y? y not!

Attended Y Combinator's demo day yesterday. It was a hoot, tons of bright, energetic kids doing their first real pitches to a pretty supportive crowd. I had to leave before the end, so I only got to see some of the pitches and missed some I wanted to see. A few broad trends:

• Do anything to capture usage data. Several different groups were using different hooks to get you to install tool bars, applications, or use bookmarklets to better track web browsing. As next generations search heats up, building large corpora of browsing, buying, and social habits will become increasingly regular parts of business plans. Alas, not one of the projects I saw made privacy a serious part of the pitch, which will lead to the kind of problems Ed Felten has been writing about.
• An unfortunate meme is propagating that "productivity" can be determined by looking at how many hours you spend using various applications. This is bunk on so many levels. Certainly, a 10 hour-a-day habit of browsing ferretbondage.com is unlikely to make you a workplace star, claiming that we know it won't is foolish. Richard Feynman worked on problems that stumped him at strip clubs. 8 hours of email use in one day could either result in zero productivity (you were coordinating timing with the Emperors Club), massive global increases in productivity (you are program managing three critical, distributed projects), or anything in between. Now, much like counting calories, getting more information is good, but perpetuating silly ideas -- and worse, trying to monetize those ideas by selling it to managers to "monitor" their teams -- is a very pointy haired choice.
• Many of the Y Combinator companies look more like features than companies, but I'm OK with this for a couple of reasons. First, I'd take a good collection of features looking for business and product people to partner with over the reverse any day. Second, if you look at Y as a talent scout, it suddenly seems incredibly valuable. These teams proved they could execute on good ideas under pressure, which is incredibly valuable, especially when combined with the networks they're getting exposed to. Not sure what this does to Y's business model if a high percentage of teams get scarfed into other companies, but as someone hiring in the Bay Area, I love the fact that these teams are moving out here.
• Although it had nothing to do with demo day, I had a chance to talk to Trevor Blackwell and see what Anybots is up to. Very exciting stuff! The science fiction convergence of augmented reality, telepresence, and virtual worlds is not as far away as you think.
  

So, on to the projects (in the order they presented):

• Omnisio. I am torn. Omnisio is almost exactly a small side project I was spinning up my flash knowledge to write for fun. Video markup, sharing, and annotation done right. Cool features like being able to sync a powerpoint presentation to the video, tag sections of video, share video with social comments, and compilation clips. All of this adds searchable data to videos, which is interested. Not a business yet, but a great collection of features. I love this idea and think we'll see more uses they haven't thought of.
• AddHer. Hot or Not with automatic link exchange and easy publishing to social networking sites. This is all about trying to drive more profile traffic and time will tell how valuable that will be. Good statistics interface, although user experience -- even in the demo -- was rough and had many extra clicks.
• Snaptalent. Google meets Monster. A targeted ad network for job postings, using all kinds of neat tricks to achieve better targeting -- "Oh, you're coming from Google's IP range? Want to work at Facebook?" -- and a better user experience. Ads expand -- go DOM manipulation -- into a picture and movie covered page that allows a candidate to submit a resume, visit your site, or just learn more. All three options are instrumented, so tons of user data is generated and collected. Easy tools for building the ad, too. Snaptalent has the distinction of being the first Y company to be profitable before demo day!
• Rescuetime. First of two "productivity == more time focused on visual studio" plays. Nice UI, cool graphs, and if lots of people use it I suspect they'll generate very valuable collections of user data, but marketing this is managers bugs me. I love the idea of bringing social networking news feeds to team collaboration, but showing programmers that they are below average for the week in terms of time spend with an editor open is not going to make your team more productive. First off, half your team will be below average for hours. Second, measuring time typing is like judging code by the number of lines produced. Third, I already talked about the dangers of too long a work week, so positively reinforcing what is already a destructive trend is a mistake.
• MightyQuiz. Just go click on it. Come back in a few hours. User-generated trivia game with very nice social networking tools for tuning questions, easy question generation, easy syndication, and done by Harvard classmates of the Windward Mark team I acquired when at Linden. Cool folks who presented lots of compelling user data. 94% of people hitting the site answer at least one question, average is 19 and 8 minutes per day. Tons of metadata being generated.

• TipJoy. I've already written about them, but TipJoy seems to have improved quite a bit in the last few months. I love their argument -- that micropayment attempts in the past failed by focusing on the payment part -- although if they get traction fraud may destroy some of their secondary positives (such as the Digg-like ranking on the home page). Good points about usability -- PayPal takes 8 steps, Amazon 7, to pay someone. 200,000 impressions per day at this point. Not at all sure there is a business here, but their usability lessons are worth remembering.
• Mixwit. Slide plus iTunes. Collect online resources of photos, music, and movies and build widgets to share those collections on social networking sites. Widgets for programmableweb.com. Very early, but the widget they did for a mix tape looks very polished. 47% of people who start a mix tape publish it.
• Wundrbar. ZOMG, it's a COMMAND LINE for the Intraweb! Yeah, you think I'm joking. It's a command line. The URI to list commands is "http://wundrbar.com/command/ls" Command. Line. Google could add these features in, oh, about 8 seconds. Now, to be fair, it's an interesting use of javascript and one that could probably be taken further, plus the referral business could be decent if they get a lot of use. But it's still a command line. I'm waiting for the GUI interface to create text for wundrbar. Then we'll have Windows for the Intraweb!
• 8aweek. Another "if we just know where you surf you'll be more productive" product. I know, I shouldn't be harshing. A bunch of very heavy handed tools -- like "no, you can't go back to Digg, biatch! (click here to be lame and go anyway)" Polished, earnest presentation. If they collect data, the search implications will be very interesting, but again no mention of privacy issues. Plus, they generate great web analytics, which will also be valuable.

So, an interesting collection of bright people generating cool ideas and features. I don't know if Y is the right way to convert that energy into companies, products, and profits, but I'm glad they're doing it.

entrepreneurs and bloggers

There has been a recent spate of posts about startups, hiring, spending, and how you must -- or shouldn’t ever -- hire workaholics. Comment sections have been ablaze and everybody has an opinion. There is a great deal of certainty, although many opinions seem to be extrapolated from a sample size of 1 (and we all know how many lines can be drawn through a single point.)

Much of the advice is of the “buy low, sell hi”-variety. Obvious, known, and tautological:

• “Startups must hire the right people.”
• “Working with more interesting people is more interesting than just working.”

History is replete with examples of startups that were formed by whacky creative types, by armies following specific procedures, brilliant kids riffing off great ideas, mad geniuses, and everything in between. There are, in face, enough different and mutually contradictory ways of succeeding that selection bias allows one to defend almost any approach. Nearly every argument in all the blog posts can be supported or negated by examples.

Even though I’m a little late to the discussion, I do have some thoughts on this topic.

Rule #0: Have a Vision Driving Both Product and Company

Know what you are trying to do and why it's worth doing. Make sure the user experience flows from this vision. Duh.

But, remember Conway's Law.

Your organization's structure and culture will be reflected in your products. More than that, how your company operates will shape the possibility space for products you can create, will determine what you can create. So it is critical that Conway be extended. Vision must drive organizational structures capable of realizing the vision.

Rule #1: Don't Be Dogmatic

You won't always be right. Shocking, I know. So, be ready to change. Particularly when you deal with scaling -- whether in company size, code size, product complexity, number of customers, etc -- you need to assume you're wrong about something. After all, very few systems are scale free.

This rule may be the easiest to say but the hardest to follow. It is an example where how your startup is organized, how you work and collaborate, will define what you are capable of.

Rule #2: Tired (and Stressed) Employees Are Stupid Employees

The various posters spent a ton of time arguing back and forth on workaholics while ignoring a great deal of research and evidence that long work hours hurt productivity and efficiency. Studies going back to Henry Ford’s production lines demonstrate again and again that excessive hours reduce productivity and increase turnover. Since nobody was arguing that lowering productivity would contribute to success, how can anyone argue in favor of extended sleep deprivation?

Worse, for programming – especially in the knuckle dragging languages that I’ve spent my life using – the most subtle and difficult bugs to track down are usually memory related. Fatigue mirrors the impact of alcohol consumption and reduces peoples’ ability to assess their own competence. So, tired programmers are like having drunk, overconfident programmers. How much time do you lose to the memory leaks they introduce?

That’s before the costs related to increased turnover.

Now, there are qualifiers. There are lots of times when an all-nighter – whether alone or with the whole gang – can be a productive, bonding, or necessary event. A few weeks of intense effort or a weekend may be what gets you over the hump, but months or years of a culture that demands 60, 70, or 80-hour work weeks will destroy your productivity.

And cause other problems - just ask EA.

Rule #3: Know How to Do Math

This applies to a lot of areas.

Consider financing. At every startup I’ve been involved with, salaries have been the primary cost driver. Burdened programmer salaries in the Bay Area can easily hit $150,000 per year, so almost any discretionary spending will be less than 5% of the salary cost. Sure, it is possible to buy hand-stitched, gold plated toilet tissue and to burn $20 bills in a woodstove to heat your office, but my experience is that you really have to work to waste enough money to notice, especially if spending is transparent so everyone knows about your payments to Emperors Club VIP. The risk of an early employee abusing the company Visa card is way less than the lost time to someone not being able to buy something they really need when they need it.

Worse, refusing to buy a second monitor, laptop, extra computer, comfortable chair, free soda, beef jerky, or whatever it is that is keeping your key employees happy can easily reduce their efficiency by more than 5%. So be smart and do the math.

Again there are caveats. If you are scraping to get a better demo before a funding round, in a dry spell, or whatever, then change the rules. Just be honest about what’s going on.

Math also helps refute correlation-causation errors, which can be incredibly useful when trying to balance vision and change. Demand data and then analyze it honestly.

Rule #4: Have Fun

You’re going to spend a ton of time and energy on your startup, no matter how carefully you plan your time, how much jerky you buy, and how carefully you plan. So make sure the time in the office is fun for you, your coworkers, and your employees. Every company – everyone – is going to define fun differently, but figure out what it is for your vision and culture, then work to hold on to it. If growth means that the old definition no longer applies, spend the cycles to find a new one.

If you find yourself not wanting to come into work on Monday, you aren’t having fun. Figure out the problem and fix it.

But, remember that nothing is more fun than succeeding, so make sure your definition starts there.

tipjoy is pretty cool

One can make a compelling argument that much of the growth in blogging and experimental web ideas is tied to Google's Adsense and related products. By setting efficiently creating a 4-way transaction between the site creator, site visitor, advertiser, and Google, Adsense created a microtransaction system that worked. I know, you're now thinking "Microtransactions? I thought that was when I would pay the site creator 1/10 of a cent for their content. This is advertising!" It doesn't look like a microtransaction system.

But it is. And unlike the various attempts to create microtransactions in the past, it worked. See, the problem with microtransactions was that even when the amount of money being paid was trivial, the number of steps required to make a micropayment still killed the procehttp://www.blogger.com/img/gl.link.gifss. I would happily pay a few pennies to xkcd, but am I willing to do the work to keep my account full, click "I Approve" every time, etc. There were a lot of attempts but nobody seemed to overcome the user experience challenges. Adsense succeeded by piggybacking on something people wanted -- better targeted ads -- and making the payment transparent -- Google splits the payment from the advertiser with the site creator.

The results make everyone happy. Visitors only click on things they want, site creators get subsidized, advertisers reach high quality prospects, and Google makes a zillion dollars. Only downer is that some people don't like ads and want something different. In the real world, this is the National Public Radio model. Online, we're back to microtransactions and one new company -- out of Y Combinator -- is Tipjoy.

Here's a Tipjoy button:



If you click it, it will ask for your email address to leave a tip. No checkout, no big approval process. Eventually, if you go to the Tipjoy site you will be able to create an account and fund your tips via Paypal. Sort of like Digg, but with the added filter of "is noting this site worth 10 cents?" If it takes off at all, it should generate some excellent search and ranking data.

The Tipjoy folks are even doing one other clever thing. To avoid being a money transfer service -- which no small startup wants to be -- your tips can either go to charity or into an Amazon gift card.

knowledge acquisition

Recently I was asked how I went about learning new things. The question was in the context of my usual role of "go figure this out" at Linden, where I tended to draw the short straw and thus ended up researching -- among other topics -- intellectual property law, economics, learning theory, and philanthropy. I created a specific approach that served me well.

1. Identify knowledgeable academics/experts in your field of interest. Games and virtual worlds make this is particularly easy, but a bit of time with Google and social networking and you can usual find a set.
2. Reach out to them, preferably via mutual acquaintances, but otherwise with a direct, brief email. Explain what you are working on and why you are trying to better understand their field. Ask them for relevant books or articles that would help a motivated lay person understand their domain of expertise.
3. Collect the list of reaching material from each source, then pick the 4 or 5 most popular and read them. More than that, study them, reading critically and for understanding. Take notes and generate questions from the readings.
4. When you are done, write back to the people you asked for help. Thank them again, explain that you have read some of their recommended materials, and layout the questions and comments your studying generated.

At that point, you will have demonstrated that you are serious about your research and are willing to think hard about the subject. For most academics, researchers, and experts, having somebody do that much work to learn about their field is a pleasant surprise and, in my experience, they are more often than not willing to help you out with follow on questions, as well as introduce you to students who might be good fits for internships or open job positions.

Plus, you got to read about a new subject and come away with your mind stretched a bit. It is a win-win! You never will know enough to solve all the challenges you face starting businesses and creating new product, so having a network of smart people outside your domain can be an wonderful asset. It certainly was for Second Life development.

will microsoft + yahoo! be microhoo! or yahoo!soft?

Other folks are already writing about what Microsoft's proposed $44 billion buyout of Yahoo! could accomplish. Sure, maybe you can sum up their search and add businesses and create a more solid competitor to Google, although history is a bit short on examples of two losing businesses combining to create a Voltron of success. Moreover, what happens to top talent at Yahoo!? Microsoft's offer is a pretty impressive premium on Yahoo!'s current stock price, so folks holding on to complete their vesting may see this as a cashout opportunity. Many employees may have chosen Yahoo! over Microsoft because of -- in my opinion, outdated -- worries about Microsoft's culture, business practice, operating system choices, or whatever.

Worse, it seems like both Microsoft and Yahoo! are still trying to play catchup to Google by copying Google rather than by attacking whatever's next. Sure, when Google is raking in nearly $5 billion a quarter, it may be tempting to just copy them, but does anyone really believe that in a decade current approaches to search and ads will still be dominant? Adwords are the most profitable way to sell online advertising today, but there are 100 (1000?) different approaches being tried in garages and labs all over the world.

Many of those will be better than adwords. By a lot. I'm not sure that Microsoft is more likely to own one of those approaches by buying Yahoo!

Which in now way is a slag on Yahoo! Many friends at Yahoo! are among the smartest folks I know, but they don't seem to be leveraging the advantages they have -- remember, they have the most popular portal and email system on the Web -- to the degree I would have expected, and I doubt being acquired by Microsoft will change that.