Ryan Says…

Today I attended ’startup school’ which was held at Stanford - basically a Y Combinator event that is co-sponsored by BASES. I was surprised at the huge turn out - I would estimate around 500 people, and judging from those I talked to a lot had flown in from out of state for the weekend. Considering there are events like this pretty often in Silicon Valley (although this one was of high-quality and free), I can only conclude that the large audience is due to the ‘cult of Paul Graham’ :) It seems Paul, and Y Combinator, have a virtual monopoly on attracting hackers just out of school (or still in school) who want to start companies. This is probably because a lot of Paul’s writings (many of which I really enjoy) espouse the virtues of teams of young technical founders, as well as his willingness to fund teams in the very early idea phase (or even smart teams with no idea at all!)

Anyway, the event did live up to its apparent reputation. With a full day of talks I can’t hope to reproduce much here, so I’ll select a few of my favorite points from each speaker. You can find more here, here, and here. I believe there will also soon be podcasts on the startup school site.

Mark Macenka - Partner, Goodwin Procter

• It’s important to have confidentiality agreements in place when you discuss your ideas with people, if for no other reason than to show intent (in case you need to later prove in court that it really was confidential information - this would be hard if it can be shown you have talked to a bunch of people openly about it).
• Founders should agree early on about things like who owns what percentage of the company, who is on the board etc.
• Look into I.R.C 83(b) for a tax break on founder stock - you only get 30 days to use this!
• Overall with your early stage ‘legal stuff’: do it early, document things, keep it simple.

Paul Buchheit - Creator of GMail

• When defining a new product idea, try adding the phrase “that actually works” to the end (e.g. “an online email client that actually works”) to help you think about ways of doing something differently than what’s out there
• On scaling your infrastructure: treat disks as sequential access devices. Keep your ‘big data’ (e.g. media) on disk or a service like Amazon S3, and your ’small data’ (tags, meta-data) in RAM. DB updates should avoid seeks.

Chris Anderson - Editor in Chief, Wired Magazine

• On his ‘Long Tail’ theory: The Gaussian (normal) distribution is an artifact of scarcity - applies if there is a significant cost to moving many SD from the mean. E.g. distribution costs a lot of money - a 20th century view
• Forces leading us towards the new ‘long tail’ world: 1) democratization of production (PCs, Rails etc), 2) democratization of distribution (Internet), 3) Connecting niche supply and demand (search, recommendation engines etc).
• Old: dozens of markets of millions; new: millions of markets of dozens (attributed to Joe Krauss).

Paul Graham - Partner, Y Combinator; Founder, Viaweb

• Discussed 16 reasons why people go and get jobs instead of starting a company, and proceeded to show why most of them aren’t very good reasons
• Basically a good motivational speech for the type of hackers in the audience.
• Interesting stats that 50% of the first batch of Y Combinator companies were ’successful’, 25% are now rich, 0% would have traded the experience for a cube job :)

Michael Mandel - Chief Economist, BusinessWeek

• Productivity growth has started to decline since 2003. It is very important that we reverse this. Need to unlock more possibilities for growth (new applications of technology).

Max Levchin - Founder, Slide; Founder, PayPal

• Statistics are important - product managers should measure everything.
• Analytics packages out there are not good enough - need to build your own. Google Analytics is a good start, however.
• Tail your logs, graph results, look for anomalies (shows what users want to do). If all your metrics are flat this is a big problem.
• The color blue is used extensively on the web so sites are more accessible to those with red/green color blindness
• You can now design sites that require scrolling (don’t fit on one screen) because scroll wheels have become quite good.
• Your business should appeal to at least one of the seven deadly sins. This is a good motivator.
• If you make things too simple, there is no commitment from the user. Make them do a little bit of work to get value (but not too much - find the balance)

Ali Partovi - Founder, iLike; CEO, GarageBand.com; Founder, LinkExchange
& Hadi Partovi - President, iLike; GM, MSN.com; Founder, Tellme

• Make sure you can explain your customer need in one or two sentences.
• Network effects are a good barrier to entry to establish (value increases with the number of users in the system - makes you harder to displace)
• Listen to your customers - read forums yourself, know customers top 5 likes and dislikes about your product.
• Align the top X problems in the business with the top X people in the organization.
• Hiring and recruiting should be top priority. It’s better to have false negatives (turn away good people) than false positives (hire sub-par people).
• Company culture takes work. There is a natural tendency for a company to become political/bureaucratic and once this happens it’s hard to reverse. Usually occurs when you hit around 20 people - hold this off as long as you can.

Rahoul Seth - CFO, Adteractive

• Plan your cash requirements early - before you seek seed stage funding.
• VCs often approach valuation in ‘reverse’ - working backwards from their desired percentage ownership
• Founders should aim for 5-10% at liquidity
• Employee stock option strike price is usually at a discount to the preferred stock price of the latest round.

Mitch Kapor - President, Open Source Applications Foundation; Founder, Lotus

• Considers himself a (very) early adopter, and able to spot trends
• Current trends he thinks are important:
• Social production (e.g. wikipedia)
• Virtual worlds (e.g. second life)
• Gave some interesting stories about the early days of Lotus, and how things for early stage startups haven’t changed much in 25 years.
• On corporate culture: every action and inaction sends a message. Walk the talk (mind the gap between the message and practice). Hold people accountable - don’t tolerate abusive behavior (even from star performers)

Greg McAdoo - Partner, Sequoia Capital

• Introduced as the world’s best VC firm :)
• Very useful talk on what Sequoia looks for in investments:
• Clarity of purpose
• A spectacular market
• Alleviate a customer pain
• Team DNA
• Incredible product focus
• Real operating margins
• Frugality
• Inferno from a single match
• Get your product out early and innovate
• Know your competition - and respect them.
• Sequoia likes it when there are large potential competitors (means the market is worth pursuing)
• Can’t just be linearly better than competitors - need a reason why big competitors really can’t serve this market as well as you can.
• As a startup you can’t create large waves, so you need to ride large waves
• Understand TAM, SAM, SOM (who is and is not a customer)
• Make it clear early on whether you see yourself as a long-term CEO (e.g. Bill Gates, Jeff Bezos) or if you will need to bring in a CEO. Be realistic.
• Figure out what corners you can get away with cutting, and cut them.
• Look for folks whose hair is on fire, and build a fire hose (doesn’t matter what color the hose is or if it leaks a bit). Similar to the vitamin vs aspirin metaphor.
• For a software company a TAM of 0.5B+ is attractive, for HW 1B+, for services 2-3B (due to lower margins)
• It is OK to not have a fully baked business model at the beginning
• Big waves aren’t necessarily created by R&D but are a consequence of a confluence of many things.
• Recommends locating your management team in silicon valley or Route 128 corridor. Engineering team can be in a lower cost area.

Mark Zuckerberg - Founder, Facebook

• Quite a good speaker for his age - used no PPT and commanded the stage
• Fundamental point: when starting a company it’s important to be smart and young.
• Seems very fond of iterative development (they release code nightly). Repeated this theme often.
• Even your ‘business people’ should be technical
• Don’t keep marketing teams, product management teams etc in silos. Give power to people so they can iterate
• Instill values in your organization and then empower people to make decisions
• Wants everyone to build stuff on top of facebook :)

Joel Lehrer - Associate, Goodwin Procter

• Tried hard to make a fairly boring topic (patent law) interesting
• PTO is not allowed to search sites like wikipedia for prior art - Google searches were only recently allowed.
• The allowance % of patents has dropped from ~65% in 2002 to 50-55% today.
• To get a patent your invention must be inventive (new and non-obvious, a high bar) and useful (low bar).
• After disclosing your invention you have a 1 year grace period to file a patent in the US. Not so in other countries
• Getting a software patent takes ~4 years and ~$15k
• The best reason to file a patent is because it forces you to think about and articulate your differentiation in the market
• When building your product, work on your own time and equipment (not at work)
• Understand your current obligations (employment contracts etc)
• Document contributions from others, get assignments
• File a provisional patent before you launch your site
• Do some (not too much) searching for prior patents
• Make sure you use open source software in the right way (less of an issue if you are a hosted service and are not distributing software). Keep a list of what open source projects you use, and the relevant licenses
• Sometimes its better to keep a trade secret than file a patent
• Recommends drafting and filing provisional patents yourself (only costs $100). Get a well-drafted example from a patent attorney
• Search firms like NERAC, INREA may do some searches for free if you tell then you are an exciting new silicon valley startup (and imply you may bring them future business :)

Whew, long entry!

Share This Story