Joshua Schachter started the collaborative bookmarking
site del.icio.us in 2003. As often happens with startups,
del.icio.us began as something Schachter built for
himself. He needed a way of organizing his collection of
20,000 bookmarks, and he hit on the idea of "tagging"
them with brief text phrases to help him find links later.
He put del.icio.us on a server and opened it up to other
people, and it began to spread by word of mouth.
For the first several years, Schachter worked on
del.icio.us and other projects, like Memepool and
GeoURL, while working as a quantitative analyst
at Morgan Stanley. But all the while, del.icio.us was growing. By November
2004, a year after its release, it had 30,000 users.
In early 2005, Schachter decided to turn del.icio.us from a hobby into a
company. In March of 2005, he left his job to "found" del.icio.us and focus on it
full-time, raising $1 million in funding.
In December of that year, Yahoo acquired del.icio.us for an amount
rumored to be about $30 million.
Livingston: Take me back to how you got started with del.icio.us.
Schachter: It goes back quite a while. In 1998 or so I created a website called
Memepool. There was an editor, with reader submission. We had a contribution
pool, and we'd edit and post stuff. It was chronologically sorted, updated
every couple of days--so it was basically a blog before that word came out. We
put a link at the bottom, "Send us an email. Give us good links." And people
would email us stuff they found on the Web. I would dutifully look at it and
write it down. It took me a long time to post anything, because I'm not a great
writer.
Over time, I had these links that just piled up--links that I'd found, or
surfed for, or had been sent in, or whatever. By 2001 or so, I had a text file filled
223
Joshua Schachter
Founder, del.icio.us
16
CHAPTER
with 20,000 links. I couldn't find anything in that file anymore, so I started putting
in notes. I'd put the URL, a space, a hash mark, and then a word or two
describing it. I think the first one was "math," so I could grep out all the things
that were #math and get all my items marked as math. In some sense, these
were the first tags.
After a while, I couldn't really do this, so I built a sort of next generation of
that text file, which was called Muxway, in 2001. It was a lot like del.icio.us.
There was a bookmarklet; you saved things; you could describe and tag them. It
was single-player--no one else could use it--but the actual website was visible
to other people. I discovered over time that people were subscribing to my
bookmarks. There were some 10,000 daily readers looking at my stuff. That was
interesting.
I did several other projects along the way. I did GeoURL. Something called
Reversible, that is long gone. Reversible was also like del.icio.us in many ways,
but different in a few key ways that made it fail.
In late 2003, I started working on del.icio.us, which is a multiplayer version.
I was actually trying to come up with a better Memepool--something between
Muxway and Memepool which was more vital somehow, and we ended up with
del.icio.us. I had it partially done for the first Foo Camp. I'd been invited to
Foo Camp for GeoURL, and I had stuff to show for del.icio.us, but I didn't
show anyone. I chickened out because I was embarrassed at the state of the
thing. So people were using it then, but it was more generally released later--I
think toward December of 2003.
Through 2004, I kept working on it and started to get press and lots of users.
By the end of 2004, I had 30,000 users.
Livingston: How were the users finding out about it?
Schachter: People were telling each other about it.
Livingston: You were at Morgan Stanley this whole time, right? What were you
doing there?
Schachter:I was doing data mining and proprietary trading algorithms.
Livingston: Why did you choose not to focus full-time on del.icio.us and what
finally tipped the scale?
Schachter: The economics didn't make sense. It still made sense to keep the
day job. But in late 2004/early 2005, my group at Morgan Stanley began to
come apart. There were a bunch of people leaving, so it was a natural time
to leave. It was a "Should I find a new job elsewhere?" kind of thing.
Livingston: When you were doing this in your spare time, did you ever say,
"Ugh. This is too much work"?
Schachter:Not really. I was always very careful (not anymore, because the guys
that I work with are better programmers) to structure the code--each chunk of
code wasn't larger than the screen--such that I could come in and look at it, figure
out what I'm doing, do it, and be done for the day in 15 minutes. So if I
could get one thing done a day, I was happy. A lot of stuff, if I could spend more
224 Founders at Work
time, I did, but as long as I could get one or two things done a week total, if I
didn't have time, I didn't have time.
So it moved pretty slowly. I worked on it for years.
Livingston: Looking back, do you wish you had left Morgan Stanley earlier to
work on del.icio.us full-time?
Schachter:I think it would have been very challenging to sell this as a venture
to VCs if I didn't have a great deal of user base and press to show. I think that
would have been a challenge. If I said, "Hey, I'm going to build a bookmarking
service," I would have never been able to get off the ground.
Livingston: Because the idea was so new?
Schachter: No. There had been plenty of other startups that failed doing this.
Backflip and God knows what else. So it had been tried and failed in the past.
Livingston: Why did del.icio.us succeed?
Schachter: First of all, because it was not a venture to start. I was building a
product and that's it.
Livingston: Did the others fail because they had too much money?
Schachter:I think in general being overcapitalized is a path to failure. The VCs
want you to spend. There are general ills with being overfunded.
I don't think they ever really quite thought out the problem. We live in a different
world now where people value the data differently.
Livingston: Was there anything about del.icio.us that was much better than
your competitors?
Schachter: I think the competitors had already disappeared by then. The tagging
thing was probably essential.
Livingston: Can you tell me more about how you came up with tagging?
Schachter: There was no point at which I said, "I'm inventing this wonderful
new thing." I just sort of realized that I had evolved my own filing system, and
it worked for me. I'd used it for a long time before del.icio.us even showed up.
This was the codification of that practice.
Livingston: But you were one of the first companies to do tagging?
Schachter: Yeah. For example, in Muxway, the internal table that tracked that
stuff was called Tags. The name had come along at some point, but I don't
remember exactly how it showed up.
Livingston: When you decided to leave Morgan Stanley and focus full-time on
del.icio.us, did you know you had to raise money?
Schachter:I was getting a lot of interest in acquisitions--there were a bunch of
offers/buyouts, and they were increasing in value over time. At the same time, I
wanted to be able to pay the rent, but I didn't want to chew into life savings. At
the end of the day, my Morgan coworkers were pretty supportive, "You should
go do this. Try it out and let us know how it goes."
Joshua Schachter 225
Livingston: Union Square Ventures was your VC, right?
Schachter: They were Union Square and Amazon.
Livingston: Did they come to you?
Schachter:I had met Jeff Bezos at Foo Camp, and he was very interested.
Livingston: How much did they put in?
Schachter: We never announced the amount, but it was not a huge amount of
capital.
Livingston: And that's because you didn't want to take a huge amount of
capital?
Schachter: Well, there was a lot of risk. It was sort of hard to justify a large
valuation and so on, so we sold a small chunk for enough money to work for a
while and see if it turned into something. That was the plan: see where this
goes.
Livingston: Did you hire anyone?
Schachter: We did. There were eight employees total at the end.
Livingston: Were most of them shareholders?
Schachter: We gave shares to everybody.
Livingston: Did you have vesting?
Schachter: Yes. Even I vested.
Livingston: What were some of the first things that you did once you were officially
a startup?
Schachter:One of the most challenging things was getting payroll going. PEOs
typically don't want to do less than five employees.
Union Square introduced me to this guy, Albert Wenger, who had some
operations experience. He helped a lot. I lucked out in that he's a smart guy
who knew how to do not just the corporate operations stuff, but he had a good
product sense and ended up doing a great deal of product work as well. The
first version of the Firefox toolbar, he dealt with, for example.
Livingston: What were some of the biggest technical problems that you
encountered?
Schachter: Scaling, inevitably. Scaling, dealing with bandwidth, dealing with
routing, networks. This is for consumer Internet kind of stuff, but there's a
great deal of stuff that you have to flawlessly execute on. It has to be done well,
but everybody does it well, so it doesn't differentiate at all. Like your connection
has to be up. Your office needs to have DSL. There's a great deal of crap
that has to be executed better than competently that is no value for you to actually
do yourself. So outsource that.
For example, the payroll. I was capable of going 2 to 3 months without
salary, but other employees certainly were not. So that kind of stuff.
226 Founders at Work
But you need to pay attention to the important stuff. Scaling was important
and core to the product, but dealing with the network, getting the hardware
racked, building machines, ordering stuff, getting pricing out of Dell, you name
it. That was a lot of work that was not useful.
Livingston: Outside of the scaling requirements, can you remember any
technical problems that you guys solved?
Schachter: Tagging basically was the thing. And then there's a gagillion little
improvements in marketing things. We actually thought about the product
always with an eye toward innovation. Everything we did we questioned--and I
think we didn't even go far enough. Whatever it is, question every single aspect
of conventional wisdom. "Is that the right way to do it or can we break that and
make it better?"
That's also dangerous, because, if you are doing a lot of paradigm innovation,
call it--which is not a good word--but if you are breaking boundaries
elsewhere, maybe you need to be very within boundaries on other fronts.
Livingston: Do you remember a time when you were worried about something?
Schachter: Site's down. Site's slow. Table crash--MySQL corrupted a table.
That happened all the time. A great deal of what we did was putting out fires.
We didn't have a lot of process management in place, which probably hurt us a
great deal.
A week after the acquisition, the power of the data center dropped and corrupted
every single machine. We were down for like 48 hours. That was horrific.
The power bounced in the network; the machines didn't come back up
because they weren't configured quite properly. We weren't careful about that.
In general, assume that whatever you are doing is going to go wrong. How
can you make it so that it will go faster when it does go wrong? Because it will.
For example, the rebuild script takes 24 hours, but that's not a big deal because
this part of the system isn't live yet. But when it is live and it takes 24 hours to
redo, that's a big deal. So fix it. Make it work in 2 or whatever. There's a lot of
stuff that you can't get around due to SQL, like you can't change the database
without bringing the site down.
Livingston: What kind of technology inspired you?
Schachter:Inspired? We built in Perl, MySQL, Apache. Very standard LAMP
stack kind of stuff. That was the standard mechanism for everything.
Livingston: Would you do anything differently if you could?
Schachter: Knowing what I know now, I would have designed the back-end
architecture differently, and that would have saved a lot of work now. Scaling
past one machine, one database, is very challenging, even with replication. The
tools that are there are not quite right.
For example, when you add things to a table and it numbers them, that
means you can't have a second machine also adding to them because the numbers
will collide. So what do you do? You have to come up with some
Joshua Schachter 227
completely different way to do it. Do you have a central server that hands out
number sets, or do you come up with something that's not numbers? Do you
use random numbers and hope they never collide? Whatever it is, autoassigned
IDs just don't fly. There's a stack of about 15 things that I have, a big
list of pitfalls.
Livingston: Can you remember any features from del.icio.us that the users
wanted or really loved that surprised you?
Schachter: There's always stuff. I tend to be careful about that. I think people
ask for features--they want to do something, but they don't say, "I want to do
that something." They translate it into some feature that typically they've seen
somewhere else and ask for that instead. I want a feature that does this. "Why
do you want to do that?" Then it turns out there's some better way to do that.
So, stuff that people ask for, I tend to try and dig to the root cause, before
reducing to practice.
People frequently aren't quite sure what they want. Then there's a whole
bunch of stuff that's like, "Feature 1 and feature 2 suggest feature 3; ask for feature
3." And I just know that people are never going to use feature 3 and the
implementation thereof would be quite expensive. So leave it out.
Livingston: How did your user base evolve over the years?
Schachter: I think it's still a very technical, early adopter audience. It's broadening
over time, but we're sticking with that for now.
Livingston: Can you tell me about some of the major turning points in
del.icio.us?
Schachter: Nothing really comes to mind. It was like a roller coaster always
going up, so it's always increasingly bigger, faster, more and more people.
I had a bunch of conceptual revelations on how to build stuff. For a long
time, it would go slow and I'd figure out some clever thing to do--"I know
we're doing extra work here." Figuring out caching. My own education was
kind of interesting. But that was ongoing; there was always something new that
I learned every couple weeks. So I never really broke it up into large milestones.
Getting the funding, working on it full-time, selling it--these were all
big parts of it.
Livingston: How was working on it full-time different than when you were at
Morgan Stanley?
Schachter: Constraints breed creativity. So now, instead of only having 15 minutes
two or three times a week, it would be more like, "I have the entire day to
work on it, every day." I don't work in bursts like that. I do a little bit of work
and then go wander around the city and come back. Then work all night. Once
everyone has gone to sleep and it's quiet, I can get a lot of work done. I didn't
really get to stay up late when I was at Morgan; I don't really do it now. But during
that I did, and I think it was incredibly productive. Probably very alienating
to my wife though.
Livingston: Did you find you were better at some things than you thought?
228 Founders at Work
Schachter:I could focus on it more and do slightly larger stuff. I've always had
a short attention span, so that's probably the actual limiting factor. The amount
of coffee I can consume to mitigate that and that's about it.
Livingston: Were there things about del.icio.us that users misunderstood?
Schachter: We named things differently. I wouldn't say that we had awesome
execution. It was very techy. It bred a strong priesthood, which was helpful in
getting the message out initially, but it was harder for people to adopt. We continue
to work on that, and struggle with that now.
It is a challenging product to do conceptually. It's not something like, "Let
you file your taxes better." There's no clear value proposition here. It is valuable,
but hard to understand. You will be able to remember more things this
way, and with that, people don't even realize there's a problem. So that's a challenging
value proposition to explain or get across.
Ultimately, I think people who understand it are better for it, but it's a
challenge.
Livingston: Was there anything that you learned from your earlier projects that
you were determined not to do with del.icio.us?
Schachter: There were a bunch of things. I released a bunch of projects--I've
done a bunch more that are halfway done. I keep an idea journal of stuff. I
make ideas and I work on them a bit to see what they feel like, and then I move
along. One was called Bookbook--because I never came up with a name for
it--in which you could say, "I'm at this location and I don't want these books
and I do want these other books." You would put that in an XML file on your
website, like a feed--you would provide a feed and other people would do this
and create a central crossing engine that would say, "You have this book and he
wants that book, and you are not that far from each other." This was basically a
distributed geomarket for books.
The problem was, the way I wrote it was fully decentralized. You didn't log
in and create your data; it was just, "Here's a URL to my data" and the system
would do the best it could. The problem is that it was so hard to use. You had to
make an XML file. If that's your beginning user interface proposition, you fail. I
think 12 people signed up for it, maybe. The UI was too hard. The elegance of
a distributed system trumps the usefulness of centralized UI and control.
Similarly, there was a system called Loaf that I did with Maciej Ceglowski
that was a fully distributed social network--no central server whatsoever. It
used email as a carrier and could tell people you talked to about other people
you corresponded with in an encrypted and compressed way. If I emailed you,
it would attach a Loaf file. You couldn't open the Loaf file and read the contents
of it; it just didn't work that way. It used Bloom filter, so it was sort of a statistical
object. But you could take another email address and see if it was in there.
With 99 percent accuracy, you could tell if someone was inside that file. So if
you got email, you could say, "I think Joshua Schachter corresponds with this
person." Without me exposing my address book to you, you could tell who in
your address book you talked to. It was a pretty neat idea, but it was complicated
to install.
Joshua Schachter 229
The other problem was that it didn't work without Loaf. So that didn't do
very well, but we got press for it. It was sufficiently innovative. Maybe I'll
return to that idea someday.
Livingston: So press helped you get the word out about your projects?
Schachter:I was in USA Today in the late '90s for Memepool, so it was always
from there I got a great deal of press and sort of had early training. My father
was the consumer advocate for the Long Island Railroad and was in the newspaper
all the time, so I got training sort of that way. When this stuff started happening,
I knew that you have x messages and when you talk to the press, any
question they ask is answered with one of the messages.
I understand talking to the press as an essential part of marketing. At the
same time, I understand that the consumers are the best marketers. If they love
your product and you give them the tools to market it, they will.
Livingston: What do you think about technical founders versus businesspeople
founders?
Schachter:I have never had a great deal of trust for people who don't execute
on core ideas. I understand the value of needing someone to deal with that kind
of stuff--someone's got to do the VC pitch and there's got to be a CFO, etc. But
the guy who says, "I have a great idea and I'm looking for other people to implement
it," I'm wary of--frequently because I think the process of idea-making
relies on executing and failing or succeeding at the ideas, so that you can actually
become better at coming up with ideas. It's something you can learn. It's a
skill, like weightlifting. That failed; that worked; continue. You begin to learn
how to make ideas. So if you are someone who can't execute and all you can do
is come up with ideas, how do you know if they are any good? You don't really
know if it's a good idea until you've executed it. You need to understand the cost
of execution and so on.
Also, where I worked at Morgan, they were not hyper-trustful of MBAs. All
my coworkers were PhDs in computer science, mathematics, or physics.
Livingston: New York City doesn't seem to be a place where too many startups
flourish.
Schachter: There's a great deal of technology going on in New York City. In
financial places, there is lots of high-end, high-speed transactional technology.
There are a lot of good problems in finance. The technical problems faced
there are hard. One issue is that there is a lot of money to be made there and
the companies that are doing that pay. Some of the big brokerages pay a lot of
money annually for their technology.
Livingston: Why aren't there more good hackers living in New York City then?
Schachter: There are. They work at banks and stuff and have side projects.
Livingston: So the smart hackers...
Schachter: They're around. They just know when to be quiet and when not
to be.
230 Founders at Work
Livingston: Who did you learn things from?
Schachter: Albert, the guy I mentioned. I learned a lot from Fred Wilson of
Union Square, certainly. I learned a great deal from my Morgan folks. I learned
a hell of a lot about how to understand problems. It was rough sometimes, but
they pushed me far. My coworkers at Morgan were smart people. One of the
people we worked with won the Nobel Prize for economics while I was there. It
was a fast, smart environment.
I remember when I interviewed at Google the first time around and they
were making derogatory comments about where I worked: "Well, here you'll
get to work with PhDs and computer scientists." And I'm like, "I already do."
Livingston: So it didn't work out when you interviewed for a job at Google?
Schachter: I went out there once and was rejected because I didn't know C++.
Livingston: Was there ever a time when anyone tried to trick you or take
advantage of you?
Schachter: Yes, but it would not be polite to talk about it. There were several
cases of people wanting to get equity in advance of other people, or weird deals.
Livingston: As a new startup founder who has never done a deal or negotiated,
do you think you need to be careful of getting taken advantage of?
Schachter: In general, I found VCs to be significantly politer than the folks I
worked with. The worst they did was not call me back. I'd never hear from
them again. Brad Feld does a nice blog talking about how the VC process
works. He says they never call you back to say no--they don't want to close the
door in case they want to open it again, but they don't want to actually give you
a response. Very few VCs actually said, "Sorry, we're not interested."
Livingston: How did the process of starting your own company and then selling
it change you?
Schachter: It pushes you far. You learn a lot. I did a round of funding, I was
writing code, I was hiring people, chief architect designer, negotiator, you name
it. I did all of it, for the most part.
When Albert got up to speed and was working full-time, he did a great deal
of work there as well.
Livingston: What surprised you most about having your own startup?
Schachter: It's a combination of sudden freedom to run things as you please
and crushing responsibility in which you know you have to do certain things in a
certain way at a certain time. That eradicates all of that freedom. You become
a robot on rails. You know what you have to do and you are working in a certain
direction.
Maybe other people are different, but I think that every step was sort of the
inevitable, inexorable progress due to the previous steps in the path. It's not like
I had no choice, but everything I did was the only choice because it was the only
thing that made sense at the time. It's not that long ago, but no regrets, that was
the path to take. Everything that had to be done was done.
Joshua Schachter 231
Livingston: What is your favorite bit of advice you'd give to a technical person
who wanted to start a startup?
Schachter: Reduce. Do as little as possible to get what you have to get done.
Do less of it; get it done. If you've got two things that you want to put together,
take away until they go together. Don't add another thing. Because you can
understand it better, you can analyze it more cleanly. The UI will be easier.
Doing less is so important.
People often wind up adding features, adding stuff. Making it bigger is the
typical way you engineer out of a problem, right? It's the traditional, "I apologize
for the long letter. I didn't have time to make it shorter."