Bouncy Balls

I like this new TV ad from Sony. 250'000 bouncy balls are let loose in a hilly San Francisco street. Their descent is shown in beautiful, meditative pictures, underscored with "Heartbeats" by José Gonzàlez. The video reminds me of happy days in San Francisco.



However, the making-of featurette is somewhat disappointing. It's very self-congratulatory and overhypes their "like no other" tagline. Also, it doesn't really give answers to the real, hard questions: How did they collect the balls after each shot? Did the cars get damaged in the process? And, finally, how in the world did they get that frog to jump out at just the right moment?

The New York – New Haven Train Schedule

The last entry was about flight timetables in Envisioning Information. I'd like to continue this tour de force and talk about train timetables. One of my favorite examples in the book by Tufte is the New York – New Haven train schedule.



Prior to Tufte's redesign, the timetable has many shortcomings: Too much space is allocated to the column headings: Only 21 percent of the area is devoted to actually giving train times. The many lines and boxes "create an elaborate but false appearance of systematic order". But worst of all, the times are listed in a strange pattern: the day's schedule follows a serpentine path.



For comparison, here's the top portion of the new design.



The book was published in 1990, so you'd think that Amtrak would have picked up on these recommendations and adjusted their design. However, when LT and I made a trip to New Haven, unbelievably, the schedule still looked the same. I clearly remember my confusion about following that snaky path, and the 'bold-means-p.m.' visual syntax. LT, conversant in urban matters, took the matter out of my hands and said "Yeah, ignore that, the train's at 2:33 ".

Envisioning Too Much Information

My friend Mesch is a software engineer who works on graphics- and UI- related stuff. Also, he's a very smart guy. When I visited him at work two weeks ago, the only book he had on his desk was Envisioning Information by Edward R. Tufte. "That one has got to be good," I thought, and got myself a copy.

In the book, Tufte talks about the different aspects of communicating information on paper. Often, he starts by showing a badly made illustration and then presents an improved version of it. The results are often diagrams that are more understandable, contain much information, and are easy on the eyes.


Swissair 1949/50 Winter Timetable

One of my favorite high-density diagrams is the Swissair Winter Timetable of 1949/50. It is very different today's rather boring timetable presentation in booklets with tiny print. There is one circle for each city, geographically arranged. Every flight connection between two cities is represented by a line between the two and the departure arrival time written inside the circle, next to the line. The flight number is listed in the center of each line.


Circles for cities, connections with lines instead: no tiny-print listings.

I think this is a very interesting way of representing the data, and many variations would be possible: For example, an online travel site could show such a diagram with prices and airline logos instead of connection times and flight numbers.

Tufte, it seems, would agree: His book features a similar map from the Czech airline as a positive example. There's a lot information encoded in the map, a concept the book endorses by saying: "To increase clarity, add more detail."

Why doesn't anyone use this excellent concept from the middle of the 19th century? I'm not sure, but let me venture a guess: People like to see illustrations that are similar in type to those they have already seen. This is not because they're dumb, but because it takes far less time to absorb information when it's presented in a familiar way. That's why flight plans and timetables are still presented in lists.

More Notes from Startup School

I was amazed at the popularity of the previous post. Even considering that I wrote it on a plane from Boston to Frankfurt, and it contained plenty of errors, it still tripled my website's traffic.

So I decided to sort through my hand-written notes and type them up to cover the other speakers as well. In the previous post, I only really talked about Michael Mandel and Chris Sacca, who left the strongest impression. Aaron Schwartz announced that there would be videos of the talks, so you might want to watch those once they come out. There is also a notes section at the original Startup School site.

Langley Steinert
Langley is the Chairman of TripAdvisor, a travel advice website. He had also worked on Viaweb with Paul Graham. His talk was about his experiences in the early stages of TripAdvisor, which he eventually sold to Expedia.

I remember him making a strong point for keeping an eye on liquidity: "Always have 12 months of money in the bank, raise money as soon as you fall below." There are two reasons for this: First, you have to be able to focus on running your company instead of spending all your time on chasing money. Second, once your bank account is empty, it won't be easy to get favorable terms: They have you by the balls.

In the early stages of a web startup, you're interested in getting visitors, quickly. Langley suggested three ways of doing this:

1. Do PR. The media people have to fill their pages and sites with content. They'll be especially happy if you give them statistics and nice little stories. Be opportunistic, not aggressive.
2. Have a good product.
3. Search engine marketing. TripAdvisor did this heavily in the early stages.

Overall, this was a very good talk. Langley seems like a nice guy and his slides had plenty of interesting data.

Mark Hedlund
Next up was Mark, who started with some general advice about doing a startup, and then went through a list of startups he found interesting. In the first part, he warned about founding a company that has no foreseeable chance to generate revenue – otherwise, there's no way to get VC money.

Also, he warned about starting a company on the basis of being "the Google for X". Simply offering a variation of search for a different field may not cut it.

He went on to talk about some startups and their respective stories. His examples were Flickr (photo sharing), Jotspot (corporate wiki platform), Koders (search engine for source code), Odeo (podcasting), Project PlaceSite (social networking for people sitting in cafes), Spike Source (open source heavy-weight), Splunk (search software for log files), 37signals (web-based collaboration), upcoming.org (social event calendar, acquired by Yahoo), and Zimbra (AJAX e-mail). Most of these were new to me: I only knew of Flickr, Jotspot, Odeo, and 37signals. I love the 37signals products - Basecamp is fantastic, and their simpler-is-better approach rocks.



Qi Lu
Qi (pronounced "Chi") is a VP Engineering at Yahoo. He talked about some of the new Yahoo products and their efforts to reach out to developers. Also, he wanted to recruit some talent for Yahoo. While this wasn't so directly related to startups, he did show some interesting stuff.

Previously, I had only seen the new Yahoo Mail from screenshots, and Qi demoed some of it to the audience. The new Yahoo Mail looks much like a web-based Outlook, is similarly interactive as a desktop app, I'm sure the average user will like this, as this is what he/she is used to. Maybe GMail will have its behind kicked. Time will tell.

Disclaimer: I once interned at Yahoo and still think it's a very good company.


Hutch Fishman
Hutch is a startup CFO veteran and has worked at many young companies.

He took us through the typical startup funding stages. A company might go through all of these or just the first few:

• Seed: Strategy and market research, develop business plan


• First Round: Complete initial product, build initial team


• Second Round: Bring product to market, initial customer deployment, complete team


• Third Round: Expand sales and marketing, significant risk taken out of business


• Fourth Round / Mezzazine: Working capital needed for liquidity event

I liked this talk a lot, because it didn't just talk about abstract concepts, but went through the mechanics of startup finances in detail.

At the end, he made a strong point for working with top-tier firms: If you work only with the best lawyers and the big 4 accounting firms, this will make things much easier during the preparation of an IPO or a sale.

Paul Graham
Paul's talk can be found here.

When I was a TA for a CS class last year, I motivated (or tried to motivate) my students by saying that the best way of getting new ideas is to know the old ones. And then taking them apart. Therefore, I think Paul is right about the need to find new ideas through questions. Also, Paul was the first speaker without slides: Peter Norvig would be proud of him.

David Cavanaugh
David is a patent attorney and explained the different IP protection schemes of interest to startups.

Michael Mandel
See my previous post.

Steve Wozniak
There is no need to introduce Woz. He talked about his experiences from HP, Atari, and the early days of Apple. I didn't know he was such a funny guy. At the end, the audience thanked him for his achievements with a standing ovation.



Stephen Wolfram
Stephen is the founder of Wolfram Research and the creator of Mathematica. In addition, he is an extremely smart guy.

He talked about his first company that he founded while at CalTech. He eventually left because of disagreements with management and sales. Therefore, he kept his second company, Wolfram Research, strongly engineering-based.

Eventually, he took some time off to write A New Kind of Science, and showed some examples of where this science could be used: I found his ringtone example rather amusing. Almost equally impressive was the fact that he got the auditorium's audio system to work right away – that stuff never works the first time at my school.

Stan Reiss
As the VC representative, Stan Reiss of Matrix Partners stood in a pretty tough spot; Paul Graham has repeatedly bashed VCs as being unhelpful herd followers.

Stan picked up on this and incorporated some self-humor into the presentation. Overall, he offered a good overview of the VC landscape.

One of the questions from the audience was: "How do you protect yourself against a VC who listens to your idea, and then turns and gives it to someone else?" Stan argued that quality VC companies don't do this. I certainly hope so.

Mark Macenka
Mark is a partner at Goodwin Procter, a law firm. He held a great talk about the legal traps and mistakes that startups should avoid. According to Mark, it all begins with good advice. Personally, I think having a good lawyer is important, but you should focus your energies on building the product.

It helps to have a person on board who is obsessive about legal stuff: You can avoid so many pitfalls later on if you make the founders and early staff sign papers that bind them and their inventions to the company.



Chris Sacca
See my previous post.

Olin Shrivers
Olin is a co-founder of SmartLeaf, a company that makes automatic fund allocation software for mutual funds. He is also a funny, straightforward kind of guy.

He recounted some anecdotes from his company's early days, the herd cult among VCs, and the fact that when you look at it honestly, a startup is all about money. He argued that startup founders need to have a high tolerance for feeling like a moron, as any great new idea will first encounter much initial resistance.

One of his strong points was that you have to have good technology and be respectful of the user. For example, it pays to have a one-sentence privacy policy; his went something like: "Barring legal process, we will not give out your information to anyone, at all".

Summer Founders
The event closed with a panel that featured this year's summer founders. They talked about their startup experiences, and the up and downs they had experienced.

Startup School: Michael Mendel and Chris Sacca

This weekend, I visited Y Combinator's Startup School in Cambridge, MA. It was awesome.



The night before, I attended a reception with some attendees and presenters. This allowed for some schmoozing with other potential founders and, of course, the speakers. Typically, these were surrounded by scores of kids, so getting to talk to them required interrupting someone else. I managed to listen in to conversations with Stephen Wolfram (the creator of Mathematica), Michael Mandel (of BusinessWeek) and Paul Graham (the organizer). Still, the most interesting attendee was Joel Spolsky. I have been a big fan of his articles on Joel on Software for a long time, so I was happy to have the opportunity to chat with him. I found it rather surprising that he was there: After all, his main interest would be running his business, not talking to young startup founders.

The next day was full of back-to-back presentations. They were of high quality: Langley Steinert of TripAdvisor, Apple co-founder Steve Wozniak, Paul Graham, and Olin Shivers of SmartLeaf all held very good talks. Even the VC guy, Stan Reiss, seemed trustworthy.

There were two speakers I found especially interesting, partly because they weren't really about founding a company, but the encompassing scheme of things.

Michael Mandel on US Advantages

The first one was Michael Mandel, the chief economist at BusinessWeek. As an economist, his job is to look at the economy as a whole. Interestingly, he didn't seem concerned about the US trade deficit, the holes in the budget, and the savings rate. He wasn't even concerned about the "boom, bust, boom, bust" cycle of today's economies.

His interest is startups, since they are the only elements of the economy that drive things forward. Big companies typically fear innovation and competition, so they won't risk doing the things that startups do. In his world view, the US has two unique advantages:

1. There is a working system of Venture Capital.

2. People are willing to work more than in other countries. You can't achieve much innovation in 35 hours per week.

I agree that these are important points, but they are slightly incorrect: VC is gaining a foothold in Europe, even if there are no big-name firms yet that have a strong track record. His second point stems from a misunderstanding of European work culture: Sure, union members in big European car plants do work 35 hours a week. In contrast, I don't know of technology people in Europe who put in that little effort.

Mandel pointed out that the unique US advantage is not technology, as it is very easily transferrable to other countries. This point I agree with: The whole trend of outsourcing to India proves this point.

Chris Sacca Talk

Chris is BizDev person at Google who is, among other things, in charge of deciding which small start-ups to acquire. At startup school, he found an attentive audience.




He explained what he thought were good startup ideas. In his view, these were concepts where you could "Go Big!", i.e. make a product that solves a problem that many people have and therefore can reap huge rewards in the long run. Being "Cheap to demo!" was important to him, as a presentable product will get you bought quicker. Also, Google still believes that "Geeks rock!"; they are the ones who have ideas and drive the technology forward, not the marketers and business people.

A large part of his talk, however, was devoted to discussing Google's culture. He talked in length about the value system originally put in place by Larry and Sergey and the various perks that Google offers its employees.

I found this talk to be very good and I think it resonated well with the audience. He seemed approachable and encouraged attendees to come chat with him after the event. As I walked out of the auditorium, people were forming circles around Chris.

Dislaimer: I was an intern at Google last year, but never met Chris. Also, I'd like to add that what he said was true (as of last year), although I never actually believed the swimming pool would see the light of day.

Some Final Thoughts

Holding this conference was a great idea by Y Combinator and Paul Graham, and they I'm sure they invested a great amount of effort into organizing it.

The turnout was incredible: Harvard's Science Center B, with seating for 500, seemed almost completely full. It seems like there is a strong interest in the US to build startups. This is great: After all, as Mike Mandel suggested, these future founders are the people that drive innovation and long-term economic growth.

New York, New York

During last week's visit to the City, I found out one thing: New York is out of this world.



Its scale is inhuman, the weather is mostly miserable, and the traffic is insane: Just getting from one end of Manhattan to the other can easily be an hour-long undertaking.

On the other hand, it is the center of the universe, although not in all respects. It certainly is the undisputed financial center, with a large portion of the world's money washing through Manhattan every day. It is also a cultural hub, (although not the only one): There are numerous world-class museums, all just a few miles from each other. The city appears in almost every movie, news broadcasts and late night shows are produced there.

When in a foreign country, it's always interesting to listen to what people chat about in public. In New York, these conversations typically revolve around relationships ("So Carol really is getting married to Tom!") and, more importantly, real estate ("My friend John just paid $1.1 million for a 1 bedroom apartment – isn't that a steal!?"). This corresponds pretty closely to what I had expected: After all, the local paper, the New York Times, is the only paper to feature a gossip column with dollar amounts: The Real Estate section.

MIT's Stata Center

Last week, I toured MIT's new Computer Science building, the Stata Center. It is a building by Frank Gehry and has become the most photographed building on the MIT campus. Before that, it quickly rose to fame for its large cost overruns - $300 million was spent on construction, instead of the planned $120 million.



Aside from the architecture, what left an impression was the fact that almost every part of the building had been named seperately. The Stata Center has two towers - the Gates and the Dreyfoos -, and even the cafe has been named after the Forbes Family. These naming grants covered a large part of the construction cost.

On the other side of the pond, universities in Europe rarely have buildings with naming grants. There seems to be a concensus that this practice is a sell-out to egocentric interests, and all buildings should be financed by the state.

I disagree.

When European universities erect new buildings, they are financed through taxes - the costs are a burden shared by all. Why not have rich people - of which there are plenty - give money? It would take a burden off the public's shoulders, and leaving more money on the table for doing excellent research.

Fantasizing about Free Wi-Fi in SF

Good news for San Francisco residents: Google is planning to offer free wireless Internet covering the entire city area.

The media's visions are those of you sending e-mail to granny from Union Square. However, I think it's interesting to take some time to fantasize about the long-term effects of this.

By offering a free, always-on network, they're basically launching a large experiment in ubiquitous computing. As more and more mobile phones / PDAs are coming out with WiFi included, I think people may start downgrading their cell-phone plans. They will use the same phones, but talk for free via Voice-over-IP. [1]

Another effect is that devices will be able to talk to other devices for free: Your MP3 collection could be synced with your car's stereo, without having to connect it to your notebook. Vending machines will be able to ping headquarters and tell them they should send someone with more soft drinks.

Predictions are hard, especially when they concern the future. I'm very interested in how the free Wi-Fi will turn out.

I'm sure this will pay off handsomely for Google. Still, I think they shouldn't display an ad on each page. They will make enough money by being able to show very finely grained contextual ads on their search results or on Google Maps. (e.g. "$3 dollar coupon for the Starbacks to your left - click here!")

---

[1] Of course, I'm disregarding some technical details here, for example the need for a decent wireless hand-off protocol like SyncScan to make this work right.