Buchheit on Google, Friendfeed, and Start-ups

Paul Buchheit, developer of Gmail and founder of FriendFeed, talks with EconTalk host Russ Roberts about the evolution of the Gmail project, how innovation works and doesn't work in a large corporation, how Google has changed as it has grown, and corporate culture generally. The conversation then turns to social networking and what might be coming next. The discussion concludes with Buchheit's observations on Silicon Valley and the power of failure.

Explore audio highlights, further reading that will help you delve deeper into this week’s episode, and vigorous conversations in the form of our comments section below.


Joe Green
Sep 21 2009 at 8:20am

For anyone interested in strategy and/or entrepreneurship, this is fascinating. I am particularly drawn to the discussion of organizational challenges to sustaining profitability/growth over time. Reminds me of John Roberts great book “The Modern Firm”.

It is curious that Paul, who was desirous of a certain type of organizational environment (ie., not the bigger Google and its changed incentives), was willing to sellout to Facebook. What are the differences in the organizational structures of the two firms; Google v. Facebook?

Sep 21 2009 at 12:52pm

The discussion on privacy made me post this note I saw on another website.

The government’s position with respect to law enforcement recalls a joke from the old Soviet Union:

Policeman: “Don’t park there! That’s where our leaders park!”
Citizien: “It’s OK, I locked my car.”

Sep 23 2009 at 10:38am

Dr. Roberts,

In your conversation you made the comment regarding your guest’s summary of “Born to Run” that it “sounds like a hoax”. While I have not read the book, I did do some research regarding running shoes for a biomechanics seminar while I was a graduate student at Berkeley. The scientific literature was so counter to our perception of running shoes in popular culture that I titled my talk: “The Great Shoe Conspiracy” (albeit somewhat tongue in cheek).

The short version is that there is plenty of research to support the notion that running shoes increase the risk of ankle and knee injury through the combination of: (1) increased moment arm of the ground on the ankle joint because of high heels; (2) neutralization of a corrective reflex that is triggered when fascia connecting the tarsal bones is stretched by stepping on something that could penetrate the foot.

A somewhat outdated but reasonable review can be found in “The biomechanics of running” T.F.Novacheck, Gait and Posture v. 7 (1998) pp. 77-95.

From the economic perspective, please remember that “the invisible hand” of the market can only protect us from bad products when the harm they cause is obvious for all to see. Whenever the injury is delayed and the dissemination of questionable research can muddy waters and protect a lucrative market, the economic incentives align against the public good (at least in the short term).

Scott Link
Sep 24 2009 at 10:27am

The discussion about the shoes led to the most provoking observation: [Organizationally we have a] “tendency to put up too many barriers between ourselves and reality and consequently we end up making bad decisions ’cause we’re not really getting the right feedback” (about the 45 minute mark…)

Great discussion!

Russ Roberts
Sep 24 2009 at 1:35pm


I said it sounded like a hoax. I didn’t say it was a hoax. The way I heard Paul was that these new shoes caused you to run on your toes. In my mind I saw a ballerina-like gait induced by the shoes. That’s why I said it. As someone who is always looking for more comfortable shoes, I have no trouble believing that current shoes or most shoes aren’t so good for us.

I like your point about information. I would say it differently. The truth is elusive. It’s hard to know what is real and what is a claim to advance someone’s special interests. I think markets help us distinguish between what is real and what is fake, but it can take time and it may not always be easy.

Sep 26 2009 at 5:12pm

Russ, you might also add, as one always can, regarding markets: They may be imperfect, but they always beat the alternative (what exactly that would be in running shoes is hard to fathom).

I like to remind myself when I see incredible business ineffeciencies (which seem to be almost daily) – as I am sure we all do – that we live in the wealthiest, most efficient time and place in the history of the world with a tailwind of two centuries of unprecedented exponential economic and technologic growth.

(It’s also true, by the way, that we don’t need the heels on our calfskin oxfords and bluchers, designed as they were, originally, for fitting in a stirrup, but men’s dress is inherently conservative. By that I mean, we like to look like each other and like our fathers, and to not have to be concerned with buying new garments everytime the leaves change colors. And, of course, for those men who claim to not like to look like everyone else and who foolishly decide to act upon the urge, we like to mock them incessantly. Usually men’s dress only changes slowly by dropping the most formal get-ups once or twice a century, typically a decade or two after major political events involving large scale revolution, death and destruction [e.g. knickers and wigs after the American and French revolutions, frockcoats after WWI, etc.] So, give the heels on men’s shoes time, we’ve only been off of horses for a century, or so. [Please not that this analysis does not apply only to mens’ dress clothes: nothing else can possibly explain the ubiquity of “cargo” shorts. I posit that it is an unalterable feature of male human nature, applicable in all times and places to varying degrees, irrespective of political affiliation]).

Sep 28 2009 at 12:32am

Dr Roberts,

I was going to ask, in relation to your response to Shai’s comment, if you would concede that markets tend towards certain types of mistakes while (for example) representative democracy tends towards other types mistakes.

I realised in phrasing the question the way I was one of the possible reasons economists have such a difficult time maintaining authority in public opinion. I was phrasing the question in a way that you would for a politician, trying to get them to break from their media message. Economists, particularly those close to the Austrian side of the equation try to stay “on message” in order to convince people. This is similar to politicians. When you hear Silicon valley type intellectuals (like many of those invited to this show), they often have strong libertarian tendencies, but they are very experimental too. This seems to lend them the credibility of a scientist.

*As an analogy, try to think of biologists (or thrown into) the creationist-Darwinist debate.

Sep 28 2009 at 1:12pm

I really enjoyed this conversation this past week. I’ve listened to it 5 or 6 times and find something new each listen.

One quibble. The problem with gmail downtime isn’t that gmail is down an unreasonable amount of time compared to other mail servers. It’s that an outage affects more people simultaneously than outages of private email servers. The same reliability factor weighs against centralized social networking, like Twitter and Facebook. Twitter was out for nearly a day recently because of a denial of service attack. A lot of people didn’t know what to do with themselves and a lot of businesses that rely on it were sunk for the day.

In the future social networking space, I think that instead of users picking formats (e.g Twitter, Facebook, Friendfeed, etc.), users will choose service providers or have social networking service included with their Internet. While it might turn out to be less reliable in terms of total uptime, the downtimes should be distributed better.

Another issue social networking needs to resolve is the issue of multiple audiences. There’s the common joke/gripe about the Twitterer who tells the world what he had for lunch. Perhaps some of that Twitterer’s friends are interested in that and some aren’t. Social networking needs to make it easier for the poster to sort and filter below the account level. On Twitter, I deal with this multiple audience issue by using several accounts. Unfortunately, that doesn’t work well when using the cell phone gateway.

But Russ and Paul… This is an amazing discussion. Maybe there aren’t a lot of EconTalk listeners that these nuts and bolts entrepreneur discussions. But they are the ones that get me thinking the most. Thank you both!

About this week's guest: About ideas and people mentioned in this podcast:

Podcast Episode Highlights
0:36Intro. [Recording date: September 14, 2009.] How did gmail get started? Work on email or personal information system, wide-open request. Useful information locked away inside of email. Built system to index and search own email, iterated from there, tried new things. Couple of years--actually relatively long process, just himself working on it, starting around late 2001. A little slow from the start, many ideas never launched; others launched much later, such as IM integration, added maybe a year after the initial launch. Initial launch was April 1, 2004; had to be invited for first couple of years. Originally difficult to get an invitation; selling on e-bay for maybe a couple hundred dollars. How many people working on it at its height? Left Google 3 and a half years ago. At time of launch, about a dozen people. What was the process for deciding what features it would have? Iterative. Brainstorming sessions; talking to people about their email needs. Two different systems--the system they wanted to be using and the system they were actually using. Big complex folder hierarchy, sorting mail. There would be Inbox 2; renamed Inbox when it got too big; leaving everything in the Inbox because they would fall behind. Usage pattern not well supported by existing programs. Look at email, either respond or star it for later; then when done want it to go away. Archive; can search to find later on. Advice? Get over the guilt. Some people love organizing their email; others don't. Automatic organization into conversations; sort by subject to find other messages in the thread; one line in the Inbox.
7:12Which feature most proud of? No one thing. Influential aspect: front end done all in javascript; at the time, people considered Webmail to be a toy. People inspired by it and modern browser technology. People today are building whole office suites on the Web. Correct that you were employee number 23 at Google? Fairly standalone entrepreneur within a company. How did that integration work out? Occasionally a little bumpy; at the time, most of the rest of the company was focused on web search and advertising. Questioned why email at all? Technically infrastructure turned out to be wrong. If you are crawling the web, not a big problem if you lose a few web pages; can fetch them later. With email system, you can't lose anything. Assumptions worked their way down to the low levels of decisions--need to be reliable. Distraction: off in a corner working on it. What happened within the company that made it more complicated? Argument can be made that companies can only do so many things well. Perception that Google was a search technology, back-end for products like Yahoo. Yahoo not too pleased when they discovered gmail.
11:32Successful because so many people are using it; but two issues. Gmail was down last week for a few hours, sporadic--people felt the tremor, not 100% reliable, impossible. Financial side: use of advertising on gmail. Reliability issue really important; important also to keep some perspective. Annoying to not be able to access email for an hour; but that can happen even if you are operating your own email. Power at house, lose electricity more often than email access. One provided by government, other provided by corporation. Seems really important, but at a practical level, doesn't cause that many problems. Financial side: little ads at the top of gmail. Ads on the internet somewhat annoying; find g-mail ones to be somewhat less annoying; even click on them every so often, much more often than on other pages. Directed advertising: payoff is pretty good, actually deliver an interested customer. Sufficiently optimized that you don't need to make much money off it to cover the cost. Keeping operational costs low. Recent podcast with John Nye, levels of trust and comfort when interacting with strangers; how after an economic crisis, like current one, people are less comfortable with the capitalist system, markets. People have raised about Google and gmail question about stored personal information--which is what allows Google to make all those ads so well-targeted. Gives people the willies. Emails, searches, all stored somewhere on Google. Doesn't concern Buchheit so much; has broader concern that all email can be subpoenaed. Google has incentive to be careful with the data; but they are subject to the law. Not google-specific; can be compelled even if it is on your own server. Obstruction of justice. Getting together in a room with someone versus doing it online, over IM, or in email, which can become evidence in a trial. Encourages people to destroy the information; unfortunate for the world to create incentives to destroy historical records. Trust google; assume the incentives are in place; assume some kind of firewall/security procedure in place to keep people from going through people's email. Safer because some of the best people in the world are there doing it.
20:23Buchheit credited with the phrase "Don't be evil." True? Any other significance? Yes, true. Company meeting in early 2001, trying to decide on company values. Typically bland, forgettable company values. Took on a life of its own. Has a lot of value: gives everyone in the company license to question the decisions. Important process; often attitude is that it's not your job to worry about that, it's your job to shut up and do what you are told. Can quit, but better to question than quit. Everyone had permission, created a more thoughtful organization. One in a list, but the only one anyone remembers. Ritz-Carlton's motto: Ladies and gentlemen serving ladies and gentlemen. Memorable. Only so many things can be on your mind at a given time. Left Google around May 2006; got there June 1999--seven years. How did company change while there? Got a lot bigger. Little startup; had low expectations. Search market crowded in 1999; Altavista was the giant in search, hundreds of engineers, money to burn, huge traffic. But the people seemed smart, good learning experience, a lot to learn. Really liked working there; prior had worked at Intel, big company, gray cubicles. Post-graduate education. How did it change? Culture is hard to keep. Innovated on a lot of fronts, interesting products and technology; organizational methods relatively conventional, same strategies as done elsewhere. Incentives change over time. As a little startup, if the company fails you all fail, so you are all in it together. Bigger company, incentives more complex. Stock price might be driven by something in the ad market rather than the cool thing you produce. Negative incentives: People are incentivized not to screw up, not to cause disasters; positive incentives tend to disappear. Which will you do if a project could cause a 10% change of being an enormous success versus a 10% chance of making you look dumb? Will your boss be okay with that? Will they take over the world? Will continue to do well. Primary competition is with Microsoft. Monitoring, oversight, things slow you down. Industrial era methods and thinking. Not like running a factory, can't measure output very well.
29:25It is hard to measure, but when leading the G-mail team, with 12 people, if one not doing a very good job, probably knew it even if you couldn't quantify it. The people on the team know who is contributing and who is not. Let people pick who they want to work with and for rather than top-down picking people. Generally want to work with people who want to get things done, if you are the kind of person who likes to create new products; people with complementary skill sets. Engineers. Building effective teams is pretty tricky, want perfect mix of personalities and abilities. Classic economics problem: how to attribute team output to individuals. Allow people to freely associate. Subtle things; output is not the only thing, who compromises at the right times. If you cause everyone else to quit, you are probably a net negative. Hard to figure out from the outside.
32:27Bing.com--is it a threat to Google? Not obviously a threat, but clear that Google is trying to take it more seriously. Credible product. But there is a lot of technology behind Google, so it can be hard for people to know how far behind they are. Yahoo: redoing ad system; came out with something Google had three years prior. Coming up with the new iPod is hard. Lead, remain focused, hard to overcome. Trying keeps them on their toes.
35:03FriendFeed.com. After leaving Google, took off about a year, year and a half. Friend, second person on gmail, Sanjeev Singh, talking about what's next; started working on different ideas, keep an eye on other people. Taylor, North, behind Google maps leaving to start a startup of their own; dinners, chatting about what to do next. Pry into the year and a half off: Did you plan a specific time? Always interested in startups; at the time just needed to take some time off, work, first child born, medical issues, cancer for brother who died. Really do nothing? Hang out at home. Most people don't get to take a year off. Turned to socializing with talented folks. What is Friendfeed? Tool for keeping up to date with friends; tool for sharing things. If I write a blog post, can put it on friendfeed, friends can read it. What's the difference between Friendfeed and Twitter? In practice, Friendfeed is more conversational, comments, long discussion. Both products evolve over time and in a somewhat convergent manner. Over time, more link-sharing at Twitter. Friendfeed allows images, richer experience; similar bucket. Twitter as status-updating. EconTalk on Twitter as EconTalker. People think Twitter is about telling people what you had for breakfast; use it as a miniature blogging platform; character is bracing, encourages brevity. Plus most of the time, though not how it started. Why successful with friendfeed? Twitter more successful; Friendfeed has loyal user base. Discussion on Friendfeed, aware of each other, more like a bunch of people in a room talking to each other. Same microblogging aspect to it. Picture of new shoes, toe shoes, running around the block; book Born to Run wanting to buy the shoes. Author had leg pains, doctor said he wasn't meant to run, problem caused by padding in shoes causing you to run in the wrong way; better feedback from the environment can work. Sounds like hoax but interesting. Dangers of insulating ourselves from reality too much. Shoes: Vibram, picture of feet.
46:00Facebook acquired Friendfeed last month. What will they do with it? Facebook and Twitter competing for how people pay attention to these micro-conversations. Team presently learning about inside of Facebook, figuring out next steps. Twitter community had evolved their product; Facebook evolving in same direction; added comments like Friendfeed. Different products, but headed to same destination. Different thing for different people. If friends are active, different from if they sign in once and don't come back. Told that what friendfeed does is combine facebook and twitter so you don't have to go to both; but no, challenge to face. Core concept is sharing. Notion of aggregation is that if you are already sharing somewhere else, it should just automatically show up here, so if you already have a blog, instead of having to make a new blog post you can just go to friendfeed; but downside is that it creates notion that Friendfeed is just a thing that aggregates. More than that. Business issues, sometimes features hurt you. Friendfeed: a way to share stuff. Sharing is in top three for what humans do. Backlash, among older folks; but technology helps us in being more human. Example: number of photographs on the web and what we can look at and share with each other. Every time you lower the friction, people share more stuff. Cell phone pictures shared with friends. Posting pictures of daughter enticed the grandmother to get a computer to join in the sharing of updates.
53:08Besides more, will the texture of it change? Certainly: as friction decreases, the kinds of things people share change. Shoes, sitting in backyard, easy to post a picture of shoes; 30 seconds to share it. Changes relationship with brands. Entire way economy works is changing. Previously, all about telling people what they needed to think and throwing money at it; now have to make customers happy. Yelp; out of bananas at restaurant for breakfast. In the old days, if you wrote a letter complaining you sometimes got a nice letter back with a gift certificate. Telling a few thousand it makes a big difference. Writing a letter to a restaurant is not something people do today--not a normal person, or you have a lot of time on your hands. Changes the kinds of feedback you get. Social element, joy of writing reviews; abused, Amazon book reviews, people plant negative reviews. Power of something like Facebook, with culture of real identities, see the truth. More meaningful, harder to do fake reviews; and also can see if someone is friends with the owner. Art to the inauthentic review; often pretty evident in both directions--rave review by Mom.
57:47What next? Working at Facebook. A lot of exciting opportunities. Silicon Valley: how has it changed and where is it going? Bubble at end of the 1990s, could start company without any real product. Dynamic place, swing of technology very quick. Core pretty solid; comfortable with taking risks. Moving from Intel to small startup, Google, not seen as a bad idea. Outside of the Valley not seen the same way. Important mindset. Difficult decisions to stay at current job or go somewhere else; will others around you respect you or think you are foolish?

More EconTalk Episodes

Search Econlib