I think most companies in the Valley would acknowledge this.
The thing is - most companies are not looking for the best talent, they're looking to build the best teams. And one of the things that helps a team to gel is having all the members co-located together. It would be wonderful if the physical presence barrier was broken and you could setup virtual presences that were as efficient as physical ones, but something intangible seems to be lost when you only interact over e-mail.
It works for GitHub because it makes the company a very incentivized user of their own product - they're developing for precisely the sort of organization that is their market, and so they understand their market better than if they acted like a normal start-up. But if they were playing in a fast-moving consumer market or trying to capitalize on a recent platform or innovation, they'd get eaten alive.
You'll be surprised what google hangouts on a wired connection on both ends can do, especially when on-ramping people.
Why wired? Packet loss from wifi seriously degrades the conversation. Wired on both ends and packet loss is practically non-existant.
There are downsides to not being distributed too. Being distributed for example lets engineers control their interruptions and environment far more. If companies really really cared about building the best teams, they wouldn't do the nearly industry standard open office layouts. Open office layouts make for easy management and harder development.
I use Hangouts extensively; I work for Google, we eat our own dogfood as far as videoconference software goes. That's one of the main reasons I'm pessimistic about distributed teams. Despite having probably the fastest fiber backbone in the world, and despite having an inside track to the VC software we use, it still turns out that being physically present with the people you work with is pretty important. Enough that I'm moving offices this week, 3 cubes down, so that I can be 30 ft. closer to the rest of my teammates. There seems to be something intangible that is lost when you move out of shouting distance, and even though we have the technology to shout over 3000 miles, it doesn't seem to be the same.
One thing (of many) that simply must be handled better for video conferencing to even getting close to be a viable replacement is the "eye contact" problem. People need to be able to look in someones eyes on screen, and see them looking back at them.
For the occasional call for e.g. an interview, I make it a point to look into the camera, rather than on screen, and it makes a big difference in how I get perceived, but that turns the video aspect into a one-way thing - like looking into a TV camera. While looking at the screen means you loose a lot of the signals you'd get face to face just from having a harder time reading facial expressions, and from the way peoples interaction change when they don't get eye contact with you.
There's lots of other bits and pieces that impact its usefulness too, but to me solutions to the eye contact issue needs to be there before there's much point. E.g. in my experience with teams seated close together, eye contact is often the primary indicator that you've gotten someones attention and can start asking a question and know you're not interrupting something important.
Without that, you're either back to the keyboard, or "polluting" the environment with distracting speech - often I see team members using IM even to someone at the next desk over when failing to get eye contact, exactly to avoid disturbing everyone else by speaking, so that's a non-starter.
It's an incredibly important part of communication that most current video conferencing seriously hampers.
Moving out of 'shouting' distance for me in my office actually helped me considerably. That little bit of extra distance filtered out many trivial interruptions. Maybe your teammates and managers are good about it, but I feel like many are not. That combined with the noise machines open offices can be it can get pretty bad. Also some people can get bad with the shoulder surfing.
To create that 'we're in the same room' experience, you need to leave G+ hangouts on the entire time. You can't just initiate calls and then end them and do it several times a day.
If distributed working was a no-brained with no downsides we would all be doing it. However I do believe the upsides are several factors greater than the down, so the cost benefit does win out - if the effort is put in.
I have got most out of distributed working within weeks of physical meetups for sprints, so my view is meetup every quarter for a weeks sprint (somewhere nice of course!)
The benefits of easier to control working environments, focus on deliverables, honesty, greater team choice and flexibility, easier professionalism, all make winning easier
You must be on a bad connection. I use Hangouts over WiFi daily and rarely have problems with degraded video or audio quality. Hangouts has plenty of other bugs but quality isn't one of them.
If your in a busy wifi environment, such as a hotel, apartment/townhouse complex or similar, all of the other wifi traffic can significantly degrade the connection. Not everyone lives in a suburban house with only 3 wifi networks on their laptop and an ability to completely control their environment.
Video conferencing is really sensitive to congestion, latency and packet loss, not so much throughput. Missing a word here and there can make it fairly frustrating, and with wifi you are bound to have that happen a few times an hour even with good wifi connections on both sides.
I know you pointed out github but I'd have to disagree about the physical part for consumer facing markets. I'll point to Fictive Kin[1] and 37signals[2] as just two very successful examples.
37signals is B2B2C: their primary market is small businesses. I'd never heard of Fictive Kin before, and a quick Google doesn't turn up anything other than their website, where it looks like they're more of a consultancy than a product.
That depends entirely on the people involved. Some people find they need to be physically co-located with their teammates, but some of us find it's easier to work with other people when relieved of the distracting noise of physical presence and granted the clarity and time for thought of text communication. If you fall into the former category, fair enough, by all means you should organize your situation accordingly, but don't assume what's true for you is true for everyone.
I think you make really good points. I'd turn it around though a little. GitHub can break the mold (scale a startup/grow a team) where others have failed partly because it's in the DNA in their product. But once they show how it's done it'll be modifiable/repeatalbe and usable by folks focused on other markets.
It could be; I'd very much like them to succeed, because I love working with a lot of smart people in tech but hate paying Silicon Valley rents. Maybe that's why Andreesen Horowitz invested. But it's very much an open market thesis; I don't really see any evidence this is repeatable across markets yet.
Wrote a relevant blog post about this issue not too long ago. The article states the obvious of course. Talent isn't evenly distributed, but it is distributed. What's most compelling for perhaps Valley and New York companies only, is the wide disparity with regard to compensation:
Two Silicon Valley startups raise $1M and they each have need for a great team. Team One is on site. They spend a lot of otherwise productive time attempting to hire great engineering talent away from other equally impressive startups in the area. They pay $150k per person (let's call it $200k fully loaded) for talented, but not phenomenal people. Team Two is fully distributed. They decide to hire the best, no matter where they are, and at $150k per person everyone outside of Silicon Valley gives them a look with much less effort on their part. They save time, they save on office space in SV, and, best of all, they build a team of truly phenomenal folks who are happy to be making $150k, because it's likely $50k higher than they were making at whatever job they had before they got recruited.
Assuming you can build a better team in the Valley is only relevant if you have a reasonable expectation that you'll find similarly talented folks at the same price no matter where you are. That is simply not the case.
But that's part of the point, isn't it? If SV companies need to adjust for cost of living in SV, the same salary could attract a better employee in an area where adjustment is not so necessary.
The cost of living in California is high only compared to...the rest of the USA. But if you are talking international, I find San Francisco to be a bargain on everything but rent, and for many cities it is very much a toss up even there.
You should see the ridiculousness in the Santa Barbara/Goleta area... Same cost of living as Silicon Valley, Less Pay. Higher cost of living than LA, San Diego and Seattle but engineers get paid less than those areas.
I know that Montecito was one of the highest priced zip codes in the USA, as of a few years ago. SB is stuck between the ocean and the mountains, with no room to grow.
I am the sort of experienced full-stack hacker many Valley companies would love to have on their team. However, I like smaller cities, commuting exclusively by bicycle, having an enormous garden, eating locally grown food, easy access to amazing outdoor adventures, and mild winters. You'll probably never find me in the Valley or in the Boston or New York area. The bay area is too populous and northeast winters are terrible.
No amount of money can change those things. Quality of life is worth more to me than high salaries.
I do, however, like interesting work and brilliant coworkers. I mostly work with local businesses that are doing interesting non-high-tech stuff, like mass-custom lean manufacturing, but it would be great to have more options for working with people who know programming at least as well as I do.
Working locally with people does seem to beat the isolation of remote work, however. It probably makes sense for a company focused on collaboration tools to have a team that relies on collaboration tools to function.
Why do you like having a giant house. I've never had this feeling and would like to understand. (I currently live in what I would consider a reasonably sized apartment in SF.)
Parking. My own fenced in yard to BBQ, dog to run, kids to play in (if I had any). Tons of space to put furniture, exercise equipment, etc. My own garage to park in. Ability to easily have dozen+ people in my house for gatherings, etc. Think of it like hard drive space or RAM :-)
GitHub is a pretty amazing company and story. Their culture, the tools, etc are all very inspiring, and the whole community definitely benefits from their experiments. This meme always gets me thinking about how to attract the best talent for our company.
That said, holding them up as a model of how to build a tech team never quite feels like a good idea.
GitHub is a developer product, built for developers by developers. That essentially makes all of their employees domain experts in the product as well. They can also robustly dog-food their product. Both of those factors in my opinion make it much easier to have a successful distributed team. They can trust everyone to make good decisions absent a strong product team.
I've worked at 5-6 different startups with varying levels of distributed teams and never felt that distributed communication of product, architecture, etc was ever as productive as with the co-located teams. Trying to involve the remote people always felt like "extra work" and it wasn't a smooth process.
I would be very curious to hear from non-developer product companies about their successes (or failures) with remote teams. Are new tools like Google hangouts, shared whiteboards, and github really enough to allow you to have a productive product/engineering team?
If you have to do "extra work to involve remote people" you don't have a distributed team. Adding remote members to an on-site team that works like an on-site team seems unworkable in my opinion. You can have remote developers that get assigned work like a contractor, but IMHO a remote developer can never truly be part of a team that functions like an on-site team.
Now, a truly distributed team is a beautiful thing and can in my experience outperform an on-site functioning team. I've worked on highly performing distributed teams that only saw each other in person once or twice a year. When a distributed team has a conference call, everyone generally calls in individually from their own phone/skype, even if some are in the same room. A distributed team working in an office uses the office more as a private co-working space.
The key tool in every distributed team I've ever worked on is chat, usually IRC. The great thing about chat is that it's asymmetric communication, but targeted and instant unlike email. In a distributed team, I can throw out a question on irc without worrying about interrupting anyone or throwing someone out of the zone by tapping them on the shoulder. A lot of non blocking discussion happens throughout the day that prevents having official meetings.
If a company genuinely wants to move from functioning as an on-site team to functioning as a distributed team, I'd suggest everyone on site rotate taking a week and working remotely, especially management! It can't feel like you are doing extra work for the remote members, but that you are changing your team to work in a distributed manner for the benefit of everyone.
We have tried that for the local team but it didn't get used despite trying 3-4 different chat tools (IRC, HipChat, I forget the others). People end up talking in person b/c it's easier.
Love this. Talent is all over the world. This is also why companies of the old style, non virtual offices, have multiple offices in some aspects.
New, distributed companies, such as github, 37 signals, small game companies etc, they will always have the best employees and even if an employee moves they are still actively on the team. For some industries like games, it might be the only way to gather all the people and skills needed as especially with games it is hard to find people in the same area with required skills.
Virtual teams lead to much less disruptive change to a team which leads to stronger teams. But like it is stated, all important communications need to be virtual and through that channel first, it will fail otherwise and some can't do it. The world is big, cities are big, why destroy competitive success because it is to hard to coordinate? Traffic, moving, costs (living/goods/travel) are all better suited for easy access and ultimately company survival with virtual teams.
The absolute best thing about virtual teams is it is all about delivery and getting things done. Minimizing BS, time wasted and shipping product.
Virtual teams can come together for integrations, planning, celebrations and be as strong or stronger as physical logistical teams. It doesn't work for all industries but in technology/creative it can truly free up people to focus on work. How many offices have you worked at that after a while actual productive work hours go down to 3-4 due to all the excess stuff, then you have to work at night or after hours just to keep up with all the wasted time at the office and travel and general life that pops up. Virtual team are delivery first and that is the right way to be productive.
This is somewhat off topic; but I have to say I am quite disappointed by the way in which PandoDaily puts out its events.
I was quite looking forward to seeing Tom talk; so I got my first PandoMonthly tickets. The event started at 6, with the talk scheduled to start at 7. I only really cared about Tom, not so much networking, so I showed up at 7. Made it there by 7:01, only to be greeted by a bouncer that told us that they had just closed the doors. They were instructed by PandoDaily not to let anybody in after 7. When we mentioned that we had paid for tickets; we were told "too bad."
PandoMonthly, from this experience alone, is already the worst event I have ever seen put on. Way to tell your customers "fuck you."
If the talk started at 7, it would be rude to arrive anything later than 6:50. I think this is great, blocking people at 7. There is no other way to get things running on time and without disruption.
I feel like developer-heavy organizations would be more successful at this. A major factor is the tools facilitate a lot of the communication (commit messages, bug tracking, build notifications, documentation, etc.). They're also comfortable using tools and generally adapt quickly to new platforms. And in my own personal experience, it worked with their personalities.
I'd like to hear more examples of this working successfully at scale in highly-cross functional teams, with distributed marketing, sales, design, support, and product groups. Communication always seems to be a complaint at companies, and I'd like to hear more about how this was addressed.
I think that the chat room seems obvious but it is not to a lot of older managers and it is very important.
I have worked with some (generally older) developers who just didn't have online communications skills and even refused to use a shared chat room. They might be seemingly able to exchange a few emails or IMs, but then the next week they would be busy, stop using emails/IMs, but the day you go into the office or call them suddenly the communication increases 10-fold. Or they did it but didn't know how to use it. I.E. they would spread the chat out into many many rooms and fill them with irrelevant design speculation and idle chat.
I think the title of this is a bit misleading. The message is much less about "the valley" and more about building a culture around geographically distributed collaboration.
Not that I disagree with the sentiment or to rag on GitHub or their management (I think they have a nice product). But this article is simply quoting the CEO of a small company that has been around for very little time. Is the Valley so insular that they think this is truly a management success or is this just a terrible article?
I'm not a huge fan of the Valley having lived there for a long time and moved away however.......what the Valley does really well is understand very quickly when something new and revolutionary is afoot. That's the difference between picking the next 100 bagger or being a sheep. Github is a rapidly growing company hiring some of the best talent globally in a game changing way - and so far it's going against all the "common perceptions" about how you build and scale out a startup. I'd say watch this space closely.
The Valley does this by assuming that everything is new and revolutionary and forgetting that almost nothing is. The hits are remembered, the far more numerous misses are forgotten.
That's why it works at all. If the Valley was stocked with grumpy curmudgeons like me, there wouldn't be a Valley.
You aren't giving Github nearly enough credit for their importance today. Sure, they haven't been around long, and they might not be so critical in 5 or 10 years, but at the moment Github is arguably the most important internet company in terms of the number of other startups and established businesses that are completely dependent upon them.
Yes, of course we can all move our repos to Bitbucket or wherever, but currently there are tens or probably hundreds of thousands of the most innovative companies relying on their product. It won't be a minor inconvenience if Github goes away.
Tom Preston-Werner is a very influential leader. Aside from Github and other projects, he is also the author of Semantic Versioning, which has played a significant role in the success of Node.js and its standing as the most advanced system for modular server-side development.
> Yes, of course we can all move our repos to Bitbucket or wherever, but ...
> It won't be a minor inconvenience if Github goes away.
Yeah, I don't think you made a very good case for your argument there.
Some companies are completely dependent on Windows as their staff or software can't work with anything other than Windows. Not only won't it be a minor inconvenience, it would be an existential crisis for the company if they couldn't use Windows.
Github has done a lot for the internet. But I'm pretty sure we can live without it. There are many other competing services out there. I would say the best thing about Github is the fact that lots of big open source projects are hosted there. If they all moved to Bitbucket, would it be that big of a deal for everyone?
Some people would miss some features that were only available on Github. Most people just need a git:// URL to push and pull from.
Sure. It will be no big deal if 50,000 companies have no source code repository next week and have to switch to Bitbucket immediately before they can continue working.
Guess those VCs who were throwing money at github had no idea what they were doing.
There was no big reason for Github to exist in the first place. Git is distributed anyway. People could always just fire up their own servers for repo hosting.
LOL.
Github is so obviously critical and more importantly quite relevant I can't believe that I am actually arguing about it.
I'm going to give you the benefit of the doubt for minute and reply to your comment.
If 50,000 companies had no source code repository next week and had to switch to Bitbucket, it would be a very big deal. That wasn't what I was arguing. I was refuting your idea that companies are dependent on Github. That isn't the case. If you took any tool away from 50,000 companies overnight it would be a big deal. That doesn't make those tools important.
Take away all the desks and chairs from 50,000 companies and switch them to standing desks immediately before they can continue working and it would be a big deal. Are desks and chairs as important as your Github account? If you answer yes to this question, just stop reading at this point.
If not, then you should see how ridiculous your earlier argument was. Github has a ton of features, I don't know how many of them you use but I personally just download repos. That's about it. That's the central feature of Github. Bitbucket has that feature. If a project moved from Github to Bitbucket, it wouldn't make a difference to me at all. I imagine my situation is not unique. That was my argument. I hope I've made myself clear this time.
Not sure what your "Windows is a dinosaur" remark was about. Do you deny the existence of companies that are completely dependent on Windows?
P.S. Having a whole sentence that consists of LOL is just extremely weird and out of place here at HN.
Irony is that this is coming from a San Francisco based, VC-backed media site, of which most of their writers (kinda assuming this based on their staff list) are located in SV.
I've done both and I have to say distributed is guaranteed to win in the long run.
In my last venture we had the Palo Alto office on University Ave, and I've walked a few miles up and down SandHill Rd, so when I say distributed works I'm contrasting it against past experience of trying to hire developers, architects and (worse) project managers in SV.
The challenges raised above in many of the comments about "being within shouting distance" and "eye contact" - together with the fear of distributed workers being passed over for promotions because of their lack of water cooler time - goes away when the ENTIRE team is distributed.
In a situation where the "core team" is within a fun, social campus and George is working from Seattle in his bunny slippers then it is very difficult to convince the in-office workers not to have chats over their cubicle walls and innovate during lunch on the bean bag chairs. Using the collaboration platforms like GitHub and Basecamp becomes a pain in the ass to them. But, when EVERYONE is distributed, then a lot of that goes away, and we've found adoption and proficiency with Google Hangouts (wired or WiFi, sorry) and all the other tools go way up.
In our current team for Infinite Monkeys[1] we've built the entire venture with funding from founders and SV sources, but a team that is purely distributed using over 200 contractors through oDesk[2], literally on six continents. It has allowed us to move beyond "hiring the best people who coincidentally live within 30 miles of San Mateo" to just hiring the best people. And so we have designers in Bulgaria, developers in Spain and India, marketers in Singapore, a bookeeper in Mauritius and a PR Lead in Michigan. It has allowed us faster release cycles and massive savings in our GoToMarket burn.
There are still challenges no doubt, and the fact is building this kind of startup even 36 months ago would have been almost impossible, but tools like Google Drive, oDesk, GitHub and Hangouts, I think are making it possible.
We have replaced the water cooler with 15five[3] (ironically a SFO startup) - asking each team member to write a 15 minute shared report of what went well this week, what challenges there are, and what ideas they have for improving the business. I'm getting better feedback now through my distributed team than I ever did with our hub and spoke teams before.
Lastly, and only because it hasn't been mentioned above, but I think another example of a long-term distributed company that has done it well is Mozilla. Say what you may about their products and direction, but the fact is their developer base is massively distributed and has been for years.
Most Silicon Valley startups don't need top talent. They're marketing experiments with a small bit of technology and a lot of painful support work (due to massive, accumulating technical debt) that can only be done out of VC-istan dues-paying as young people take on pager duty for the job they think will get them investor connections in 6 months so they can do their own gigs (ha!)
They need some flashy talent (young, clueless Ivy/Stanford grads) so they can tell investors that they bought a bunch of Ivy stock on the cheap. They don't need top talent.
On the other hand, hungry talent (in a world where most people will never buy a house) and prestige-seeking talent and clueless/young talent are useful to them, especially if it congeals in one metropolitan area (such as Silicon Valley).
The Valley's more than good enough for most of these firms. They complain about the "talent shortage". There isn't one. They're just complaining about paying more for talent than they think it should cost, and "what they think it should cost" is Dickensian.
Of course, there are companies out there (if few) that demand extremely high levels of talent, but at that rarefied level, you're either (a) going to be small enough that you can do it within one location, or (b) going to ignore location outright, as OP suggests.
> Most Silicon Valley startups don't need top talent. They're marketing experiments with a small bit of technology and a lot of painful support work (due to massive, accumulating technical debt) that can only be done out of VC-istan dues-paying as young people take on pager duty for the job they think will get them investor connections in 6 months so they can do their own gigs (ha!)
The same thing is true about lots of Google (your favorite company) jobs. For many positions, they don't need brilliant people hacking Lisp, but rather they need competent Java or C++ programmers to make sure that their code base is "good enough". The result is that you get things like Google Calendar. What's wrong with Google Calendar? John McCarthy put it best:
"Now of course Emacs is merely and editor but nevertheless is does permit significant user modifications and the modern operating systems do not. Let me give an example: one thing that I have done is to add a feature where you can put in a file, the name of another file and even also a location in that file. And then with a single key press it will search forward till it finds its file name and go to that filename in place. And I use it for putting references to email messages in my calendar file. That's just one of the uses of it and I asked about the new Google calendar file. Can you put a reference to email messages in there? And the answer is not only can't the user do it, but even the implementers can't do it, but to me anyway it's very useful. When I receive email about a meeting or a seminar, or something like that, to be able to put a link to the email itself into the calendar file and Emacs permitted me to do that myself, without my even having to become a real expert in the inners of it, and I think this will be important. "
The same thing is true about lots of Google (your favorite company) jobs. For many positions, they don't need brilliant people hacking Lisp, but rather they need competent Java or C++ programmers to make sure that their code base is "good enough".
Yeah. It seems that most of the large technology employers want to hire the best people to do middling work, ignoring the unhappiness that ensues when people are over-leveled for what they're asked to do.
Closed-allocation needs to die in a fire, in other words.
You often refer to something called "VC-istan". That name implies that the VCs have some sort of scam going on where founders/employees are the victims. Am I correct?
VC-istan is the hypothesis that the VCs and the companies they support function more like a postmodern corporate entity than a fleet of competing investors and companies.
The name itself doesn't imply anything negative. "-stan" just means country. I could just as easily have called it "VC-land" but that sounds light.
Of course, my view of the existing VC-istan is negative-- and the culture of co-funding and note-sharing is both unethical and illegal-- but there could be a good VC-istan.
"Dickensian" is right, I once knew a poor soul that were making 1/3 what I was in a startup. I've known several that were ~1/2.
I'm not sure being full-stack really accounted for the full difference. Probably more to do with being viciously acquisitive and enjoying sales and negotiation.
Also, I'm not working on the fashion thing anymore, thought of a better marketing experiment to try.
Startups will pay well if you know how to negotiate and have leverage, but if not, you'll get screwed. Usually the equity is pretty pathetic too. They're much more likely to exploit transient weakness than larger corporations, which have lower bounds on salary range as well.
The thing is - most companies are not looking for the best talent, they're looking to build the best teams. And one of the things that helps a team to gel is having all the members co-located together. It would be wonderful if the physical presence barrier was broken and you could setup virtual presences that were as efficient as physical ones, but something intangible seems to be lost when you only interact over e-mail.
It works for GitHub because it makes the company a very incentivized user of their own product - they're developing for precisely the sort of organization that is their market, and so they understand their market better than if they acted like a normal start-up. But if they were playing in a fast-moving consumer market or trying to capitalize on a recent platform or innovation, they'd get eaten alive.