps that online tools can never facilitate.

    1. Familiar with the first one. I agree that online isn’t great at everything, and in fact teams at Automattic get together in person once or twice a year which is where a lot of the rapport in the team is built and there are good brainstorming sessions, but the default (the other 50 weeks) is online.

SHARE YOUR THOUGHTS