Synthetical vs. analytical

My whole pre-MBA career, including the 4 exciting entrepreneur years, I was fundamentally a R&D engineer – I designed and developped crucial parts in highly complicated integrated circuit chips. By definition, the jobs of me and my colleagues were to "create new things" without a doubt.

One day I was having a coffee chat with our department head – we'll call him Mr. D for the sake of convenience. He said something that stays on my mind until now:

I think we need to change our technical interview process. We need to ask more synthetical questions rather than analytical questions.

This might seem obvious for a candidate that is gonna be doing R&D jobs, but the truth is our part of the design – analog, mixed-signal and radio-frequency integrated circuit design – was very very difficult. It took most of us 6~10 years of experience to be able to design a circuitry that's reliable enough that the project managers can therefore stay hands-off and focus on the integration, which is also a huge challenge.

If something takes 6~10 years to become adept in designing, it's closer to artisanship – in fact, many of my peers pride themselves as artisans or even artists instead of engineers – and it involves a lot of analyzing existing designs to learn from them. Those analyses are already hard enough. Even young graduates from the best schools could struggle to understand the odd nature of the cross-domain modeling of a phase-locked loop, the modern version of which is often credited to a paper by frenchman Henri de Bellescize in 1932. Despite having studied and analyzed tons and tons of legacy designs in engineering schools or even in the first jobs, most candidates would still fail to solve an analytical question completely. It's therefore quite common that in most IC design companies the technical interviews involve predominantly analytical questions, which would already be enough to screen capable candidates.

A classical model for modern phase-locked loops

A classical model for modern phase-locked loops

But analytical questions are backward-looking, which is the point of Mr. D. For the technology to move forward, the designers have to be synthetical.

An analytical question would be something like "Please develop the small-signal transfer function of this linear amplifier with two-stage feedback."

A synthetical question that follows up the above-mentioned analytical one would be "The amplifier is now shown to be unstable, please add some circuitry to stablize it."

An ultimate synthetical question might be something like "Try to design a phase-locked loop so that it can work as an analog-to-digital converter which has the potential to be much smaller and more power efficient than a traditional analog-to-digital converter."

Even though the day job of an R&D engineer consists of 60~70% of the 1st question, it's the 2nd and the 3rd questions that move the technology ahead. Obviously the 2nd and the 3rd questions are very hard. Some large companies like Broadcom even give people who focus on the 3rd question the title of Scientists, instead of the usual glorious Member of Technical Staff. Titles aside, the 3rd question remains the most crucial question for any IC design companies to stay ahead of the competition.

The same, if not more, shall apply to entrepreneurs when they hire people. And it's not just designers, but also sales and marketing people.

The reason is very simple – you're trying to build, market and sell something totally new with limited resources, you therefore need the most synthetical and constructive minds you can find.

For example, instead of asking an UI designer candidate what he or she has done before and how, ask him or her how they would design a new adaptive interface that could replace the static (dumb) menu icon when the web page is displayed on a tablet or a smartphone.

Instead of asking a marketing candidate what he or she has done in Walt Disney and how, give him or her a hypothetical product and ask him or her to design a viral campaign.

Instead of asking a sales how much sale he or she has generated in Cisco over the years, ask him or her how they would design a sales incentive system for your specific B2B cloud services that is going to forever change the way clients view the license fees.

Only by doing that can you find the forward looking people that will help you identify new problems and innovate to move things forward, instead of dwelling on the impossibility of solving certain well-defined problems.

And you know what? You should apply the same principle to selecting VCs – if you have such a luxury to select. Instead of going simply for the biggest VC, highest valuation or the largest sum of money, you shall identify the complementary resources a VC could bring to you in addition to your existing investors. If your existing investors, maybe angels or seed funds, are already introducing lots of useful contacts in the digital advertisement industry where your new products compete in, your next VC shall ideally be not mainly offering the same thing. They might be someone who has helped many startups scale their infrastructure exponentially, or they simply have decades of experiences in helping startups expand internationally. For you, they have to be the synthetical partner for the next stage, someone who could synthethise an even bigger future with you.

Keep Mr. D's words in mind, and you'll find yourself talking to candidates in a totally different and more inspiring way.

Evaluating $5bn+ startups with Peter Thiel's 7 questions

A reminder of the pitch-to-investment funnel