Saturday, July 11, 2009

Startup Pains & Pleasures #4: Workplace Creativity & Fermi Problems

In my blog last week, I talked about my everyday creativity “trick” - if you cannot solve a problem, find an approximation of the problem that you can solve. If this solution is adequate, meaning it is still of business or technical value, if you can “sell” the approximate solution, do it. Then try for a closer approximation of the problem and solve it; this will be your version 2 solution!

Have you tried “creativity by approximation” yet? I will give you a very simple example from last week. My wife picked me up from the airport and during the drive back, said to me, “I meant to vacuum the whole house before you got back from Tokyo but as I was too tired, I just did the heavy-traffic areas”. I was pleased that a readymade example for my creativity “trick” had suddenly dropped in my lap. She approximated the “whole house” by “heavy-traffic” areas, solved the problem, and had energy left over to go out later that night. This is classic day-to-day creativity!

Is there is a *systematic* way to arrive at approximate solutions? There is: a favorite method of mine known as “Fermi Problems”. Author Hans Christian von Baeyer has a classic article on Fermi Problems called, “How Fermi Would Have Fixed It” (Sciences; Sep/Oct88, Vol. 28 Issue 5, pp2-4).

Let me explain the classic Fermi Problem method using an example. When I taught Probability Theory & Stochastic Process to EECS students, the first question I asked them in the very first class of the semester was the following: “How many piano tuners are there in Ann Arbor?” When the class acquired a collective glazed look in their eyes, I took them through these steps:

1. Take a guess at Ann Arbor’s population – say, 300,000 people.
2. Now, assume that an average family contains four members. The number of families in Ann Arbor must therefore be about 75,000.
3. If one in five families owns a piano, there will be 15,000 pianos in Ann Arbor.
4. The average piano tuner:
a. Can service four pianos every day of the week for five days and
b. Has two weeks of vacation during the year

- A typical tuner would service (4 pianos per day X 5 days a week X 50 weeks per year) = 1,000 pianos per year.
- So, there must be about 15 (15,000/1,000) piano tuners in Ann Arbor. Check the telephone directory Yellow Pages and see what you find . . .

The guesses in the first 4 steps are totally unrelated. In statistical terms, these random variables are independent and hence the errors in these guesses tend to “cancel out”. Approximate solutions arrived at using the Fermi Problem method are surprisingly good (within an order of magnitude of accuracy) and hence very useful for practical situations.

What this tells us is that when confronted with a problem, do not be dismayed by a lack of information; Get Creative! – (a) break the problem down into a number of sub-problems and solve the sequence or (b) solve an approximation of the problem.

In the first case, you find approximate solutions to a bunch of exact sub-problems and in the second case, you find an exact solution to an approximation of the problem. In either case, the approximate solution you reach is a creative way out – either when you are in a tough job interview (you may recall the infamous Microsoft and Google interview questions!) or when confronted with a rough practical situation.

Next blog onwards, I will leave the Start Up topic and move on to other interesting issues.

Enjoy the haiku . . . PG


“Summer evening
Hear the breeze caress the leaves
Honey BBQ flavor!”

- AM

Saturday, July 4, 2009

Startup Pains & Pleasures #3 - Workplace Creativity

Before I discuss creativity in the workplace, let me dwell on some topics related to “Axiomatic” HR from last week’s blog. I encourage you to check out my readers’ comments also.

First of all, “Company Priorities” are the axioms that drive more than just HR (even though I do not use them to identify what markets to address or products to produce). Let me give you an example beyond HR. When we were deciding on how the architecture team members would exchange design information, one factor guiding the decision was the “Globally-integrated Business” priority. Even though there was no immediate need for a sophisticated solution, I decided to adopt “Webex” - a conferencing tool that allows employees dispersed worldwide to collaborate effectively at low cost. In general, what I found was that if I consistently used the axioms, making even simple day to day decisions became easier, and quite importantly, easier to explain to team members.

One of the unanticipated and rather startling insights that came about was related to “tiled teams”. Such concepts are never right or wrong, rather a question of whether it is right for a certain time or a stage of company growth. Anyone who has done a startup knows: startups go through many “near death” experiences! The reasons are many; ours were never due to product or market collapse (we had done the technology before, had the required licenses and had great ecosystem partners). Ours were due to the classic reason for startup near-deaths: dried up cash flow – especially after the October 2008 Great Recession.

There is no doubt in my mind that “tiled teams” are ideal for a going concern with sustained cash flow. In retrospect, however, it seems to me now that when going through near-death experiences, things may have been a bit easier if *everyone* pulled in the same direction (which is more likely to occur when all the team members are like-minded). A near-death experience is not the time to expend a lot of energy “herding cats”. In other words, “tiled teams” may *not* be the best for an angel-stage startup, but may be ideal for a later Series C-stage startup enterprise.

Summarizing, start as a “garage” operation, locate near the VCs, don’t enter into personal liability and keep over 50% ownership of the company to yourself. Find axioms, whatever they may be, that capture core company priorities; let them drive every internal decision possible. At an early stage, start with teams of like-minded people and as your company becomes self-sustaining, remake the teams into “tiled teams” by mixing in new hires with non-overlapping skill sets and behavioral styles.

Creativity: With respect to creativity, I hope my list of US patents and published articles lend credibility to what I have to say on the topic.

As you will note on the right-hand side of the page, I see creativity and tenacity as tightly coupled; in fact, as two sides of the same coin. If you are not tenacious, give up easily and move on quickly, you never arrive at the stage where you have to grapple with an issue long enough to bring to bear your creativity and ingenuity to solve it. Here I am talking about day-to-day creativity and not “high-level” grandiose creativity.

My “trick” is very simple: if you cannot solve a problem, find an approximation of the problem that you can solve. If this solution is adequate, meaning it is still of business or technical value, if you can “sell” the approximate solution, do it. Then try for a closer approximation of the problem and solve it; this will be your version 2 solution!

I have never met a problem - be it in business, marketing, product design, technology, signal processing, whatever – that I cannot solve in the sense of the previous paragraph. The key is “approximation”. Actually what makes this approach viable is that one can make precise statements about approximations – people well-versed in esoteric topics such as measure theory and functional analysis know this fact very well (the precise concept of “almost everywhere”, for example). In fact, such concepts lead to rapid expansion in certain areas of pure mathematics and the solution to a large class of problems; these developments have had a direct bearing on advances in the engineering fields of image and signal processing which have given us a profusion of incredible consumer electronics devices in recent decades.

If one cannot make precise statements about approximations, the solver and the solution user may both be unclear about which problem was actually solved and what the roadmap is to arrive at the full solution. Let me give you an example from engineering within a B2B business context. Typically, development starts with an RFP (request for proposals) from a buyer to a vendor with a long list of requirements. Even in a prototype-driven development model, something similar happens. The first step of the “approximation” that I talk about is this list of requirements. In all likelihood, meeting all the requirements in a first generation product will make it expensive and the early end customer will use only a subset of features!

The vendor and the buyer should negotiate to identify the first subset of requirements to meet (product version 1), then decide which requirements to meet (and when) in the next step (product version 2) and so on, to create a solution *roadmap*. This is the “precise statement about approximation” that I am talking about – nothing very esoteric as you can see; I have simply made it explicit that the product solution is not a “point” event but a “trajectory” and called it out using some new terminology.

If used consistently, this same approach also contributes to *personal creativity*! Whether you are arguing with your teenager about course of study or you are grappling with an engineering problem that was assigned to you, you can use this “creativity by approximation” method. I am sure you have experiences along this line and look forward to comments and stories that embody the sentiment: “creativity by approximation”. Give it a try in a potentially creative situation during the coming week and tell me about it . . .

Enjoy the haiku . . . PG

Let us conclude with a random haiku . . .




“Pedestrian mall
In central Ginza on Summer weekends
Street shamisen music”


- PG