Andrew Chen Archives

After staring at the 320 hours of paid vacation on my pay stub for months, I finally decided to take a week off of work. Now, why did I allow such an enormous amount of vacation to accumulate? Well, basically because I usually think of vacation as unnecessary and unproductive :)

So when I did take vacation, of course I used it to catch up on reading, go through an introspective bent where I evaluated where I was in my life, and other fun things like that. I’ve come to the realization that taking time off is actually very productive, as long as you are using it for the right thing. In general, it helps break up the time between long stretches of work, and forces you to think about the meta things, and meta-meta things, rather than get stuck too close to the weeds.

Rob, a wise and all-around nice guy, challenged me to think of the following, in regards to my professional life:

1. Am I getting challenged/pushed on a regular basis by my current job?
2. Am I continuing to develop new skills or augment my current skills?
3. Am I broadening the number of interesting people with whom I interact with and who will be valuable for networking as my career progress?

He also noted that within these questions, he specifically left financial rewards off. He said, for someone who is young and thinking through these questions, financial rewards will come – instead, focus on learning.

The idea, I think, is that like all "assets," people go through an "investment" phase and a "collecting money" phase. At the beginning, young people often have negative net worths because of school debts, and spend all their time learning how to be productive human beings. Now, it is true that some 20 year-olds actually create value in addition to sucking it out of the system – otherwise we wouldn’t have 28 year old billionaires – but the norm is to only start adding lots of value back into the system once you become a responsible 30 or 40 year old. Either way, the important part is to be investing when you are young, and hit your peak later on.

A corollary to this is that if you find yourself "harvesting money" too early, you are probably in the wrong job. The reason is that for most 20 year olds, you don’t have that much to give back, so harvesting money at that stage is just another way of saying "dead end job." So don’t do that.

Another way to ask Rob’s questions above, in a less systematic and formal way, is to ask, "If you had $50 million dollars, what would you be doing right now?" (Of course, substitute $50 million with whatever your "number" is – and yes, I know this question was in Mike Judge’s brilliant Office Space) Asking yourself that question frees yourself from financial constraints, and removes social status as an issue – and instead, it focuses on your "true" goals.

Then the second step of this question, of course, is that even if you don’t have 50 million dollars, as young and resourceful lads, you should be able to figure out a way to work towards that goal, regardless of your current income. This is especially true for Ivy League Microsofties, who an get a "real job" anytime they want, but somehow view their world as trapped behind a desk in Redmond.

Now hilariously enough, many people I’ved asked this question to really answer "I’d invest it," which tells you they are altogether too responsible :)

My friend Tim Higgins, whom I admire greatly, has his answer and his current life aligned completely. He worked in the financial services industry, saving money over 4 years to fund 2 years of golf. Tim’s attempting to be a pro golfer, and he would be doing this regardless of the money situation. I remember in a conversation I had with him about it, he said, it’s better to try now than when I’m older, and also that it’s better to try and fail rather than regret what "could have been" years later.

Anyway, my life is not 100% aligned right now, but will be quite shortly. But from now on I’m taking vacations more often to think about stuff like this :) Better to risk a week of productivity than a year heading towards the wrong direction.

A great article featuring The Tipping Point’s Malcolm Gladwell: The Myth of Prodigy and Why it Matters.

I remember that it was a tremendous obsession by my parents to get my sister and me into the "gifted" programs in the Seattle School District. At first, we were in a program for the top 5%, then in a program for the top 1%, and then by 7th grade, they had me apply to the Unviersity of Washington’s Early Entrance Program for kids to start college full-time at 14. There, we had people tell us constantly how we’d all change the world, and how we had so much potential, for being a bunch of "child geniuses."

In retrospect, it was all silly. It’s true that every couple years, we’d see a "true" genius come through the program that would easily outclass everyone. For example, we had a couple kids come in at 12, whiz through their classes to graduate before their 16th birthday, and finish with honors in a hard science discipline – that takes some real talent. But outside of these exceptions, 95% of the kids in the program were smart, but not exceptionally so. What they had was work ethic, highly involved parents, and that made it difficult to separate natural ability and lots of practice.

More importantly, most of the kids that then graduate from the program suffer a sad introduction to the Real World. Other than the group that go into academia (which is 40-50%), there are high levels of unemployment, or unfulfilling employment. We have people working in lumber yards, at Target as security guards, at dead-end office jobs, people sitting at home with their moms, and all sorts of other random gigs. It’s really a pretty disturbing waste of talent to see guys with 150 IQs doing hard labor for $30k/year. But it happens because of what Gladwell refers to as the "prodigy midlife crisis." Basically, people tell these kids they have all sorts of potential, but when they graduate, they’re unable to translate that potential towards something that can utilize their gifts. And more importantly, it may be that they are very good at school, but not necessarily good at innovating new things, or leading teams of people, or communicating ideas, or other important skills.

The sad truth of it is, for many of the kids that have gone through the program, it’s possible they would have had better lives by just going through high school normally, and then having a normal college experience.

Anyway, this discussion reminds me of what happens to people through their careers. At the beginning, a lot of rewards get heaped on people who are able to follow instructions well, analyze things in detail, and other individual-contributor tasks. In fact, at many companies, young people spend many years in their early career fulfilling these tasks. But as these people get promoted, the demands on them change drastically, in a way that they may not respond to well. Rather than doing the work themselves, instead their job is to help other people do, which may involve more soft skills like persuasion, information-transfer, politicking, etc.

This sometimes leads to the common concept of the "Peter Principle" which is stated as:

In a hierarchy every employee tends to rise to his level of incompetence.

This obviously means that eventually, people stop being able to respond well to the varying responsibilities they are given, and they hit a ceiling on their career.

Obviously for business prodigies – and I’ll use an example of those as obedient young Microsoft types that may or may not be thinking for themselves – you end up with young people who are kicking ass on their careers early on, but then face a "prodigy midlife crisis" as their roles change. Rather than following instructions really well, the demands are different, and they can’t adapt.

This is all related to a question that’s often posed to me – sometimes, my mom or someone in the tech industy will ask, why don’t you go to X huge, high-brand company? Or, why don’t you get Y degree? I’ve come to believe, over time, that these things are fairly similar to being a violin virtuoso at a young age.

Working somewhere like Microsoft or Yahoo only proves that you can work at Microsoft or Yahoo. You learn skills and experiences which only apply to those companies, and are difficult to generalize to all companies. People who want to learn about startups should start companies (or, in a worst case scenario, be employed at a startup). Otherwise, you’ll hit a midlife crisis too, as you realize that your 8 years of experience don’t necessarily make you better than a fresh college grad at building the next Google.

These kinds of articles are always really interesting to me: The new economics of Counter-Strike.

One of the big problems for economics, as a science, is that it’s nearly impossible to follow the Scientific Method by tweaking variables and rerunning experiments. Instead, you can only look backwards, build models, watch future data, and confirm. The weakness there, of course, is that you can never truly make sure your measurements are independent.

But video games have a lot to teach economists – through artificial like Hollywood Stock Exchange, Second Life, etc., you can adjust rules, rerun experiments, and generally have a contained little space to see how humans interact with one another. Pretty damn interesting stuff.

The most famous study I’ve seen comes from the economy of Everquest. Article here, and paper is here.

You could say the same thing about entrepreneurs and other creativity-intensive professions: 10 reasons you shouldn’t go to film school.

This made me laugh: Four Simple Steps to Becoming a Billionaire.

As many folks in usability / user experience know, don’t listen to self-reporting. People are very bad at explaining why they did something, and instead, you have to watch them objectively and study their decision-making patterns to reach any conclusions. That’s why it’s silly to ask customers what features they want and develop them specifically, without putting it through a filter of customer motivation and context.

Related to the idea that people can guarantee a string of successes is a recent blog called the Myth of the Serial Entrepreneur. How many are really out there? And how good are they compared to a layman? After all, in a hit-driven industry, it just matters that you hit it once. Plus, the difference between an "expert" entrepreneur and an amateur one might be 0.1% success rate versus a 0.2%, with the rest of it being attributed to Luck. (I’m exaggerating to make a point – proven entrepreneurs are probably several thousand times better than average entrepreneurs, as you see in other expert performance studies)

That said, whereas you can’t guarantee you’ll succeed at a billionaire level, you CAN guarantee that you won’t. I have many friends who are very smart, motivated, and capable people.
In fact, they are probably no less capable than the Mark Zuckerbergs of the world – that’s why you always see posts on that Internet that say, "Oh, I could have done that!" But of course you could have, but you didn’t. And that’s what counts.

Whereas
successful entrepreneurs took a shot with 1 in 1,000,000,000 odds, many
of my friends are stuck at Microsoft enjoying safe, reasonable
lifestyles with 0 for 1,000,000,000 odds. So there’s no possible upside
beyond their salaries. So experts might get 0.2%, amateurs might get 0.1%, but Microsofties get 0% :)

Anyway, my business plan after reading that article is to illegally acquire some energy assets in Russia and kill a bunch of people. Now I just need to find a VC to fund that…

I was having dinner with my friend Andy from college today. Andy, by the way, is an improv theater guy who also majored in Applied Math, making him the funniest math nerd I know :) Anyway, we start talking about Moneyball and quantitative approaches to traditionally qualitative things, and how sciences grow out of mythological approaches.

For the most obvious example, take astronomy. At the very beginning, you had mythology instead of science. Eclipses happen from dragons eating the moon, earth is in the center of the universe, etc. Then, over time, patterns emerge – when the sun sets, or how stars appear, and then qualitative theories can be formed. For astronomy, this is probably when people were building Stonehenge and other things which showed understanding of patterns, but not the underlying reasons why the patterns occured.

But then, as people develop the instruments and expertise to begin measuring results consistently, then these qualitative models become quantitative in nature.  Scientific models form for why they exist. And eventually, this leads to a scientific process of hypothesis, testing, and verification.

Moneyball, for instance, is a story of how baseball made that transition. At first, I’m sure people played it without keeping long, historic, records. And even when they did keep records, a random pattern-seeking person might decide that a pitcher would win games if they threw really quick fastballs, or hitter was good just by hitting homeruns. What Billy Beane was able to do was to apply scientific rigor to the process, and then shape his environment by actually making decisions in accordance with his theories.

This discussion takes us to an interesting place – the jump from mythology to science is a hugely disruptive one. And in general, it may create tremendous opportunities to build companies to capture the inefficiency of the previously flawed foundation. You could argue, for example, that Zillow is making real estate into a more scientific process, rather than one based on conjecture and human-involvement. Does Zillow represent the Billy Beane of real estate?

I don’t know what other opportunities are out there for companies that are able to exploit this – but perhaps the place to look is to understand what kinds of new measurements are taking place? What kinds of new, common-place data is now being collected, and could you apply these assets towards a previously mythologically-driven problem to create efficient, data-driven ones?

UPDATE: Additional thought – one field I’m particularly excited about seeing evolve is the social sciences. For a long time, a lot of pattern-seeking models like Freud’s id/ego/superego might have been formed for pretty silly reasons – I remember thinking Freud was all BS when I took those psychology classes in college. Economics has its flaws too, which is why it’s interesting to see behavioral economics approaches grow and flourish. Both of these might be revolutionized by the advent of two major areas: Brain-scanning devices that measure neurological responses to stimuli, and the DNA-focused initiatives like the Brain Atlas, which may drive us to build a more scientific model of human behavior.

My co-worker Basem Nayfeh made an interesting observation on YouTube versus MySpace recently – he said that he sees the YouTube acquisition as much more justifiable than MySpace for one simple reason: YouTube is a clear distruptive threat to an industry that already represents billions of dollars.

His argument that is it’ll be relatively straightforward to put together the licensing deals (being the "pipes" in the content-versus-pipes dualism) and then start tapping into the 10s of billions of dollars that represent TV. By having a close analogy in place, they can go to ad agencies and pitch something they somewhat know and understand, rather than being a purely evangelistic sale.

On the other side, MySpace is not a clear replacement for something that clearly generates revenue today. Is it just communication? If so, isn’t it just disruptive of IM, e-mail, newsgroups, etc., all things that people never figured out how to monetize? If anything, you’d expect that it might be an indirect disruption towards ALL other internet sites. Because it’s a foundation for peoples’ Internet experiences (go to MySpace first, then figure out where to go), then one could package things like video, applications, retail, and other apps that are more closely tied to monetization.

John Battelle writes on Google’s recent acquisition: First Blush on GooTube.

I have to say, I’m pretty damn surprised at the whole thing, especially
how fast it went. But in particular, I would have guessed that it’d be
Viacom, Fox, Yahoo, MSN, or AOL that would acquire them, for reasons
that I’ve stated before. To summarize my previous analysis, to really
capture the opportunity, YouTube really needs a great brand sales
organization in order to sell pre-roll and/or other brand stuff that’s
integrated into the videos. Google has the weakest team out of all the
other portals/old media. Thus, I’d assume that the synergistic value would be a lot less than someone who thinks they can capture a lot of brand dollars.

That said, one analysis of this acquisition is that it’s just 1-2% of
their market cap, and the transaction was completely in stock, so they
can gain or lose more than $1.6B if YouTube changes hands. For example,
it might be that if Fox bought YouTube, thus gaining a virtual monopoly
on video on the Internet, that might have caused Google stock to drop
$1.6B in market cap. Or, alternatively, if they can show higher growth
by buying all this traffic (and keeping the TAC they’d otherwise pay
out) then it might cause investors to get even more excited next
quarter and the stock would gain the 1-2% to offset the acquisition
cost. Either way, it might make sense.

One interesting point in Battelle’s article is the description by David
Drummond of Google (their General Counsel) that the YouTube valuation
was calculated on a "synergistic model." It may be obvious to a lot of
people what that means, but let me describe it for everyone else.
Basically, the idea is that rather than calculating the value of a
company based on their current revenues and income (of which YouTube
probably had very little), instead, you ask the question: If we bought
this company, and integrated it completely into our operations, how
much could we gain overall? Basically, you take YouTube’s revenue
potential and then amplify it against Google scale, which leads to a
big number indeed!

An hypothetical example of this: Let’s pretend that Joe can come up with a recommendations algorithm that’s 2X better any existing ones, for movies, books, and so on. Let’s also pretend that he’s making $5MM/year running this service. Now, Amazon or Netflix might come along and want to buy them. But even though they’d look at the revenue, they certainly wouldn’t pretend that they were just acquiring an additional $5MM/yr revenue stream. Instead, they’d realize that they could apply the technology to their stores and make an additional $100MM/yr. That would be the "synergistic" part of their modeling, which might get them to pay Joe $200MM for his handy new technology. Obviously I’m simplifying things here, since there’s a build or buy decision, competitive issues, and so on and so forth, but that’s just an example.

This is the kind of approach that ends up justifying
valuations like Skype as well – eBay calculated the strategic leverage
that the platform could bring, rather than a valuation based on
incremental revenue gains. So in the long run, who knows if the $2.5B price tag for Skype of the $1.6B value for YouTube are worth it – they might well be…

This is the most exuberant article I’ve read so far about YouTube: YouTube IS Wildly Profitable – No Doubts About It.

Particularly this quote:

If the site serves 100
million video streams a day, then it generates 100 million pageviews a
day.  Judging by the site, it serves 1 ad impression per page, so it
can estimated that it serves up 100 million ad impressions per page.
Looking at the kind of clients I see, and the CPM these garner, it is
reasonable that YouTube can make – every day – anywhere from $50K to
$200K (assuming $0.50 to $2 CPM).  Realistically, I would bet their ads
yield $0.75 CPM, or roughly $75,000, but of course, who am I?

This really reminds me of so-called Chinese math, the type that venture investors often see. You know, the "if we could get 2% of the total Chinese market, then we’d have billions!" top-down modeling that goes on.

Now, what *is* true is that YouTube makes $0.75 CPM on some of their traffic. The problem is, at the traffic levels that YouTube puts out, they will also get a lot of 7 cent traffic. Why is that? Well, the advertising ecosystem looks like a pyramid. At the top, you have a small number of advertisers (typically premium brand guys) that are willing to pay $15 for good space (like on their homepage) and they are willing to buy X impressions. At the next level, you have other brand guys that will pay $5, but they are only willing to buy Y impressions. This goes down a couple more levels. Then, there’s a huge amount of inventory to be had at $1, which usually go to ad networks, and then at the very bottom, there lots of impressions, in the hundreds of millions for sites like YouTube, which monetize at pennies per thousand. Each "layer" of the pyramid is structured based on CPM, and things like frequency caps that the advertisers impose. Once the publisher gets big enough, they start busting through the layers, hitting diminishing returns at each level.

If you’re a small site, you can actually get a high overall CPM since you can directly sell to endemic advertisers at $15. But the larger you get, the more "layers" of advertisers whose budgets you exhaust, until at the very end, you have a couple billion ad impressions that you hand over to CPA ad networks.

So for YouTube, the author is correct that they probably get $175,000 for the homepage. But do they get that every day, at that CPM? Maybe, but maybe not. And he’s right that YouTube sells a bunch of other inventory $0.75. But all of it? Absolutely not. Instead, the calculation is much more complex, and wholly dependent on how many premium advertisers they can get to make up the top of their advertising pyramid.

All in all, I would guess that YouTube monetizes, on average, much closer to 10 or 15 cents overall CPM. You’d want stats around user impression frequency, what percentage of their site is ad networks versus direct selling, etc., to really model it out. But my guess is, based on experience with other ad networks, that they’d be lucky to get it consistently into the double digit CPMs.

These are some of the unique dynamics of very large publishers, but also specifically of user-generated content sites which have special characteristics like huge user frequency (100+/day for some), lots of homogenous inventory, and poor user intent. If you don’t factor the above points into your calculation, you’ll be trying to sell BMWs into China expecting a market of 1.2 billion people without realizing that over 700 million of those people make $10/month.

Ars Technica article on a fascinating audience trend: Teens – E-mail is for old people.

It’s something that’s come out in interviews I’ve had with people that are 19 or younger. Rather than using e-mail, instead they use text messaging or IM or MySpace. For someone who’s an e-mail/Blackberry addict, this sounds pretty strange. But the reasons underlying the shift make sense:

1) E-mail is hard to use
Although it seems trivial to the rest of us, e-mail is actually reasonably hard to use. Does your friend use andrew1923@hotmail.com or andrew1293@hotmail.com? I forget! The last thing you want to do is memorize another arbitrary string to get ahold of people. Although address books can help, in practice, it’s slower and less convenient than other types. In our interviews with younger folks (again, 19, 20 or younger), people simply didn’t know the e-mails of their friends. Makes me wonder if all those "E-mail to a friend" features will become worthless over time.

2) I want it now now now!
It amuses me that a century ago, the only real way to communicate long distance was by letter, over multiple weeks. Then you had the telegraph. And then the phone. And now everyone has a cell phone – in fact, I remember seeing a little girl about 10 years old riding a bike here in Bellevue (an upscale suburb of Seattle), gabbing excitedly while she rode down the street. It’s as if she was practicing to be a future soccer mom in a Humvee, also gabbing on her cell phone. Either way, all the types of media and interactive communications tools make consumers more and more impatient.

E-mail has a great characteristic, at least for me: It’s asynchronous. That means, for introverted extroverts like me, if I feel like just relaxing for a bit, I can. But for other audience groups, it’s really not that interesting.

3) MySpace is the replacement
One often-known fact about MySpace is that the vast majority of pageviews for the site, other than profile views is checking your messages. In fact, if you ever hang around heavy MySpace users, you’ll hear them talk about how they "need" to check their MySpace. Some people check it hourly. Does that remind you of e-mail? Well, it is – that’s exactly how they use it. But rather than memorizing random addresses, instead, they can click on the picture of their friend (which is already in an "address book" as one of their Top 8) and easily send a message.

I’m curious, though, if this trends holds over time. After all, I don’t see Corporate America moving off of e-mail and e-mail addresses anytime soon. I don’t think Microsoft will have a big MySpace page where all the employees can look each other up (Although Fox Interactive employees might). Furthermore, universities still expose their students to e-mail. So the question is, for those students that go into college and then join the workforce, maybe they will go through an indoctrination and end up on e-mail just like before. Only time will tell.

Saw this on TechCrunch today: Controversial PayPerPost Raises $3 million. I haven’t used the service, but it’s interesting thing to how they are approaching creating a new marketplace.

Building a marketplace is hard. Not only do you have to get the chickens and the eggs to all line up the right way, you have to make sure everyone trusts each other, and also, that everyone agrees what they are buying and selling. This is defined as the "ad unit."

In PayPerPost’s case, they’ve decided that the most factors that define their ad unit are:

• Price
• Minimum words
• Dates
• Tone

(This is from the screenshot on the TechCrunch post)
In analyzing this approach, you have to ask yourself, first of all, is this going to be branding or direct response focused? You don’t really have to pick one or the other (for example, TV has branding and DR infomercials), but it’s a useful exercise to understand the primary benefits of the medium.

You could argue that it’s really part of some viral branding push. This would be a fairly hard play, for a number of reasons, but if they wanted to focus on that area, they’d need more "brand" stats. For example, they’d want the number of unique users that are reading the plug, what their audience composition and demographics are (perhaps partnering with Nielsen or comScore), etc. Furthermore, they’d need some unique measurements around the "buzz" factor of a blog, perhaps based on Technorati’s in-bound link metrics or something similar.
Those are the changes they’d probably want to make on the product side. On the business side, they’d need to only deal with very reputable blogs (and would need to prove this), and create enough relationships that ad agencies can buy lots of reach. If you can only reach 1,000 people, it’s hardly worth getting out of bed. Furthermore, under the brand model, they’ll need a little more money, just to start a New York office with agency people. (I’ve written about this before)

A much more likely scenario is for them to focus on direct response. The advertisers for DR are much scrappier, will pretty much try anything, and will probably be the early adopters for that service. Furthermore, these guys are much more likely to want to buy links from random, unbranded sites, particularly if these sites can help their Google Pagerank. (This might be the real benefit of the service, actually) In that case, PayPerPost will be encouraged by their advertisers to understand what the CTR per blog are, how much click volume they can probably drive, and then, as much as possible, they will want to know the conversion rates. Ultimately, this will make PayPerPost use a mechanism similar to Commission Junction’s or Linkshare’s, where links have clickwrappers with affiliate IDs attached to them to track all this random information. The DR advertisers will then compare the performance of these new blog ad units with search, display, affiliate, and whatever else they are buying.

Either way, it seems like PayPerPost is already on a good track – it’ll be interesting to see which side of the tracks they land, within the advertiser world. The process of defining the ad unit is probably the most important set of decisions for the company, since it’ll make or break their marketplace. Another company that has a set of interesting decisions like this is NextMedium, which is building a marketplace for product placement. (That company is a great study of ad units too – it’s clearly branding, but what do you measure? Time on screen? A-list celebs? Dimensions? etc.)

My guess is that for PayPerPost, in the success case, it’ll be direct response at first, but as they grow and add money, they’ll be able to pitch this stuff as a large, viral, "brand integration"-type ad buy to agencies. Best of luck to them!

Click here to see a bunch: Lost Garden Blog.

I was surprised to hear recently that Seattle and LA are considered the big hotbeds for video game design. Cool :) I’ve always known that with Nintendo, XBox, Valve (which is located in the same building as me!), Popcap, and others, that we had a lot of talent here.

Anyway, recently I’ve become very interested in game design as a systematic approach to creating "fun" within applications. I’m mostly interested in how these design processes could be applied to websites in order to make them more fun. It’s interesting to me that after all these years of photo-sharing, you’d get a site like Flickr that’d come out and people would describe it as fun, or like a game, or whatever.

The game community’s emphasis on rapid prototyping, user testing for "fun," designing in mechanics like uncertainty, pacing, etc., all have a huge appeal for me. I think as the Internet moves from task-oriented websites to more and more entertainment, these mechanics will be very useful to understand.

Haha, I like it:

• IDF: Iraqi Defense Force
• ADF: Apple Defense Force
• NDF: Nintendo Defense Force

Source: Picture it: fanboys fight harder – Joystiq.

Greg, a Seattle tech blogger writes: Not walloped by Wallop.

My favorite quote:

First, I have to admit, I am in no way in their target demographic. I
am too old and boring, and my dating days are way behind me.

I have two random comments about the target demographic of Wallop.

First, one funny thing about the "target demographic" of Wallop, or at least, the demographic for which Wallop has been the most successful, is that much of the userbase is in Mainland China. I’m good friends with one of the backend developers that helped build Wallop, and when they released it, immediately a huge audience converged in China. (Who knows why?) Anyway, just as Orkut eventually gravitated towards Brazil, and Friendster is the most popular in the Phillipines, Wallop had a pretty good presence in China. Now, it’s been spun out now, so who knows if it was all technology or included that user base. They were able to build a large, 6-figure user base for a research project.

Second random comment, from the user interface perspective, Wallop absolutely is confusing. But so is MySpace, and Geocities before that. I think the reason why geeks like me and Greg typically react poorly to these cluttered, messy, non-standard interfaces is that we’re heavy internet users that notice little things. We care about the user interface paradigms that emerge, since it makes it easier to locate and find information. People like us know what Orange buttons that say XML or RSS on them really mean.

But the problem is, I don’t think the vast majority of users care. In fact, many of them like the clutter. Remember that for every ugly MySpace page out there, with flashing icons and moving photo galleries and pink on red text, someone out there spent HOURS customizing that page. I remember that after prototyping a site targeted at 16-24 year old girls involved in social shopping, and showing them mockups with extra cool Google-like whitespace, they said, "Yuck. Too boring." They wanted lots of colors and random stuff in the background.

Anyway, that’s just a long way of saying "I don’t get it either" :)

It’s a book about the evolution of football: Amazon.com: The Blind Side: Evolution of a Game: Books: Michael Lewis.

For those of you who haven’t read Michael Lewis’s books, they are amazing. I first found Lewis through his book Liar’s Poker, about Wall Street. But after also reading The New New Thing (about Silicon Valley and Jim Clark) and Moneyball (about Billy Beane and quantitative approaches to baseball), he’s become one of my favorite authors.

Great BusinessWeek article on Design Schools: The Talent Hunt.
It’s interesting to see a huge trend towards people-focused product creation
in Corporate America. I think people are really starting to see design
as a core aspect of their strategy, rather than some magic you add in the last minute. Within the technology world, this is probably driven by Google and Apple, who often have LESS functionality than their competitors, but still dominate because of a superior user experience.

Anyway, I got the chance to be introduced to Stanford’s d.school a couple years back when I attended a series of talks there. They have a great approach, that’s fully interdisciplinary. Now, if they’re able to launch a couple multi-billion dollar startups from the program like Stanford’s CSL, that’ll really put them on the map :)

Here’s a recent link on VideoEgg raising money and trying to start an ad network: GigaOM » VideoEgg Lays on Ad Network, Funding.

Following on the previous post about YouTube, it’s clear that there will be a footrace to create a video ad network. This is all well and good, but I hope these guys realize that they have one well-pocketed competitor that will probably crush them. (Well, maybe two well-pocketed competitors) For most people, it’ll be obvious who these competitors are.

Why are they at such a disadvantage? Well, the reason is that these guys are going to fundamentally try to compete on Brand advertising. Because of how these ads will be set up (pre-roll to user-generated content), there’s no reason people will click the ad to get off of the site. Also, because of the ad unit is video, and it’s expensive to develop effective creative, only deep-pocketed ad agencies will have the money to do it.

The funny thing about Brand advertising is that it’s very backwards to what you’d expect. Most rational nerd-types would approach it with, "Well, just prove that the branding works and you’re good, right?" But NO. That’s not how it works. Branding is just not ROI driven for most advertisers. (Whether or not it should be is another question) So because of that, and because effectiveness is difficult and expensive to measure, people just use TRUST and RELATIONSHIPS as proxies. Furthermore, rather than Google’s 200k advertisers, which look like the Yellow Pages, with lots of mom-and-pop stores, instead, all the dollars belong to 4 or 5 major brand agency conglomerates.

That’s right, it means that some of VideoEgg’s $12 million needs to go towards opening a big office in Midtown Manhattan and hiring some very expensive ad sales people. And it’ll be an uphill battle even then, because they have zero track record and no brand. Why would the advertising agencies, which act as gatekeepers for ad dollars, go with something as unsafe as VideoEgg when they can put their ads on Yahoo, ESPN, CNN, and all the other major outlets? It will take a long, long time for them to develop enough track record to get connected into the insular world of NYC brand advertising.

So who has the best chance to develop a video ad network? Well, what companies have the largest online ad sales forces today? And who could just sell these pre-roll ads as part of a larger media bundle, which is always easier. Basically, you sell a $500k ad deal across your huge brand network, and 10% of that is line-itemed towards video without anyone really noticing. That’s the best way to get started.

In my opinion, Yahoo and Microsoft have the best chances to create a video ad network. They have huge brand ad sales teams, already take video input, and have the pockets to go buy a bunch of video inventory by acquiring one of these video-sharing companies. They just have to bundle this as part of a larger deal, tap their New York relationships, and it’s all good.

Of the startups, the first one to take a strategic round from one of the major ad agency conglomerates will have the best chance. (Think Siebel and Accenture’s equity relationship for a background on that) Or, they need to open big NYC offices and hire very expensive (and in-demand) advertising guys. Otherwise, if these video companies don’t get themselves deep into the ass-backwards world of NYC brand advertising, they will be screwed.

(IMHO)

Growing up in the time of the dot com bubble was a surreal thing. In 1999, I was 16  or 17 years old, and a junior at the University of Washington. I had started using the Internet early – as a progression from playing with BBSes to using apps like gopher and archie/veronica at the public libraries. Anyway, it seemed like they were handing out money left and right in Silicon Valley, and I figured that my friends and I were pretty much as smart as everyone else – so why not start our own website?

But first, we needed an idea. It turned out that I was about to rent an apartment with one of my best friends, Jake Kreuzter, which turned out to be an enormously annoying experience. We ended up wandering around the UW, calling phone numbers as we walked around, because there was no centralized place to find apartments. There must be an easier way!

So we decided to start Local-Rent, a website where local landlords could post their apartments and people could go rent it. After coming up with this idea, we recruited a couple other friends from the Early Entrance Program that were really smart and reasonably technical. This included our good friends John Richmond, Thor Sletten, and a bunch of other people. This larger group also entertained other random ideas beyond Local-Rent.

This large group, however, did not go well. It wasn’t clear what people were going to do, and slowly but surely, people got bored and it was left to 3 of us. (Jake, John, and myself)

We started building the site, but went way overboard. This, in particular, was my fault. (But remember, I was 16 at the time, so sue me) We started coded stuff up in PHP and MySQL, but we came up with a massive schema and lots of features for the website. We had tables for tenants, landlords, apartment complexes, individual apartments, transactions, rental agreements, the whole thing. Note that this was all speculative, since we really didn’t understand too much about the renting business anyway, and we also completely focused on the technology.

One thing led to another, and we ended up having a really complex application – in fact, in the living room we had put up a bunch of Post-Its representing the structure of the app, and it was huge. Furthermore, we started by coding up layers within the application. So rather than working on the user interface and all of that, instead, John and I ended up building an ORM (object relational mapper) for database rows to get instantiated as PHP objects. As an aside: It was a neat experience, from a software engineering point of view, since we didn’t know ORMs existed as a general concept and it just made sense to us. So now, to see stuff like Hibernate and ActiveRecord is pretty cool.

Anyway, after a couple months of working very hard and getting no where other than the backend infrastructure, we gave up. The feature list seemed huge, and it seemed like we’d never get to the end.

In retrospect, it was a fantastic experience. It was great to understand some of the tendencies that you can get wrapped into – and the entire engineer-as-entrepreneur thing is especially hard because it’s easy to get sidetracked by technology rather than users. So even though it was a colossal failure, I’m happy I wasted a summer doing this when I was 17. It was fun in itself, and I became a better person for it.

In the end, we never finished the site, but years later, it would become very exciting to see Craigslist succeed. Although our idea was really a subset of what Craigslist has become, it makes me happy to see a community, locally-focused classified section where random people post sublets, residential housing, and other rental opportunities. It somewhat validates the thinking of a couple eager teenagers with PHP h@x0ring skillz :)

All in all, I learned a bunch of really important ways, by screwing up:

1) It’s easy to bite off more than you can chew
When we first started, we defined way too much stuff to do. It was fun to dream up the "next big thing," and people with big vision also tend to go overboard with this. But the practical reality is, defining something really big can be demoralizing. It takes too long to build and see results, and you take on a ton of risk by making your invention a big, high-stakes bet instead of a bunch of little bets.

Instead, the right thing to do would probably be to define the vision, make that your North Star, but then step back. You’d look at all the little micro-releases you could do to build up to your North Star, and start to go from there. If you have to integrate some things, do it incrementally, and not all at once at the end.

For Local-Rent, it would have been better to make a little forum-like application for people to post generic listings right away, and try and reach out to rental people. Probably the project would have gone to the next level right away, as we started thinking about the major business hurdles to seed this kind of marketplace.

I see people making this type of mistake even now – you’ll hear startup people that need to be in stealth for one or two years for a consumer media application. It’s just silly. Launch your site already!

2) It’s easy to chase shiny-technology things, if you’re a nerd
This applies to me, but maybe not to everyone. It’s often tempting to use each startup opportunity to learn something new, and thus, take on additional technical risk. In our case, we really, really didn’t need an ORM to make the business work. So why did we work on it? Because it was fun :) Thats a good reason if the goal of the project is to play with technology. But it’s a bad one of you actually want to get stuff done.

Hilariously, I still find myself making this mistake from time to time now. The fundametnal aspect of it is, I do these types of startup projects to learn new technical things anyway. That’s an integral part of the goal. But getting too distracted is bad, and I can catch myself easily from that standpoint.

3) More people at the beginning isn’t better
Although this didn’t create too much of a problem in the long run, it seemed like a really good idea to bring in lots of smart people early on. We brought in friends just because we thought they could contribute one way or another. Frankly, it’s just distracting. You end up creating busywork that doesn’t need to exist in the first place, just to give people stuff to do.

The group eventually evolved into a mix that did balance out. You want everyone to be able to contribute very concretely in the beginning. So no finance people or business people that just muck around with PowerPoint and Excel. That is 1/2 of one fo the guys’ jobs. Instead, you grab one or two engineers that are really good at shipping stuff really fast, and you have one business-y technical guy. (In recent cases, I’m the latter). That way, you can focus on making things rather than talking about making things.

4) Focus on the customer (and make it concrete)
Another symptom of playing around with technology instead of focusing on the business was not talking to any customers. Although Jake and I, as renters, were part of the consumer market, we really should have spent a lot of time with the other side of the equation (homeowners) as well as potential partners (classifieds, newspapers, etc.).

The right way to go about this is to be very specific about your target audience, and vet ideas and prototypes with them first. I’m a huge fan of thinking through and incorporating user personas into your design process. If you and your developers are arguing about whether or not "Landlord Larry" really uses the site in X or Y way, you’re on the right track.

5) Failing ain’t so bad
And finally, as I mentioned above, trying this project and then giving up was a rewarding experience. I learned about lots of stuff that didn’t work, which then implied some hypotheses around what could work. And although your ego can take a hit, at the end of the day, you’ll forget it and be happy that you learned so much.

As soon as Local-Rent was over, it only took another year to dream up something else and try it. This time, we had a lot more success getting it off the ground. I’ll write more about that project some other time.

After several random, semi-failed attempts at creating consumer websites, I’ve come to a new appreciation for companies like IDEO, and industries like video games, for pioneering rapid prototying techniques. By failing faster and more often, they are able to converge to a good solution faster than a slower, more deliberate method. (Think of them as two different search algorithms for finding web concepts that engage people)

A long, long time ago, I thought that starting a business was about The Business Plan ™. That was the master document of all the things you wanted to do, the markets you wanted to attack, and the means for how you were going to do it. After very deterministically planning with god-like omniscience, you’d have a tome of Truth that no one could challenge.

Years later, I would find that a very silly approach – I’ll write more about that later.

In the meantime, I wanted to share one of my favorite rapid prototyping articles, coming from the games industry. It’s a bunch of pages – just keep clicking until you reach the end.

Link: Gamasutra – Feature – "How to Prototype a Game in Under 7 Days".

<puts on flamesuit>

Living a mere 10 minutes from Redmond, WA, I’ve come to know lots of random Microsoft people over the years. Lots of my friends from college went there. My ex-girlfriend is there. In fact, in the last month, my sister Ada graduated from UPenn and joined MSN Search. (Congratulations, I’m sorry)

One pattern that I’ve started to pick up over the years is hardcore Microsofties tend to train their brains in a way that makes them eventually unfit for startups. (But who cares, they never wanted to work for one anyway)

It all comes down to how their business is organized, and the audiences their products target. I love it when Microsofties use the word "platform." That’s how they think of everything. Windows is a platform. Office is a platform. MSN is a platform. But think about what that implies – a platform doesn’t solve any problems for anyone. You need stuff that works with the platform to do anything productive. But it compells you to make things broad and generic, while requiring you to develop a huge ecosystem of vendors and partners to execute.

While this works when you have a monopoly, it certainly doesn’t work when you are a puny startup (even one with venture capital). No matter how much money you spend, it’s impossible for everyone to suddenly jump onto your ecosystem. Instead, it makes far more sense to go the opposite direction – specific instead of generic, and problem-solving instead of platform-building.

That also applies to your audience. Obviously, if you target a huge, nebulous audience, you are going to get in trouble (unless you can bundle your app into, say, Windows). It’s very very hard to find a killer-app for anything but a focused audience. So study that group, solve their problem very well, and build out from there.

So if you are just creating a new startup, and your smart ex-Microsoftie says any of the following, you are suffering Platformitis:

"We should build a big suite of all of this stuff!"
"Let’s be the portal of X!"
"If we code X (some cool thing) and launch an API…"
"Let’s expand into these 5 verticals!"
"We should build a platform for X"

If they say that, you are in a lot of trouble.

Instead, embrace your constraints as part of a startup. Pick one target audience. Find out what they are really passionate about. Pick one thing. Build it really fast, and see if they are as excited as you are. Rinse and repeat, until you have a great user experience. And once you are a bigger, uglier, slower company, and have couple successful business lines, then by all means diversify. You might even want to build a portal, or a suite, or a platform.

But when you start out, and you’re in the beginning stages of your startup, focus is everything. "Crossing the Chasm" is really about the same thing, in a specific B2B context. So be a man, make your decision already :)

My buddy Eric Mattson asks my opinion on YouTube:

Andrew,
Do you really view things like YouTube as sustainable businesses?
Bandwidth is cheap. Storage is cheap. Why do people need YouTube?
Eric

A couple weeks back, I had a conversation with my friends about consumer early adopters and was wondering how to define them. Although they’re described with some detail in "Crossing the Chasm," that group seemed to be more B2B focused rather than consumer-oriented.

Anyway, we discussed a couple different variations of this, and came to a couple interesting conclusions. Pretty obvious, in retrospect, but still interesting to formalize.

Characteristics of Andrew as an early adopter of technology/websites:
    Like to read about new things
    Like to try out new things for pleasure
    Am confident in my ability to spot the importance of new trends and products
    Develops an awareness of all possible options and how they relate to each other
        How else would you know whether one thing is better than the other?
    Takes pride in the expertise of random subjects, and have interest in offering opinions
    May be a manifestation of competitiveness in out-knowing other people in random areas
    Similar to the snobbery of people that have indie/obscure movie or music expertise

Characteristics of John as an early adopter of media:
    Being careful to spot problems and on the lookout to solve them (not actively looking, but passively absorb)
    Saw Firefly ads, taped the TV show, then if writing/characters are good, then develop the motivation to know everything about the show
        Then follow the creator of the show
        Sometimes, try out new shows and then won’t follow it if it sucks (like Enterprise, etc.)
    Found Veronica Mars through TelevisionWithoutPity (recaps of TV shows), and the reviews seemed to be good, then developed a stronger motivation
    Goal-oriented TV watching, unlikely to randomly find shows
    With sci-fi books, read the books that known authors have read
        Or, read books that the known good authors like
    Randomly look at sci-fi books at Barnes & Noble, then grab a book based on the cover
        Motivated to do this so that there’s something to read – if there is a book queue, then not necessary to randomly look for books
    Is influenced by secondary reviewers and other sources

Characteristics of Craig as an early adopter of gadgets:
    Hear about new devices then research new features, reasons to buy it, etc.
        Hear about e-mail, newsletter from a carrier, word of mouth
        Then search online to get more information
        Then go to a store and ask about the device
    For new features, just looking for differences, nothing specific – know about things he doesn’t care about
    Gets bored with current gadget, and then starts looking for new things
    When the product sucks, then he gets annoyed – or when a trial goes bad, he feels bad, then move on
    Short attention spans with phones
    Likes to show gadgets to friends and "trend set" with friends – whether or not it’s bad or good
        Talk to friends about it because it might benefit them to know certain things
        Will have different levels of discussion depending on the audience
        Because he has the expertise related to the product, he can influence and offer personal experience
    "Show my friends that I’m a baller" – maybe a status thing, or show of importance
    Have to possess things that other people don’t have – now that Blackberries are common, need to get the newer Blackberry

—

Common characteristics?
    Always on "browse" mode, keeping the mind open for new good things
    Short attention span when browsing for new things, going for volume
    Low cost for "failure" – willing to try out new things
    Able to form an opinion or judgement about the product independently of other people
    Once you’ve find something good, continue push on that button to find additional good results

A couple weeks ago, I wrote a bunch of notes on the Internet advertising market. Specifically, it seemed like there were 4 major categories for creating value on the web:

1) Make existing high-value inventory even more valuable

Rich media, video ads, and “profile” targeting fall into this bucket, since it mostly has to do with brand advertising. As we discussed, there’s a flight to quality for branding, since advertisers are just dipping their toes in online right now. These “known good” publishers are typically very entrenched, have invested millions of dollars into NYC relationships, and generally have ties into old media. Although there are some plays here to help advertiser buy reach more effectively, it’s maybe that we’ve hit diminishing returns already.

2) Make existing low-value inventory more valuable

More likely, advertiser spend will go towards “gray area” inventory that turns to good inventory through some magical process.

Just as you mentioned the “media gap” between Internet and Internet ad spend, there are many areas of the web that gets lots of attention from users but don’t monetize well. The “usage gap” maybe? Included in these are casual games (Pogo, Miniclip, etc.), blogs (LiveJournal, Xanga, etc), social networks, media-sharing sites (Photobucket, YouTube, etc.). There are lots of ideas tactically on how this could work, which we can talk about at a future conversation.

If a company can help monetize some of these categories, it would be akin to Google/Overture figuring out how to monetize search better. (Which used to be a poorly monetizing category, http://bnoopy.typepad.com/bnoopy/2005/03/index.html)

Revenue Science generally falls in this category, BTW…

3) Create ad impressions where none currently exist

This may be intertwined with the above (2) point, but obviously for a long time, search engines just had very few ads, but now there are lots of text ads. So what are other areas which are creating a lot of pageviews, but not ad impressions? Ads integrated into flash applications like casual games? In-game ads like Massive? Watermarked ads in photo-hosting sites? 15-second frontend clips shown before YouTube ads? Product placement in blog entries? Social networking profile widgets? Audio ads before Skype calls?

Obviously another way to go about this, rather than being an intermediary, is to actually collect the content yourself and then create an ad network on top of this. Google did this, and YouTube seems to be doing the same.

4) Use the Internet to revolutionize off-line media

And finally, obviously companies like SpotRunner, dMarc, NextMedium, etc., are using the Internet’s strengths in aggregating constituents and automating processes to create transactions in TV, radio, and movies, respectively. Pay-per-call is another big area here, as well. This area is not a 100% online pure-play, but obviously could be quite interesting since it taps into huge, established budgets, and could force old-media to evolve rather than be up-ended completely.

I don’t remember being a nerd before 5th grade.

In 5th grade though, I had a teacher that would change my life forever, and set me on my path to nerdiness. His name was Kent Daniels, and he was my 5th grade math teacher. Now, Kent was no ordinary teacher. His class had a kumbaya-like setup, with a circle of big, cushy, used sofas where kids would sit down and talk. Furthermore, he was able to somehow amass a bunch of old Macs, probably 1 for every kid. (In 1992, this was a big deal)

He assigned us a project on probability, where we rolled 2 die, added them up, and repeated. We kept track of the totals and how often we rolled them. Of course, over time, the totals would converge into particular percentages, which indicated their probabilities. For example, there’s a 1/36 chance to roll a 2, 2/36 chance to roll a 3, and so on and so forth.

Even though he had all the kids roll the dice a 100 times or so, he also challenged us to write a computer program to do it. Kent said that he had a Mac Plus in the back room that we could use to run it for a couple weeks, and simulate rolling 2 die a couple million times. This was a tall order for a bunch of 10 and 11 year olds.

Of course, my friend Kelsey Woods (who already knew a little BASIC) and I decided we wanted to do it. Even though I knew nothing about computers, he started teaching me what he knew about BASIC, and I started learning on my own. I still remember the exhilaration of watching a screen full of random numbers when we figured out how to seed the generator.

Weeks after the original assignment, we had a beautiful little program in BASIC that generated the numbers, calculated the totals, and even drew little graphs with *s :)

In a bit of fanfare, we loaded the program onto Kent’s Mac Plus in front of the entire class. We started the program, and left it. Everyone clapped.

Minutes later, smoke started emerging from the computer. Kent looked over, grabbed the power cord, and pulled it out of the electrical socket. He quipped, "I don’t think the computer liked your program." Haha.

I never figured out what happened to the computer, and why it started to smoke. Years later, I heard that Kent had left the Seattle Public School district after a dispute with the administration. It made me sad to hear that. I haven’t heard from him, or my friend Kelsey in years, but I’ll always remember them and the smoking Mac Plus.

All right! I finally decided to make a blog to collect all the random e-mails and notes I send to friends. Anyway, here’s the first post out of many.