tag:blog.garrytan.com,2013:/posts garry's posthaven 2017-07-25T09:42:19Z Garry Tan tag:blog.garrytan.com,2013:Post/1092890 2016-10-04T19:49:14Z 2017-07-25T09:42:19Z Consumer, SaaS, or Enterprise?

Early founders often ask whether the classifications of startups (consumer, SaaS, or Enterprise) matter a ton in terms of what kind of startup they should try to start. Is there a such thing as "founder market fit" when it comes to these stages? Unless you're an ex-Oracle sales manager founding a company, probably not. 

It's backwards to rely on the consumer/SaaS/Enterprise distinction early on because that's tactics, only after you've picked a problem to solve. In a nutshell:

  • Consumer — lots of people (tens of millions or more) have the problem, and the problem is medium to severe
  • SaaS — fewer people (hundreds of thousands or more)  have the problem, but the problem is severe
  • Enterprise — a few people (as few as dozens) have the problem, but the problem is extremely severe

Start with what problem to solve, and then choose your tactical approach given what you know about that problem. That'll help you get to where you want to be, and let you optimize for solving a real thing that you actually want to do. 

Finding good startup ideas is hard enough. You don't want to complicate it by setting constraints at the wrong level of a if you can help it.

]]>
Garry Tan
tag:blog.garrytan.com,2013:Post/1049735 2016-05-12T01:00:00Z 2017-07-09T13:18:28Z What to do when you just start making something useful

A founder in China I met asked me what he should do with a business he started that has a few hundred users that are happy. He's not sure what to do next yet.

Here's what I wrote him:

The main thing is to a) figure out how you can make money, and then b) figure out how you can grow. How much you know a and b will determine whether or not you decide to raise money, or do it as a lifestyle business, or don't do it at all. It's important to figure that out as soon as you can, since those are really the only next steps from where you are. If you can grow and make a lot of money (high unit economics / margin) then you should raise money. If you can grow but can make some but not a lot, then you can probably be a good lifestyle business and you should just try to get to profitability. If you can't grow or you can't make money doing what you're doing, then don't do it.

I think the most remarkable thing about this trip to China is while the markets are radically different, the fundamental problems founders face are absolutely the same. 

]]>
Garry Tan
tag:blog.garrytan.com,2013:Post/971235 2016-01-15T17:00:05Z 2016-11-16T06:11:38Z Babbage's broken dreams: A brilliant design for the world's first mechanical computer, sabotaged by personality

Brilliant inventor Charles Babbage designed the Difference Engine, a pioneering mechanical computer capable of storing and operating upon up to 1,000 numbers with a limit of 50 digits each. Remarkably, it wasn't until the modern era when researchers used 19th-century mechanical fabrication techniques to verify Charles Babbage's designs did work after all. 

A lecture in 2008 at the Computer History Museum by Doron Swade shed light on what went wrong

In fact, no one at that time had warned that these calculating machines could not be built. Many other machines - some of them useless - were in fact built using those same Victorian era mechanical parts and tools. Instead, there were a myriad other factors contributing to the failure to physically realize the Difference Engines. These included: an argument with his lead engineer over worker compensation, run-away costs, muddled financial arrangements, wrangling with a changing British government for funding, and a resulting discontinuity of negotiations over funding and construction.

Perhaps, Babbage's personality had more to do with the failure to complete the Difference Engines then the underlying technology of the time. Known by his first biographer as the "irascible genius" Babbage had lots of pride and hubris. He behaved as though being right entitled him to be rude. His quarrels with the British government and with chief engineer Joseph Clement might have doomed the project.

When Clement resigned in March 1833, the practical construction of Difference Engine No. 1 stopped - 11 years after it was conceived. Negotiations with the British government continued, but funding was finally axed in 1842. Difference Engine No. 2 was conceived and designed between 1846 and 1849 - long after the design of the (never completed) Analytical Engine.

Regrettably, none of these three machines were built in their entirety in Babbage's lifetime. Only years later did we recognize the potential and power of the underlying machine architectures, which were independently "discovered" by later day computer pioneers.

A preventable tragedy. It was lack of interpersonal skills leading to inability to raise funds (and inability to manage execution!) that kept such a vitally important innovation from surfacing. Not the first time, and certainly not the last.

Irascible geniuses everywhere must take heed as they pursue their dreams. Being smart, right, and brilliant is no license to be rude and treat people poorly. So many startups crash and burn, not because the tech didn't work or the founders weren't smart enough, but because the interpersonal neglected. 
]]>
Garry Tan
tag:blog.garrytan.com,2013:Post/926338 2015-11-06T17:07:56Z 2016-10-26T11:22:40Z Thank you, Y Combinator

Dear friends-

I wanted to let you know I am stepping away from YC. I've been involved at YC since early 2011 and in that time it has been my privilege to work with over 600 startups and over a thousand founders. It would be impossible for me to have predicted the success the team has achieved since then. After nearly five years, I’m ready to take a break and plan for my next adventure.

It's been a great time and I wanted to thank everyone for being awesome. PG, Jessica, Trevor, and RTM have built an enduring organization and I can't wait to see where my friends and partners take it. Led by Sam, the YC partnership has taken the founding vision and put it on track for tremendous success and impact. Between the incubator, YC Research and YC Fellowship, the future is bright indeed. YC is engraved in my heart. To my fellow alums: I will always be there for you.

My wife and I are headed to Southern France and Spain for a few months to relax and enjoy the wonders of parenthood with our 3 month old infant. I don't know what is next yet, but making great software and helping others do the same will always be my life's purpose. See you all back in SF in the new year.

With much love,
-Garry]]>
Garry Tan
tag:blog.garrytan.com,2013:Post/876251 2015-07-01T22:28:17Z 2016-04-05T15:49:04Z Deprogramming corporatism

There’s a culture to big corporations that is unnatural and detrimental to founders who have spent too much time in them. Some experienced founders have a really hard time coping with starting a new company, and are very frustrated and surprised when things don’t work even when they're working really hard at it. And even when founders overcome that, often they’ll try to grow the team by hiring experienced directors and managers from corporations that are well known and successful. Many of them fail pretty quickly, and it’s always a surprise to everyone, especially the experienced hire. 

These are well-known startup tropes for a reason. There are specific aspects to that culture that cause problems. If you know what they are, then you can at least recognize it and try to counter them.

Doing things without results

People become corporate do-nothings because it’s easy to become disconnected from the act of creating when you’re at a big company. Even at the best big companies, employees can just do things that look like work that aren’t, and you wouldn’t be able to tell. Getting customers is easier (sometimes trivial) with a large installed base, sales force, or huge brand name. There are large protected revenue streams that cover up failure on the quarterly financials. 

Founders from corporate backgrounds aren't stupid. They're just used to doing things and having something happen— they have the wind at their backs at a big company, and there are a lots of things that you can do at a big co that would never work on your own. It's bewildering to work really freaking hard at something and have no results at all come back. But it happens all the time. 

There’s no air cover for startups. You don’t have backup troops. It’s just you versus the world. So naturally, if your product or service sucks, then you die. You can’t just look like you’re doing stuff. You actually have to make it, and almost totally on your own. 

Cover your ass culture

If doing things that aren’t effective don’t get you fired, then what does? Usually making mistakes that make your boss look bad. Big organizations are just groups of people, and people sure like to talk shit. The one thing you can’t do is look like a bozo. It’s fine to work really hard to no effect (hey, you worked hard!), but if you become a social liability, you’re donezo.

What’s worse than just covering your ass is actually taking credit for what other people do. This seems correlated with people who climb high in organizations. Obviously this works poorly in a small startup environment because there’s nowhere to hide. Someone's got to do the work.

In some sense this dynamic is unavoidable, since we are all social animals after all. Startups can sometimes develop a cover-your-ass culture too, but that’s the job of the founders and CEO to keep people focused on things that actually matter. It helps that startups are just smaller, so this toxic effect of group dynamic is blunted. 

Buzzword thinking and trend-following

How do you avoid looking like a bozo? Well, for one thing, if everyone in the world out there is saying it, then nobody can fault you for it. So getting the right corporate whitepapers, or latching yourself to the right buzzword du-jour (e.g. big data, Internet of Things, NoSQL, etc.) is necessary to blend into the pack. Oh, that big data initiative failed? Everyone else was doing it, so our ass is covered and we won’t get fired. 

Founders get this confused all the time and then wonder why they fail. We said all the right secret words! Why am I failing? It was never about those words to begin with.

Startups can’t survive blindly following buzzwords or whatever trend is hot because you actually have to know what’s coming in the future and be right. That’s all there is. If you chose the wrong market, or you’re wrong about what people want, then you’re toast. I’m not saying all things with buzzword labels will fail. I’m saying that startups for big data, for instance, actually have to make life better for specific customers such that people are willing to pay for it. It has to make sense. It’s not enough to be attached to that name. 

This was a tough lesson for me to learn personally. At 23, I turned down the shot to be first engineer at Palantir (now rumored to be worth $20 billion!) even though Peter Thiel personally took me out to dinner to recruit me. I thought the buzzwords were signal, and absolutely zero of the mainstream press or tech blogs were abuzz about the latest hot government enterprise software startup in 2004. It turns out you have to work on things that a) you know are right, and b) most people don’t know yet. This is why Peter likes to ask: What super valuable company is nobody building yet?

Deprogramming

Experienced founders who have grown up in these environments are not doomed to failure. Quite the contrary, those who succeed have avoided, overcome, or escaped the problems described above. 

The recurring theme seems to be simply results. We spend a lot of time trying to get founders to focus on action and results — build product, talk to customers, that’s it. Think in terms of concrete numbers, whether it is user growth, savings to customer, or revenue. There are lots of places in the world where you can survive without results of your own, but startups are not one of them. 

Just as corporate culture is a culture that is learned, not innate— founder culture is learned as well. I think one of the reason why YC works for founders is that it takes a village. It takes a bunch of people who all believe a thing, and practice it daily. It takes fundamentally changing your surroundings and the people you’re around. It takes avoiding the coworking space [0], and working harder than you ever have in your own space. 

This is also a big reason why they say YC is a concentrated form of Silicon Valley. For decades, Silicon Valley has been the place where people can escape their corporate cultures and create something new. Now you can do it with a lot more like-minded people by your side. 

[0] http://www.quora.com/Why-doesnt-YCombinator-have-a-physical-space-for-its-startups-to-work-out-of

]]>
Garry Tan
tag:blog.garrytan.com,2013:Post/859340 2015-05-21T17:14:38Z 2015-09-24T19:34:35Z InboxSDK launches — The Biggest, Most Unexploited API Frontier in 2015 is now open

The Facebook API. iOS. I remember when their API docs hit the scene, and what a greenfield opportunity it was. Today, the opportunity for Gmail apps is much bigger than when those greenfield APIs were released. There are over 1 billion daily active uniques on Gmail now, and it’s a fundamental pillar of how people communicate today. But until today, it was cumbersome and nearly impossible to consistently write an app that had a good user experience that worked as well as being a part of Gmail itself. 

InboxSDK, released today by YC-backed startup Streak CRM, is that solution. It provides the missing things that would be extremely cumbersome to do with Gmail’s backend APIs alone. Lots of companies like Dropbox, Stripe, Screenleap, and Streak CRM themselves actually use InboxSDK right now in production. 

An incredible amount of work was put in to make InboxSDK possible. A Gmail API is a monumental task — just look at all the possible interactions, with multiple inboxes, chat, preview panes, conversations, different Gmail Labs features, and many kinds of email compose windows. 

As with any brand new opportunity, we don't know all the amazing things people will build yet. But this is a rare opportunity for hackers and founders to think about ways to make awesome stuff that works on top of email. A billion users are going to be in for a treat this year.  

]]>
Garry Tan
tag:blog.garrytan.com,2013:Post/853016 2015-05-07T22:38:55Z 2015-11-06T21:28:44Z Share buybacks: Big cos say "We don't know what to do with the cash anyway!" and why it's good for startups.

Stock buybacks are the biggest force influencing equities since 2009 — over $2 trillion have been done.

Some are probably good, but many aren't. In 2011, Warren Buffet wrote about how you can tell: (via Seeking Alpha blogger Chuck Carnevale)

Charlie and I favor repurchases when two conditions are met: first, a company has ample funds to take care of the operational and liquidity needs of its business; second, its stock is selling at a material discount to the company's intrinsic business value, conservatively calculated.

Clearly a lot of buybacks fail the 2nd criteria — but there are so many boards and management teams that are compensated by EPS that the share buybacks keep happening even when the price is not a discount. A natural result when incentive structures meet personal self-interest. One blogger points out, it's actually a form of managers looting their companies. 
By using large stock buybacks to manage the short-term objectives that trigger higher compensation for themselves. By using those stock buybacks to manipulate the share price, which allows them to use inside information to time their own stock sales. By using buybacks to funnel most of the company’s profits back to shareholders (including themselves).  They use the stock market to loot their companies.
It's a claim echoed by the Atlantic in February of this year in an article called "Stock Buybacks Are Killing the American Economy." Rather than new jobs, new factories, new products, it's straight back into the financial system.

Since the companies themselves have no specific better ideas about how to use the capital to grow in real terms (e.g. real new products that drive real new revenue) all their professional managers can do is buy shares back regardless of price. In this world, buybacks are directly related to Thiel's ideas around indeterminate optimism of the markets, where companies are encouraged to be as profitable as possible.

Who cares if you're buying back at a ridiculous price? We don't know what to do with the cash anyway. 

This is interesting because a traditional criticism of whether we're in a bubble is whether P/E ratios are high or low. The E part is Earnings Per Share, and so if managers manipulate the denominator for both P/E and EPS, they can make it seem like things are fine (and get their quarterly bonus to boot!) when functionally there's little happening. 

On the bright side, the lack of innovation in traditional incumbent businesses means that startups have a chance in more arenas than ever. If big companies have management that won't take the risk of failure, and aren't hiring teams to attack new markets, then it's wide open for new players to get capital, hire people, and make these things happen. 

If buybacks are killing the economy, then startups will save it. Pretty sweet if you ask me. 

Chart via Factset
More reading here: ValueWalk 
]]>
Garry Tan
tag:blog.garrytan.com,2013:Post/850225 2015-05-01T21:41:13Z 2015-11-08T23:03:46Z Defending A16z: Noam Scheiber mistakes a VC portfolio for his 401K

Labor and workforce reporter for the NY Times Noam Scheiber takes on Andreessen Horowitz in this recent piece:

Rather than profiting like Mr. Ovitz and his fellow agents, the venture capitalists may be more like the Hollywood studios — chronically overpaying for projects whose costs they can rarely recoup. Mr. Andreessen and his partners have invested so much in so many start-ups that it would take a remarkable string of successes to make the approach pay off. For all their skill — the firm bought into the likes of Airbnb, Instagram and Pinterest relatively early — their track record suggests it’s unlikely. Already, they’ve suffered a few impressive flameouts, including Fab, on which they are likely to lose tens of millions of dollars.

Even when they pick well, they often bid so much for stars that the return is relatively modest. It’s easier to triple or quadruple your money when you’ve invested $10 million in a $100 million company than when you’ve invested nearly $100 million in a $1 billion company, as they did with the daily deal site Zulily. There are only so many companies that are acquired for billions of dollars or reach that kind of price through an initial public offering. Fewer retain such valuations — Zulily’s stock price has fallen sharply since last year.

Noam doesn't make a data-based argument. He uses an anecdote. Because data would ruin a really good story in this case. He compares Silicon Valley juggernaut Andreessen Horowitz to the excesses of Hollywood and CAA.

There's a big difference here. Let's take Avatar, for instance. It's a film that was made in 2009 for $237 million. It grossed over $2.7 billion worldwide — a roughly 11X return on capital. That's the highest grossing film ever made, and a good proxy for how profitable Hollywood can be at best. 

Let's take another example from Silicon Valley— Facebook. Peter Thiel invested $500,000 in the fledgling company in its first seed round in 2006, and from public records held 22.4 million shares of the stock at IPO. Those shares, if he hadn't sold them, would be worth $1.9B today (at about $74 per share). That's a 3,800X return on capital. 1 

Multi-billion dollar companies happen when non-obvious ideas and huge market needs meet perfect execution. We've seen it before our eyes — Uber, Airbnb, Dropbox, Stripe, Instacart — and when you have the potential for 100X to 4000X returns, it's not about avoiding loss or minimizing downside. A proper venture portfolio is not like your 401K. The only way startup investors truly lose is if they miss the Uber. 

And that, in a nutshell, is why using anecdotes (e.g. Zulily in Sheiber's piece above) as evidence against Andreessen Horowitz makes no sense at all. An individual investment may fail but it's just one in a portfolio. The returns that are possible in early stage technology investing far outweigh anything Hollywood has ever seen or ever will see. Software is eating the world, and the numbers bear it out. 

1 David Hammer suggested a better comparison would be Accel, their Series A partner. Their 10% stake at IPO is now worth $14.8B, so their $12M investment yielded roughly 1100X return.
]]>
Garry Tan
tag:blog.garrytan.com,2013:Post/775153 2014-11-25T22:37:24Z 2015-09-24T19:34:56Z Spotify as a simple case study in making something people want

A passage in the New Yorker long form piece this week on Spotify really illuminates the key parts of what we spend a lot of time trying to help founders create: Something people want. 

Problem: Piracy was how people did it at the time. 

“It came back to me constantly that Napster was such an amazing consumer experience, and I wanted to see if it could be a viable business,” Ek went on. “We said, ‘The problem with the music industry is piracy. Great consumer product, not a great business model. But you can’t beat technology. Technology always wins. But what if you can make a better product than piracy?’ ” Ek continued, “Piracy was kind of hard. It took a few minutes to download a song, it was kind of cumbersome, you had to worry about viruses. It’s not like people want to be pirates. They just want a great experience. So we started sketching what that would look like.”

They started with a simple problem statement. All of the great era-defining startups seem to start that way. What was broken? They knew exactly what was broken. 

Solution: 200 milliseconds

Their “product vision,” in tech parlance, was that the service had to give the impression that the music was already on your hard drive. “What would it feellike?” Ek asked. “That was the emotion we were trying to invoke.” The key was to build something that worked instantly. Streaming, whether audio or video, tends to have built-in delays while you wait for the file, which is stored on a server in the cloud. But if the music starts in two hundred milliseconds or less—about half the time it takes, on average, to blink—people don’t seem to perceive a delay. That became Ek’s design standard. He told his lead engineer, Ludvig Strigeus, a brilliant programmer he had worked with before, “I don’t accept anything that isn’t below two hundred milliseconds.”

What would solve that very direct problem? Well, a great streaming service that was superior in a specific way. They knew how to measure success. Clearly if you could make a streaming service act as if it were on your local hard drive, you could win. 

Hard-to-build proprietary technology — the first time it was done, just enough of an advantage

Strigeus responded, “It can’t be done. The Internet isn’t built like that.”

“You have to figure it out,” Ek insisted.

The solution involved designing a streaming protocol that worked faster than the standard one, as well as building their own peer-to-peer network, a decentralized architecture in which all the computers on it can communicate with one another. In four months, they had a working prototype.

“And I knew when we had it that it was going to be very special,” Ek said.

This is a signature piece of why Spotify was special early on. There are two parts to this. The first is could it be done? Until these early prototypes, it hadn't. The second part is can it be cloned? As with almost anything software related — yes. Novel tech is not an infinite defense, but it was enough of one for Spotify to get off the ground. You don't need a technological advantage that lasts forever (though of course that'd be preferable) — just one that will last enough such that your competitors can only copy your innovation from 6 months ago. It can take that long to fast follow, which is enough to keep your unoriginal would-be competitors at bay while you blaze ahead. 

The Schlep = The Moat

Ek’s original idea was to launch Spotify in the U.S. at the same time that he launched the service in Europe. Ken Parks, Spotify’s chief content officer, said, “Daniel thought he could just go down to the corner store in Stockholm and pick up a global license.” He didn’t realize that he would have to negotiate directly with all the different copyright holders, a herculean task. Not surprisingly, the labels weren’t interested. Ek was an outsider—a techie, and a Swedish one at that. Parks, an attorney who’d worked at E.M.I., recalled, “We needed to overcome the music-is-free mentality that Spotify represented.” Of the labels’ attitude, he went on, “If you have something you’ve invested a ton of money in, and you’ve been selling it for a lot, and you feel raped by piracy—to say to that person, ‘The only way to beat this is to co-opt the people who are stealing from you,’ that was a challenge.” Ek said, “If anyone had told me going into this that it would be three years of crashing my head against the wall, I wouldn’t have done it.”

This is a common refrain. If founders knew how much work these things would be, often they wouldn't do it. But they do, and that's protectable long term enterprise value. 

This analysis is of course with 20/20 hindsight. In the moment, Ek and the Spotify cofounders and investors had no idea they'd be right. But they figured they might be, so they built it. 
]]>
Garry Tan
tag:blog.garrytan.com,2013:Post/774542 2014-11-24T18:05:22Z 2017-07-06T07:44:10Z Why Flow, a new low cost super high precision controller, is important for designers and creatives

Flow is launching this morning. They're YC alums who have created a low cost, high precision wireless controller in the form of a dial. It's highly programmable and most designers and creatives will find this to be super valuable because that's where precision really matters.

When I'm in Photoshop making pixel-perfect mockups, or when I'm in Lightroom editing photos, I'm constantly making micro-adjustments on specific settings, whether it be brush size, exposure, etc. I have to acquire the target, then move my mouse, and then click-drag to the point where I'm happy. We're exercising one fundamental law of UX over and over again - Fitts' law. 

Fitts' Law states that the difficulty of an action is determined by the movement time needed to complete that action, which is in turn defined by the size of the target to be acquired. Sliders are by nature long and thin. If I had to guess, a good chunk of the cognitive load of doing creative work is just moving a mouse pointer to a tiny slider bar.

Not only is it a tiny target to acquire, but there are finite number of steps in those sliders that can make a mountain of difference. For instance, photographers are always looking for that absolutely perfect exposure or temperature. With a slider, you're limited to the number of pixels that slider has on screen - 200px? That's only 200 gradations, and in my experience that perfect level is always in between two of those notches. 

Enter Flow. There are over 3600 distinct values in one full 360 degree turn of the device. And since you can link them directly to specific values e.g. exposure or brush size, you don't have to acquire the target over and over and over again. 

That's why I bought one, and that's why Flow is an important programmable hardware device that creative people should keep an eye on. They're accepting preorders now and are on Product Hunt, and if you get one early it's an extra good deal. 

]]>
Garry Tan
tag:blog.garrytan.com,2013:Post/755841 2014-10-15T20:06:27Z 2014-12-30T14:15:05Z Silicon Valley's Congressional Hope: Time to sweep away a congressman "who mostly votes the right way"

Silicon Valley hasn't had a real voice in Congress before. We've got a shot at one now though. In about 20 days, voters will go to the polls in a race for a congressional seat for the 17th district, and democrat Ro Khanna has a fighting shot at toppling incumbent Mike Honda. The polls are running in a dead heat, so this is one you should care about. 

There's nothing wrong with Honda — he's an old-line Democrat. Except that's actually the problem. He hasn't done much in the way of defending the things we really care about: Immigration reform, free Internet rights, supporting entrepreneurship, and reforming education. If Silicon Valley can elect Ro, then we as citizens are making a statement that business as usual for the Democratic party just isn't going to fly. 

That's why the San Jose Mercury News has endorsed Ro Khanna as well, saying "Silicon Valley -- whose economy, like the 17th District, stretches into the East Bay -- needs more than a congressman who mostly votes the right way... Silicon Valley's other representatives, Congresswomen Anna Eshoo, D-Palo Alto, and Zoe Lofgren, D-San Jose, are older than 65, and both are invaluable voices in Washington -- respected leaders on valley issues as well as defenders of progressive values. When they meet with us, they are insightful; we always learn something. This is not the case with Honda."

Ro is one of us. He's committed to reforming immigration so our talented friends who happened to be born elsewhere can still come here to create new businesses and jobs. Startups die every other day because of our antiquated and special-interest-ridden immigration policies. He's on our side when it comes to SOPA, PIPA, net neutrality and a maintaining a free and open Internet. 

If you're in the 17th District (Fremont to Sunnyvale), you have a chance to make history. Register to vote, and consider Ro Khanna for Congress. This race matters, and it's looking like a few hundred votes will swing this one way or another. 

]]>
Garry Tan
tag:blog.garrytan.com,2013:Post/748711 2014-09-30T05:31:12Z 2016-02-11T19:34:20Z Listen to the people for whom you're building: A low income housing development that chose bath tubs over hot water

I was watching an urban planning documentary called Urbanized on Netflix recently. It brought up a fascinating example of participatory design in the context of building homes. A housing development in Santiago, Chile called Elemental ran into a problem. The builders had to make a tough decision: should they build in hot water heaters, or should they build in bath tubs? The budget could only support one or the other at the outset.

A typical top-down approach would dictate that of course you'd want hot water. A first world view of the situation would say that you'd rather shower standing up with hot water than sit in a bathtub and have to heat water separately. 

Yet that's the exact opposite of what future residents of Elemental actually wanted. Architect Alejandro Aravena went out into communities and talked with residents and discovered what typical bureaucrats would never find - that people moving to the low income housing from slums would unanimously choose bathtubs instead. Hot water heaters and gas furnaces cost money, and are unfamiliar. Bathtubs, on the other hand, were very familiar (in fact what residents typically did in their existing living environments due to the extra privacy) and didn't generate additional energy cost. 

Further, hot water was one of the things that people typically added later, once they had acclimatized to the new living environment and improved their station in life. 

It seemed to me this was the sort of thing you could only tell by actually talking with the people who would use your creations. It is the ideal situation for us to create things for ourselves. But when you aren't doing that, you have to be extra careful about the assumptions and values you bring to the table. 

]]>
Garry Tan
tag:blog.garrytan.com,2013:Post/727579 2014-08-14T21:28:08Z 2014-12-08T01:09:56Z Always-on video recording will prevent tomorrow's Missouri police state — light is a disinfectant

The ongoing Missouri police fiasco has prompted a lot of discussion about the need for police transparency. My friend Jeff Lonsdale writes:

Active duty police officers need to be automatically recording everything they do. With recordings, incidents such as those happening in Ferguson can be quickly resolved one way or another. When tested in Rialto, California, recording reduced both complaints filed against police officers and the incidents where use of force was required. There will still be cases where police officers use excessive force in murky situations but by and large transparancy via recorded police and citizen interactions should protect the innocent parties, see more guilty parties punished and cause better behavior all around.

Smartphones, wearables, and always-on high bandwidth connectivity is converging in the next ten years to make this happen, not just for police officers, but for private citizens too. 

Always-on video is already shining light in Russia with the omnipresence of on-dash cameras. They say light is the best disinfectant, and that's exactly what video can be in the future. It's time to build. 
]]>
Garry Tan
tag:blog.garrytan.com,2013:Post/714512 2014-07-15T23:45:41Z 2014-07-16T01:56:30Z Software eats apparel — What MTailor means for how you'll buy clothing in the future

Miles and Rafi are two founders in the current YC batch who just launched their new startup, MTailor. It's a iPhone and iPad app that lets you get accurate measurements of your body so you can order made-to-measure dress shirts that fit you perfectly. You put the iPhone or iPad on the floor at an angle, and the app walks you through the 30 second process of turning around in place in front of the front-facing camera. The amazing thing is that the entire process is 20% more accurate than what a professional tailor would do in person. 

That's pretty damn cool. The founders are Stanford CS and math grads and devised and perfected the computer vision algorithms themselves. In the past similar founders would have tried to find some way to license the tech, but MTailor is building a new brand from scratch. It's an ambitious way for the company to fully create as much value as possible without middlemen and enterprise sales. 

Mass customization is finally hitting the mainstream, thanks to software. In the past it has always been difficult and time consuming for clothing to be made with specific measurements. It was a tedious process to begin with since it was hard for people to get accurate measurements, and even experienced tailors have trouble getting it right. Add a whole lot of waiting to the mix too — it'd take six weeks or more to wait for a shirt to be made. Those two problems taken together reduce demand. And as a result, the cost goes up, even further reducing demand. 

That's where a new software capability can come in, like MTailor's computer vision algorithms, and radically change the equation. Since it's easy to measure yourself, the major stumbling block is removed. And with a steady stream of orders, they can bring the price and wait down. That's exactly what the team has done, coming in at the $69 price point and a 2 week lead time. I was once a die-hard made-to-measure Indochino dress shirt fan, but I know where all my future purchases are coming from. 

That's the power of better, cheaper, faster, and how software eats the giant apparel business. Keep an eye on MTailor and try download their app. It's available right now. 

]]>
Garry Tan
tag:blog.garrytan.com,2013:Post/711546 2014-07-07T18:02:23Z 2016-08-30T07:04:47Z How a bug in Windows might be costing humanity over 600 years of wasted time per day

There's a Windows bug that I have been hitting for over a year that has driven me insane. My Downloads folder takes 20 seconds to show up in my File Explorer in Windows 8. To fix, I had to set the folder's type to "General Items" instead of "Photos" because Windows was trying to generate thumbnails over and over again. 

A program manager at Microsoft made the wrong decision. Microsoft has been a program manager culture for decades now. (Former president of Windows division Steven Sinofsky writes at length about it here.) PM's at Microsoft have the cat-herding role of trying to make things happen with little authority. As with most things that are started with very good intentions, there are unintended consequences. Bugs like the one above are one of them.

How did a bug like this stay in Windows for years? It's not like they didn't know about it. Someone on the team hit this problem and filed it as a bug in their bug DB. There are a lot of really dedicated software engineers and software testers out there who really do care about building great stuff at Microsoft. But it is the PM who has the unenviable task of sorting through hundreds if not thousands of these types of bugs and figuring out what gets fixed and what doesn't. On the day in question, the PM of the File Explorer probably saw this alongside 50 other bugs that were must-fix. They had a specific date to ship, and a fixed number of engineering resources. They knew their bug count had to glide down to zero by X date, and some bugs had to sadly be resolved "Won't fix." If a PM doesn't hit their schedule and ship, they get fired. This bug had to be punted. 

Say a billion users hit this bug and lose 20 seconds every day -- that's 633 years of human life wasted in the world. Normal people using their computers have no idea why this bug is happening, and just assume the computer is thinking. These wrong calls on small decisions have wide reaching impact when you multiply by the reach of software today.  

Great software is built by people making the right call on thousands of small decisions. We who create software have a responsibility to our users. Make the right choice. Fix the damn bug. And if you run a company that builds software, know better than to put a good person in a role that forces them to make the wrong decision. 

]]>
Garry Tan
tag:blog.garrytan.com,2013:Post/704718 2014-06-17T17:19:32Z 2014-07-02T20:40:11Z What Instacart taught us: Don't be a dreamer, be a creator

YC S12 company Instacart is well on its way to being one of the iconic companies of our times. How it started has been covered before: Instacart founder Apoorva Mehta hacked his way into YC by sending me a six pack of beer. It was a cold and refreshing pack of 21st Amendment's Back in Black, as pictured above. Apoorva sent it over using his own service two months after the deadline for applications at YC. The hack wasn't the thing that mattered, really, though. What mattered is what Apoorva had already built something amazing. 

I had already seen dozens of startups say they were going to do grocery delivery. But we had never actually seen someone build an app with thousands of products in it, with a few real drivers using the app to deliver real groceries every day. I downloaded the app that afternoon and was blown away at how much had already been done. That was absolutely remarkable. Everyone else was dreaming. Apoorva created it. And then everyone wanted it. 

That's the key lesson here. Don't be a dreamer. Manifest that dream. Build it. Write the code. Spin up the service. Hire. Create.

]]>
Garry Tan
tag:blog.garrytan.com,2013:Post/693639 2014-05-20T14:27:57Z 2014-07-11T08:33:37Z Living in a paperwork-less future

"Once you're living in the future in some respect, the way to notice startup ideas is to look for things that seem to be missing” —Paul Graham

"The future is already here — it's just not very evenly distributed." – William Gibson

Imagine: it is the near future. A mortgage application is a one-tap action on your smartphone. So is renewal of your driver’s license with the DMV. Paying your last parking ticket is a just one tap too. A car loan? A visa application? Life insurance? Car insurance? An estate plan? All a few taps. 

The promise of this has been a long time in the making. The next 20 years is not going to be like the last 20 years, however. Now's the time. Why? Because of a number of overwhelming new platform forces: 

Smartphone adoption — Late adopter industries are being unlocked because now far more people on the planet can have an always-on, cheap, ubiquitous computer. Android devices are $150 with no contract, today. A cheap tablet or smartphone can replace kiosks, point of sale devices, and PC’s on desks. You'd be crazy to say construction would be an industry where you grow a software business even 5 years ago, but that's exactly what Plangrid is doing bringing blueprints online. 

Digital signing — HelloSign is at the forefront of this. Instacart recently used the HelloSign API to be able to onboard their contract delivery workers far faster than ever. If you can sign legal documents instantly, you bring the friction of getting business down to nearly zero. 

Payments systems — StripeCrowdtilt OpenBalanced and WePay are at the forefront of this. It’s never been easier for a startup to stand on the shoulders of these giants and be able to do real transactions instantly. Coinbase does the same allowing anyone to receive payment using bitcoin, unlocking worldwide commerce with a universal currency. These things just didn’t exist even a few years ago. 

Legacy API's — Any system must work with all legacy systems. They’re getting filled in by both platform API services like Lob, Twilio, Plivo, and HelloFax. Not everything in the stack has to live in the future. 

Anti-fraud — Services like Sift Science are bringing world-class machine learning and anti-fraud technologies to every site and service. Online verification and personal data room services are coming online that will let people do large transactions with people they just met. Airbnb was able to use Facebook and Twitter data to verify people and make them more human, such that you could do business with them. LendUp is doing the same for payday lending, adding more data to the antiquated to the scammy world of payday lending. 

Killing paper-based purchase orders, application forms, approval queues and replacing them with all-digital direct-to-consumer experiences is happening right now, and in every walk of life. 

In the 1980's, Walmart started using IT to start working directly with manufacturers and cut waste out of their supply chain. Today, the company represents 2% of annual US GDP. That same thing is finally coming to all the other industries, and they're being brought to the world by small teams of talented founders building on the fundamental platforms mentioned above. 

New services are better, cheaper, and faster. As a result, these new businesses will be built shockingly fast. It’s not twilight for young startups trying to make something great. It is dawn. 

]]>
Garry Tan
tag:blog.garrytan.com,2013:Post/679095 2014-04-19T06:46:23Z 2014-05-02T21:03:44Z One exception to the "solve your own problem" approach to startup ideas

I got an email from a founder recently who wanted to help create software so that you could find other cofounders. It's a common idea that is so frequently attempted that I usually try to dissuade people from working on it. This is what I wrote him: 

It is usually the right thing to create something you yourself would use. Finding cofounders is a common problem that founders face, but it is probably a special case exception to the guideline. There probably aren't enough people who want to do startups, and you won't be able to make enough money from those startups to actually support your business. 

I would suggest that you work more on things that a lot of people in your world can use. If you could pick anything, the ones that are most valuable tend to come from there. 

There are probably specific aspects of your local economy and markets that you know about that nobody else knows about. Make software to solve those problems. For instance, I've met founders in the past who have backgrounds in the an unsexy business like textiles, or manufacturing. They end up making marketplaces or management software that make those kinds of operations a lot more efficient, and they're the only ones who can do it because it is rare for someone who knows how to code who also understands all the ins-and-outs of that particular industry and use case. 

You can think of this being true for any economic activity: banking, real estate, retail, wholesale, shipping, food, entertainment, transportation. There is an infinite amount of software just waiting to be written, around use cases that people actually want and care desperately about. That's what you should focus on. 
This is an unusually ripe time for people to make software that touches late adopter industries like the ones mentioned above. 

It's not enough to be focused on your own needs. This is why empathy is so important as a founder. You really do need to focus on the needs of others. That's the path to creating something people want. 
]]>
Garry Tan
tag:blog.garrytan.com,2013:Post/669415 2014-03-29T15:57:04Z 2014-05-12T04:16:22Z The outrage that shouldn't be: Oculus Rift pre-orderers aren't angry that they didn't get equity. That's not how it works.

People are up in arms about the Oculus Rift selling to Facebook — some like the normally reasonable and insightful Barry Ritholtz are livid that people paid to the tune of $2.4M in presales and aren't sharing in the equity returns

I bought an early Oculus Rift developer kit as a part of the Kickstarter. I paid $300 for it and it was worth every penny. I had no expectation of equity from it, because the product was cool and something I wanted. It's the wrong thing to criticize Oculus Rift for the simple reason that hardware crowdfunding has unlocked a whole new world of things that could be built if only you could prove people want it.

The crowdfunding campaign Oculus Rift ran proved people wanted it. We see this over and over again — if you can create a new class of consumer behavior (Uber, Google, etc) then you can create a valuable company.

Articles like this are sensational in that they sometimes incite misguided lawmakers to regulate. But in this case, crowdfunding is creating new economic activity, and any new regulation would actively squelch that — and kill viable new products and companies in the process.

]]>
Garry Tan
tag:blog.garrytan.com,2013:Post/664760 2014-03-18T15:00:05Z 2017-03-15T09:55:36Z Last Minute Advice for YC Applicants, Updated for 2014

In 2009 I wrote a blog post here about advice for startups applying for YC. Little did I know years later I'd become a partner at Y Combinator and have the chance to shape hundreds of new startups since then. 

I thought I'd refresh that article since five years is a long time. Here are my six tips for applicants, updated for 2014.

1) YC is a multiplier for companies, both early and late. 

The most common question I get from teams is wondering whether they are too early for YC, or too late. I think YC can help any company prior to Series A. 

YC teams span the gamut. Some people are just out of college and might only have an early prototype. Others might have already raised a seed round and have real revenue and employees. Some of the YC companies who have done the best (e.g. Crowdtilt, Thalmic Labs) came into YC having already raised seed rounds. They do so well during YC that they have gone on to raise competitive Series A and subsequent rounds of funding right out of demo day. It turns out to be a multiplier on success. 

On the other hand there are teams that are just super early. That's OK too. If this is your first money you're seeking, realize that this is unlike a summer program or getting an internship. You're committing yourself to building a serious business, and you're telling your investors that you're going to do what it takes to returning their money plus a healthy return. It's not a grant. We do startups because we want to create massive value, reward ourselves and reward the people who help us create that value. 


2) Solve a hair on fire problem, or do it right

Great hackers get caught up in technology, but technology doesn't create value in and of itself. Technology is only useful for solving people's problems. 

One of the first pieces of startup advice I remember coming across was Paul Buchheit talking about startup ideas. Successful ones fall in one of two categories:

  • They either solve a hair on fire problem that has never been solved before
  • OR they solve a problem that has other solutions, but they do it so much better that they are "done right."
If you're a hair on fire problem, you have to clearly state how bad that problem is. How do you know the problem exists? Who has it? How specifically have you solved it? When you're doing something new like solving a brand new hair on fire problem, it's an uphill battle just to let people know you exist. We want to know that you are capable of even getting people to know you exist. 

On the flip side, if you're focused on a more crowded space and you're trying to do it right, then you've got to be a lot more clear about how your solution is unique and different. It's not OK to do exactly the same thing. Help us understand what you have figured out that nobody else has. It's so important, this is actually one of the questions on the application. 

If you're new, focus on explaining what you are. If you're not new, focus on explaining how you're different. If you're neither of these types, you're doing it wrong.



3) Have a capable team

The ideal startup team involves really two major roles — builder, and hustler. I used to say it took three roles (designer, engineer, hustler), but in practice I've seen enough evidence to the contrary that I have changed my mind. In reality, I think designer / engineer can be abstracted to builder. Some enterprise software companies don't need a ton of design. All tech startups need tech cofounders though. 

I've seen teams of hustlers fail because they can't create any product on their own. I've seen teams of pure builders (amazing designer / engineers) that fail because they can't put the product in the hands of users. In the end, you need both. Your cofounding team needs these things, and if you don't have them yet, you should go find your most talented and trusted friends who have those skills and get them to come join you. 

You need these skills because they're the basics for success. It doesn't have to be one person specifically for one role or the other. It's just important that people are capable of those things.  Be realistic with yourself about what you and your teammates are good at. You are getting married to your cofounders -- you're essentially getting out onto a life raft in the ocean with them, and you're going to need to work with them (and well) to survive.

Doing a startup without a kickass team who can really clean up on each of the three skills is going to war without guns, ammo, training, or all of the above. 



4) Write well

In the past five years, this has become more and more obvious to me as the examples pile up. The best startups are incredibly good at writing and communicating in as few words as possible. Being concise and clear matters. PG taught me years ago: If a founder can't express their ideas clearly to others, then the simplest and likeliest explanation is actually that the idea is not particularly clear to them either. 

PG's essays are an indication of what you should be striving to do in your application. You should read these essays now if you haven't already. Everything you write should be crisp, articulate, and to the point. 

One way to do this is to actually write every last idea down. Write copiously. Brainstorm to get it all. Then edit. Connect similar concepts and put them next to each other. Edit mercilessly until there is not a single word you could remove without losing significant meaning. 

One common pitfall that plagues practically everyone is the urge to sound professional. Don't do that. Just use normal plain English that you would use to explain things to normal human beings who have no special background. It's a mistake to try to sound corporate, or professional, or like a press release. Those are all among the worst ways to communicate in the history of language. What you say should communicate, not get in the way.

Simple rule of thumb: Does the writing sound like something you would say to a friend? If not, rewrite. 



5) Get a little help from your friends

You've got mentors, right? Get as many trustworthy and intelligent eyeballs on your written application as you can. This goes for any application for anything, really. More eyeballs will always help you flesh out your concepts, spot weak links and strengthen your case.

Email founders of YC companies, or founders of any company, to get feedback. We're here to help -- someone helped us.



6) What are you going to do if you don't get in?

You've got to keep working on the startup no matter what. Realize that you don't need anyone to give you a permission slip to create your own startup.


The application deadline for Summer 2014 is March 28. Apply now.

Good luck to all the applicants, and I hope to see you at interviews.
]]>
Garry Tan
tag:blog.garrytan.com,2013:Post/646834 2014-01-27T21:14:59Z 2014-01-27T23:45:54Z Like an inflatable hammer: Unusable software and its origins

We use software to get things done. We hope these products will help make our lives better, and that's why it's such a blessing to be a software engineer and startup founder. There are an infinite number of things we can make that potentially make things better. Yet as software creators, we fail — over and over and over again. 

Just in the past day, two big brands failed me in major ways through poor software engineering. Last night I bought an Xbox One game and tried to play it. I let it try to install overnight. The software is stuck now — complaining the installer hit network connection issues and won't continue, even though clearly the Xbox One can connect to the Internet. There's no retry button. I'm in a broken state and I don't know how to resolve it. This is a pretty basic use case that is completely broken. 

This morning, I tried to use Silicon Valley Bank's new deposit-by-mail feature. First off, it took me over a week and two emails sent in order for them to enable it on my account. I shouldn't need emails to enable a feature like that, let alone wait a week. Then, when they finally did enable the feature, it didn't work at all for the handwritten checks I needed to deposit. The error? "The amount doesn't match the check image." The engineers probably used some off-the-shelf OCR system that barfed on handwriting. There's no workaround. Yet handwritten checks are a pretty core reason why someone would use a mobile phone app to deposit a check. The core use case is completely broken. 

This is the inflatable hammer problem in a nutshell — too often, we are given tools that cannot be used for their described purpose. They're broken in fundamental ways. 

In the end, SVB and Microsoft suffer from broken product and engineering organizations. These organizations have plenty of resources to make these simple core scenarios work. But it's not a resource issue. These failings usually happen because designers and engineers who actually make the software are divorced from the management organization that sets the deadlines. There's no sense of ownership — merely hitting milestones and knocking off items on a to-do list. I hit my goals — ship it and let's get a beer. 

Large orgs also make the mistake of insulating their product and engineering people from the people who actually experience the pain. There are layers of engineers and management and emails go to account managers who can't do anything about it and don't know anyone who possibly could. When I was a PM at Microsoft, the only interaction I had with end users was reviewing Windows Mobile help newsgroups once a quarter. It was like wading through an ocean of pain. It was so sad and ugly to look at, and that was the software we were making — that was the pain we were inflicting on our users.

Silicon Valley is obsessed with the auteur-CEO for good reason. When MobileMe shipped crap, it wasn't OK with Steve Jobs and clear action was taken. When idiotic product decisions are made, it takes someone with power to lay the smack down and enforce a culture in which that kind of thing isn't acceptable. 

Startups don't have a creaky old dumb organization yet — it's just founders, right? Yet startups ship bad software all the time. Why? Well, founders have limited time, and the advice to ship-fast/fail-fast is a strong driver. We ship software when the basic case works at all. This is probably still the right thing for people do. But then founders forget to fix things that are broken. The moment your software works at all feels like you're 80% of the way there, but you're really at 20%. A great product is made, not born. 

Startups have an even greater chance of success at products because they can listen to users. You don't have many, so you can actually read and answer every email you get. Take a page from Wufoo, who made their engineers and designers do support every single day. This is the best thing you could possibly do. Connect with the pain of your users, and fix the papercuts, no matter how small. That's how you get to a great product. 

Don't make the same mistake the big orgs make. If it sucks, cut scope or slip the schedule. Don't cut quality and don't strand your users in frustrating bugs in core scenarios. Fix it and make it better. Listen to users. Fix it. Repeat. 

SVB and Microsoft can endure a lot of failure before it affects their bottom line. They've got a lot more things going for them — one is a bank, and the other has a few durable monopolies that have decades of life on them still. But for startups, all you've got is whether the damn thing works. So ship it and make it better. 

]]>
Garry Tan
tag:blog.garrytan.com,2013:Post/645760 2014-01-24T22:37:36Z 2014-01-24T22:37:37Z Tenth Grade Tech Trends, 2014 Edition: Snapchat usage up 3X — 39% of teens use it regularly

In early 2013, I posted results from Survata that indicated that teens really did use Snapchat, Instagram and Tumblr. About one year later, Survata ran the survey again, and this is what they found. 

A year ago, only 13% of teens used Snapchat at least several times a week. That has since increased to an incredible 39% as of January 2014. These are very impressive numbers alongside social media stalwarts Facebook, Instagram, and Twitter. 

I've heard some Valley insiders note that the growth and traction numbers for Snapchat would justify valuations of upwards of $20B if you used comparable metrics from valuations of Asia-based messaging services like Line. 

The attention economy continues to evolve, and not all the spots on the social media periodic table have been filled quite yet. An upstart with a unique model that taps real user behavior can still gain traction in an incredibly fast rate. 

As a side note, YC-backed Survata remains a really valuable tool to gain insights and run surveys against lots of Internet users quickly. See the full methodology for their teen social media study here.

]]>
Garry Tan
tag:blog.garrytan.com,2013:Post/616016 2013-11-07T00:21:30Z 2014-04-21T18:10:38Z The API-ization of everything

Everything that used to be fax machines, contracts, and invoices and purchase orders in triplicate are going away. What's replacing it? The API.

Before Stripe, you had to fill out a ton of forms and get approval to get a merchant account. Now you can sign up on a website and make simple REST API calls. Before Twilio you had to go through similar hoops just to be able to work with phone numbers. Once again, the API prevails, and whole new capabilities are unlocked. 

Recent YC company Lob is bringing this same kind of capability to print anything — on any kind of paper, any size, and mail it to anyone — all via a simple REST API. This is a big deal because paper is actually still the interface for the rest of the world — the rest of the world that still uses paper contracts, paper checks, and paper invoices. The company is already printing millions of dollars worth of checks!

Think about most of the bad customer interactions we end up having to have on a typical day — to do anything in business, really. We get on a phone and we get voicemail. We wait a few days for our purchase order to be approved. If software is eating every industry and every type of economic activity, then the force that is replacing it is actually software talking to other software. That's the API.

This world has been a long time coming. We've heard enough about e-business thanks to huge ad budgets by enterprise IT consultant behemoths! But rather than those huge wasteful IT firms, the people bringing it to us are the hacker-oriented dev-savvy API companies like Stripe, Twilio and Lob. Where there is paper to push, a call to answer, or a purchase to approve, there is an API coming to replace it. 

]]>
Garry Tan
tag:blog.garrytan.com,2013:Post/611781 2013-10-22T18:38:49Z 2014-10-19T21:40:26Z Fear is like fire

One of the first lessons Cus D'Amato taught Mike Tyson was about fear: 

“Fear is the greatest obstacle to learning. But fear is your best friend. Fear is like fire. If you learn to control it, you let it work for you. If you don’t learn to control it, it’ll destroy you and everything around you.

“You think you know the difference between a hero and a coward, Mike? Well, there is no difference between a hero and a coward in what they feel. It’s what they do that makes them different. The hero and the coward feel exactly the same, but you have to have the discipline to do what a hero does and to keep yourself from doing what the coward does.”

That fear exists in startups too. There is much to fear in every new endeavor. Are you shipping fast enough? Are you making things people actually want? Are there enough people who want what you have created? Can you reach them and sell to them? Is this going to work? Are you going to live?

There are some startups that raise so much money that the fear is gone. There is infinite tomorrow. This is a great problem to have, but a problem nonetheless. For them, the fear must come from elsewhere. There may be time, but not for this idea and this market. Competitors are coming and they're coming for you. Ah, there's the fear again... it's useful. Use it. Harness it. 

There are others, which is to say most, where the opposite is true — there are six months left. Three months left. Or less. Then there is an impenetrable 20 foot tall wall of fear. Death is around the corner. Again, this fear will focus you. Make you ship faster. Force you to ask the hard questions, and make the changes you knew you had to do. 

All startups feel the fear. Like Cus said, there's no difference between what a hero and a coward in what they feel. It's what they do that makes them different. 

]]>
Garry Tan
tag:blog.garrytan.com,2013:Post/602112 2013-09-17T09:37:48Z 2014-10-19T23:02:54Z I am Jack's desire for a ice cream cone: Your brain is 100 billion cells working together

Earlier this evening I was debating whether or not to eat a Drumstick, one of my favorite ice cream treats. I really wanted it, but some other part of my brain short circuited that decision and said nope, fatass, you're not having it. I felt viscerally torn. I had some desire that must have emerged from the neurons of my brain having to do with eating, and I had to override those desires using some sort of higher order logic around the fact that I lead mostly a sedentary existence and reallly don't need those calories at 2AM. 

Shortly after that internal debate was won, I happened on an interview with Tufts University Professor of Philosophy Daniel C. Dennett that pointed out perhaps those cells really are warring

Each neuron is imprisoned in your brain. I now think of these as cells within cells, as cells within prison cells. Realize that every neuron in your brain, every human cell in your body (leaving aside all the symbionts), is a direct descendent of eukaryotic cells that lived and fended for themselves for about a billion years as free-swimming, free-living little agents. They fended for themselves, and they survived.    

It's an interesting thought experiment. At some point, single celled organisms started banding together. Oh, you know, about 1 billion years ago (clipped from an amazing infographic recently at waitbutwhy.com)

After about 400 million years, those few early multi-cellular lifeforms got progressively more advanced, resulting in the first fish about 500M years ago, then insects 400M years, then reptiles 300M years ago, then mammals about 200M years ago.

So that's a lot of years for things to progress from just a few cells going on a date to now about 50 trillion cells coordinated in unison in any given human body! Though perhaps the most important part, the brain, consists of somewhere around 100 billion of them. 

So that's wild. The experience of one human being, my experience, and your experience, seems to be the product of 100 billion neurons working together — fighting, debating, forming alliances, and ultimately making decisions. 

That gives me some hope for humanity overall, then. If 100 billion in-it-for-themselves descendants of selfish eukaryotes can work together for the greater good, whatever it is you choose it to be — then perhaps 7 billion people can do the same for the humankind. 

]]>
Garry Tan
tag:blog.garrytan.com,2013:Post/595384 2013-08-16T21:22:45Z 2013-10-08T17:28:39Z Lost to history

Kansas City sportswriter Martin Manley committed suicide yesterday at the age of 60. Before his death, he posted a website explaining why he did it. He still had his health, his mind, and felt that he had contributed everything he could to society at that point. 

On why he posted his website, he writes

It’s incredible what is lost to history. Even one generation back, so little is really known about the vast, vast majority of people. It’s a shame - especially considering existing technology has enabled us to store everything ever said, done or thought by every person on earth a thousand times over. There is no reason lives have to continue to be forever forgotten. Everyone can be remembered if they put in the effort to be remembered in some concrete form rather than simply being dependent upon fading memories of those who knew the deceased person.

From beyond the grave, it is a particularly striking thought. The technology exists to store everything ever said or done, and the cost is coming down all the time. I hope that's what Posthaven can be. It'd be nice to leave something behind. But there is much left to do to make that happen.]]>
Garry Tan
tag:blog.garrytan.com,2013:Post/592249 2013-08-04T21:14:56Z 2013-10-08T17:27:59Z Goldman Sachs sent a brilliant computer scientist to jail over 8MB of modified open source code uploaded to an SVN repo

In 2009, a brilliant software engineer Sergey Aleynikov was arrested by the FBI at Newark Liberty International Airport. The allegation? He stole Goldman Sachs source code, about 8 megabytes of it. But it wasn't purely GS code — It was open source code mixed with Goldman Sachs proprietary code. If anything, if the source code was LGPL or a similar license (common among open source projects), Goldman Sachs was actually supposed to release this code back out to the community.  (Clarification: If the binary is distributed, it must be released back. Not required legally in this case.)

In Vanity Fair, Michael Lewis writes:

Serge quickly discovered, to his surprise, that Goldman had a one-way relationship with open source. They took huge amounts of free software off the Web, but they did not return it after he had modified it, even when his modifications were very slight and of general rather than financial use. “Once I took some open-source components, repackaged them to come up with a component that was not even used at Goldman Sachs,” he says. “It was basically a way to make two computers look like one, so if one went down the other could jump in and perform the task.” He described the pleasure of his innovation this way: “It created something out of chaos. When you create something out of chaos, essentially, you reduce the entropy in the world.” He went to his boss, a fellow named Adam Schlesinger, and asked if he could release it back into open source, as was his inclination. “He said it was now Goldman’s property,” recalls Serge. “He was quite tense. When I mentioned it, it was very close to bonus time. And he didn’t want any disturbances.”

Open source was an idea that depended on collaboration and sharing, and Serge had a long history of contributing to it. He didn’t fully understand how Goldman could think it was O.K. to benefit so greatly from the work of others and then behave so selfishly toward them. “You don’t create intellectual property,” he said. “You create a program that does something.” But from then on, on instructions from Schlesinger, he treated everything on Goldman Sachs’s servers, even if it had just been transferred there from open source, as Goldman Sachs’s property. (At Serge’s trial Kevin Marino, his lawyer, flashed two pages of computer code: the original, with its open-source license on top, and a replica, with the open-source license stripped off and replaced by the Goldman Sachs license.)

Aleynikov decided to take another job, but in the meantime he stayed on at Goldman to help out. That's where it got sticky for him:

He agreed to hang around for six weeks and teach other Goldman people everything he knew, so they could continue to find and fix the broken bands in their gigantic rubber ball. Four times in the course of those last weeks he mailed himself source code he was working on. (He’d later be accused of sending himself 32 megabytes of code, but what he sent was essentially the same 8 megabytes of code four times over.) The files contained a lot of open-source code he had worked with, and modified, over the past two years, mingled together with code that wasn’t open source but proprietary to Goldman Sachs. As he would later try and fail to explain to an F.B.I. agent, he hoped to disentangle the one from the other, in case he needed to remind himself how he had done what he had done with the open-source code, in the event he might need to do it again. He sent these files the same way he had sent himself files nearly every week, since his first month on the job at Goldman. “No one had ever said a word to me about it,” he says. He pulled up his browser and typed into it the words: Free Subversion Repository. Up popped a list of places that stored code, for free, and in a convenient fashion. He clicked the first link on the list. The entire process took about eight seconds. And then he did what he had always done since he first started programming computers: he deleted his bash history. To access the computer he was required to type his password. If he didn’t delete his bash history, his password would be there to see, for anyone who had access to the system.

...

The story the F.B.I. found so unconvincing—that Serge had taken the files because he thought he might later like to parse the open-source code contained within—made complete sense to the new jurors. As Goldman hadn’t permitted him to release his debugged or improved code back to the public—possibly in violation of the original free licenses, which often stated that improvements must be publicly shared—the only way to get his hands on these was to take the Goldman code. That he had taken, in the bargain, some code that wasn’t open source, which happened to be contained in the same files as the open-source code, surprised no one. Grabbing a bunch of files that contained both open-source and non-open-source code was an efficient, quick, and dirty way to collect the open-source code, even if the open-source code was the only part that interested him.
And so this case is disturbing to me, because a software engineer like any of us is assailed by one of the more infamous financial institutions in the world. He violated confidentiality, but 8 years of jail, really? He's spent 11 months in prison already. He didn't steal the crown jewels or the secret sauce— his acquittal was on the basis that the code saved to SVN wasn't the proprietary trading strategies at all, and it was extensions to open source software that he wrote himself. 

Yet Goldman Sachs pursued criminal charges against him anyway. And continues to pursue him. 

I'd love to know what specific pieces of software they were. Software engineers would be able to tell what was kosher and what was not. It scares me that a jury of non software engineers (truly, not a jury of Aleynikov's peers) will likely be responsible for deciding his fate. (Edit: Note, 8 megabytes is actually a good deal of code. But that's why what code it is matters so much. There's little transparency in this case here, so we are left to speculate.)

Serge was acquitted via the 2nd Circuit Court of Appeals, and released in February of 2012. (photo above) He has since been re-arrested and is being tried by the state of New York. In the United States we have a thing called double jeopardy — you can't be tried for the same thing twice. Somehow that doesn't apply here. Not when Goldman is after you. 

Sergey Aleynikov faces two felony counts in New York. 

]]>
Garry Tan
tag:blog.garrytan.com,2013:Post/584312 2013-06-15T20:29:27Z 2013-10-08T17:26:27Z David Brooks doesn't know technology, which is why we shouldn't listen to him when it comes to his opinion on surveillance

From a recent PBS transcript, David Brooks has been very publicly engaging in ad hominem attacks on Snowden. That's just fine — you can tell his case is weak because he's resorting to the weakest of all arguments: the ad hominem attack. But the justification for mass data sweeps makes no sense. He says:

He betrayed the cause of liberty, because, if you don't have mass data sweeps, well, then these agencies are going to want to go back to the old-fashioned eavesdropping, which is a lot more intrusive.

This statement is plainly false. "Old fashioned" eavesdropping is actually far less intrusive than what we're talking about today. Eavesdropping on a telephone conversation on an analog telephone (e.g. back in the day when J Edgar Hoover wiretapped Martin Luther King Jr.) required a court order, and actual people to record and listen in on specific conversations in realtime. 

You could only eavesdrop on a specific person at one time. Now there's infinite lookback on anything you've written, who you've talked to, and any email, text message, or IM for your entire life. That was one of the most important topics of Laura Poitras's NY Times mini-documentary "The Program", with NSA whistleblower William Binney. A 32-year old NSA insider turned whistleblower, William Binney describes the actual goal of NSA surveillance:

Think of a domain as an activity. A specific type of activity. Phone calls. Banking is another. If you think of graphing each domain, and then each graph turn it in the third dimension... pulling together all the attributes that any individual has in every domain, so that now I can pull your entire life together from all those domains and map it out and show your entire life over time.

This is the shift. Now we live in an age of infinite storage! We are only now capable of keeping every aspect of our lives accessible at any time, within milliseconds. We find great utility in it, and we entrust our greatest tech companies in the world with our data: Facebook, Twitter, and Google. 

We use these services believing that they are personal, for ourselves. Yet this is the same capability that now, via PRISM, the government can potentially have, all through secret courts and little oversight. All the things you do, and everybody you talk to, and down to the very thoughts you care to put down in both private and public communications. 

Temporally, that means this new surveillance (what David Brooks dismisses as merely mass data sweeps) is actually past and present. Whereas the "old-fashioned" form of surveillance could only be one the present. It's a capability not even touched upon by Orwell when he wrote 1984. It's not enough to be clean, sober, free and clear right now. You've got to be spotless and perfect over the course of your entire life. 

To dismiss mass data sweeps as less intrusive than old fashioned wiretaps is either ignorance, or a deliberate attempt to mislead the public. David Brooks isn't an engineer, and doesn't seem to understand the technical differences between the old regime and the new, so I'm willing to give him the benefit of the doubt. 
]]>
Garry Tan
tag:blog.garrytan.com,2013:Post/581407 2013-05-28T00:35:39Z 2013-10-08T17:25:52Z Rageskitching Facebook 2013 Edition: 3px of padding makes all the difference in the world

I'm sorry for rageskitching about Facebook, but I can't help it. I've seen this bug too many times and I would really like someone with commit access to the Facebook CSS to fix it. 

It's a matter of 3 pixels, but it matters. Here's what it looks like right now live:

And here's what it would look like fixed. 

Can you see how much more crowded the text is in the unfixed version, and how much more balanced it could be?

I also couldn't help myself and fixed the bottom area of the comment addable extra too— that stuff needs just a tiny little bit of love. I can't stand uneven padding. A site that fixes these problems is a site is being taken care of with the loving attention to detail of someone who can see the difference.

3px means a lot more for a site like Facebook than the pixels themselves. 

]]>
Garry Tan
tag:blog.garrytan.com,2013:Post/421462 2013-04-19T21:46:05Z 2013-10-08T16:52:31Z Bitcoin is not a toy - it has the potential to disrupt credit cards and banks and make payments zero-fee. That's the real deal.

I spoke with Michael Copeland of Wired recently for his piece on Bitcoin businesses, What’s Riskier Than Bitcoins? Bitcoin Companies

Y Combinator Partner Garry Tan is making the bet that Bitcoin succeeds with his investment in digital wallet startup Coinbase. He agrees that having some Bitcoin squirreled away is wise (and you can bet the craftiest VCs are leveraging up their own Bitcoin caches by investing in companies that takes a whack at improving the Bitcoin ecosystem).

“I have done some of that,” Tan says laughing. “But the buy and hold thing, that is not interesting to me.” What is interesting to Tan is backing companies like Coinbase that he believes can help Bitcoin make the transition from speculative “toy” that people horde like gold rather than spending, to something that people can actually use.

Tan’s vision for Bitcoin is as a universal currency that will make money transfers easy and cheap from any person or business to any other person or business across the globe. In that scenario Coinbase becomes a far-cheaper version of Visa. “Bitcoin has the potential to destroy credit cards and banks as we know it, maybe that is a good thing or a bad thing,” Tan says. “But I like the idea that if someone needs to remit payments they can do it without being gouged.”

If Bitcoin starts to become a widely used digital currency, something that people actually spend on products and services, Tan’s vision, and that of every Bitcoin economy startup and their investors, begins to make sense. 

Read the full article on Wired

Anything that has the potential to break Visa/Mastercard/American Express's stranglehold on 2% of every consumer transaction is awesome in my book. It's an absurd relic of the past and should not continue to exist, yet it does.

In my conversation with Michael, I did say that it felt to me that Bitcoin has established itself at this point -- barring a major technical flaw, I think Bitcoin is probably the right currency and others that come along will find itself like Beta in a world where VHS has already established itself. He cleverly inverted it in his conclusion. It's an open question who is right as yet -- but as investments go, Peter Thiel is widely known now for saying that the most disruptive revolutions start with a simple idea where everyone says I believe X, but the founders believe NOT X. 

Bitcoin hasn't won the world over yet, but that's OK. This is one NOT X that I think is right. 

]]>
Garry Tan