hosted by aaron harris
episode | transcript - n/a
- started loopt at 19; it grew out of a side project and morphed into a startup only because of yc
- dormmate blake ross told him about pg's summer founders program the day applications were due
- in 2005, he know who pg was but was't "following startups"
- facebook had started the year before but no one considered them a real company yet
- loopt was born from mobile and gps growing to be important things, tied together to see the friends around you and stuff going on around you, that'd be powerful
- the summer before he worked in the stanford cs research lab on an autonomous helicopter navigation project using gps (before thinking about gps was a common thing)
- and when a palm treo was the best "smartphone" at the time
- tech trend - discontinuous jumps to new computing platforms, and between those it just gets smaller, thinner, lighter, faster
- from 2005-2008 getting into yc counted against you, it just meant you were working on this little thing
- one mistake people make is telling themselves they can't do something and cut out a huge option value in their lives... not wanting to get turned down is one of the hardest life skills to learn
- good entrepreneurs are often particularly sensitive to rejection / care too much about what others think, especially first-time founders are extremely insecure
- to get around that drive for external validation - it's a deeply personal thing... "but at some point people make a decision that you know what i am going to trust myself to do the right thing; i'm going to decide that i believe in this business, and that the haters are going to be wrong, and i'm gonna get this done."
- "if you're doing something interesting, the chances that some internet commenter will say that it's stupid or not going to work is 100% and the chances that some investor is going to tell you that it's dumb is 100%"
- when reading haters about yc or sam personally, he tells himself "you know what, i'd rather be me than the people writing this and you move on"
- dealing with both:
- personal ad hominem attacks
- business challenges - no one knows about your product and you have to push until they start discovering it
- when ben silverman was working on early versions of pinterest, he walked around cafes in palo alto and asked people to try his product; he'd go in the apple store and set all the browsers to pinterest.com before employees saw so it might seem like some official thing and maybe some people would really fall in love with the product
- "the opposite of love for a startup is not hate, it is indifference, and what kills startups that they never worry about are not any of the big scary monsters like competitors, but it's people coming to their product, their page, their site, and clicking the back button and never thinking about it again."
- "founders understand this intuitively so they don't want to launch this thing that people will say, 'well is that it?... so what?'" --> founders wait too long
- ben's example is the archetype for non-scalable user acquisition or at least discovery so you can learn what people are actually going to do
- the big press launch rarely works and it's not what they advise at yc "it is absolutely the wrong strategy to pursue"
- on yc's unicorns: "none of them launched with a crazy trumpeting victory parade, they just sort of post and say hey check this out please, they get a few friends, and those friends get a few friends"
- "the issue is that if the ceo is focused on the press and not getting a few initial users to love the product, that's what the rest of the company focuses on too"
- "i don't pretend to know everything that went wrong at a clinkle or a color, but those are common examples of these gigantic press launches with huge amounts of money raised and no one could use the product until they were ready and all that. and i think watching that as an outsider, it did seem in those cases that the press really hurt the companies."
- a few weeks ago they talked about tech being applied into legacy industries to transform the way they work... is it really infiltrating more industries or are we just seeing that in a couple places that people are talking about a lot?
- "it's hard to see an exponential curve when you're on one -- if you look back it looks flat, and if you look forward it looks vertical -- but i think it's just the continuing compounding curve"
- there's a misconception that yc invests just in consumer internet tech
- how do you decide which industries and founders to fund?
- "we want to fund every $10B+ dollar company that we can and that is such a hard constraint that we cannot limit ourselves just to consumer software, enterprise software, whatever. that is what other investors do and i don't have any interest in us doing what other investors do. i think that is a bad strategy."
- "we'll continue to look for what industries have changed to now support startups: where have the cost and iteration speed come down enough that a startup can win in a large market?"
- "the common wisdom, as late as 2007/2008, was that only pure software companies were the ones worth investing in, and that if you touched the physical world at all, you weren't a technology startup because you were going to have problems of humans and regulations and all those sorts of things. and so there was this big drive to invest only in a pure software company."
- many of the biggest startups of the last 10 years are "software plus":
- uber = software + cars
- airbnb = software + houses
- instacart = software + groceries
- people have not always loved online to offline
- aaron adds that ups and fedex has succeeded because of the way they apply technology to logistics
- in human technological breakthroughs, one of the longest reigning of all time is the hand axe back to before we even called our ancestors humans
- that's happened a few times: agricultural revolution, industrial revolution, and each is bigger than the one before; now we're in the middle of this particularly big one with the software revolution
- "the software revolution is still probably underestimated and will be bigger when the history books are written in a few hundred years, assuming that the ai doesn't kill us before then"
- founder of ridecell: on-demand access to shared ride services with transit agencies
- farebox recovery ratio: how much money a transit system makes for every dollar they invest
- large transportation systems are inefficient; for many U.S. systems FRR is < 20%;
- reasons:
- poor tools - lots done manually vs. automation; 30-year-old software; not realtime
- poor data - don't know where things aren't working, where/why people aren't using transit (esp. realtime)
- misallocation of supply/demand - ex. "in LA there is a bus that runs once an hour, but costs $125/hr to run, no one rides it because it runs once an hour, but because no one rides it they can't run it more frequently"
- low frequency of transit makes it hard for people to get to trains/subways in many cities if they live more than a mile from the train station because it's easier to drive to work
- ridecell is an operating system for transit agencies to find where the problems are, automate manual tasks, and augment/replace empty bus routes with more flexible on-demand shared ride services
- how did you end up doing this?
- became a programmer from the satisfaction of creating something and being able to see it work instantly
- after learning to drive in mumbai, failed the atlanta driving test several times and has to take the transit system
- how did you figure out where to start?
- found a CS professor at georgia tech that had worked on transit for a while. he has this grand notion of how you fix the integrated transit problem, and so he started attacking the large problem, which is a bad idea.
- "the right approach is to find the smallest bit of the problem that you can solve today"
- i started wrong by launching a consumer transportation service that competed with uber
- but then came back to the transit agencies with what we learned:
- figuring out the root cause and attacking it - focus on matching supply and demand -- the hard part about running uber is the backend -- a lot of programmers like building transit tools, but they're mostly focused on the front end
- speed - "the only thing that matters is how quickly you iterate". a lot of things any successful startup does are not perfect when they launch, then they quickly improve it.
- how does this change the conversation with transit agencies?
- labor consumes the most resources in transit (people operating a bus all the time) vs. a consumer transportation service having exactly the right number of drivers proportional to demand and they only pay for rides that are delivered -- how do you move a transit agency closer to that end of the spectrum? it's a very different problem than uber: a different set of needs for different consumers in a different environment.
- figuring out which parts of the system should be fixed schedule vs. which should be dynamic per demand
- when you went back to the transit agencies did you say that you could solve all of their problems in one swoop or focus on giving them one thing first? the latter -- they wanted apps, but only so much could be done with the underlying inefficient system, so we went back to solve the core problem: efficiency & cost (thereby improving user experience and apps)
- some customers have reduced their costs by 50% for existing on-demand by switching to ridecell
- the bigger longterm problem we're going after is empty buses to increase ridership
- "it's important i think as a community for any city, any community, that wants to increase entrepreneurship to not be critical of ideas that look like toys or that look like they're meant for only a certain part of the audience because when that technology is able to expand to its true scope, it will benefit everyone"
- for example, aws in the early days benefited a small group of devs, but today even government infra runs on it, and it's much cheaper than it used to be
- was it hard to leave the consumer business and go back to transit agencies?
- yes, because you're at this interesting place where you have to decide between giving up on something you've invested years of your life in vs. jumping into something completely unknown. some investors told him don't worry about losing my money, just stop, start your next startup and we'll fund it -- it's hard to tell them that you want to try something else when they give you that advice.
- a huge part of success in startups is always luck
- "in the face of everyone telling you 'no, this is not a good idea', you're still doing what you think is worth pursuing"
- how did you know you needed to switch from consumer?
- "we came to transportation from a missionary place vs. a mercenary place. when people talk about uber and their competitors, the view is there's this goldmine and a bunch of people are just rushing to build the next uber for x, but we've been doing this before uber started; we actually started doing realtime ridesharing in atlanta in 2009, and so it was harder for us than it should have to been to realize our product was not working."
- vinod khosla told him "don't work on something unless you can have a meaningful impact; i don't care if i lose my money -- go start something else."
- the logical thought process was that we wouldn't be able to make the kind of impact we want running a consumer transportation service because uber and lyft were already doing a great job, so what else can we do that will having a meaningful impact on the world, be satisfying to us, and be economically viable as a business? this realization unrolled over several months and we weren't confident about it even until we had our first customer.
- AH: "it's so easy to look back on things and fit them into an easy narrative, but it's the figuring out of those things in the moment that actually defines who makes the right decisions and who succeeds. and i think your point about luck is huge, you happen to know about these other things, but i think it's more than luck. it's being ready for those moments when they happen and having the right information."
- so what's the challenge now?
- 4 of the top 10 transit agencies in the u.s. are starting pilots with us, and we have pilots running. we understand this problem better than anyone else in the world, so now the challenge is just bandwidth.