At least he used the term "minimise," rather than "eliminate." There are certain problems that exist among people, particularly in organisations beyond a certain size, that simply cannot be solved or mitigated entirely, no matter how much you may want them to be.
In the end, only the first piece of advice - "hire people with the right kind of ambition" - really matters. You have to build the right kind of organisational culture from the beginning.
A few points to add:
1) I would take a page from 37 Signals' general mantra and suggest that one should really try to avoid unnecessary hiring where marginal productivity of labour cannot be clearly shown to exceed marginal cost of labour, and try to keep the organisation as small as possible. Not only does the likelihood of "politics" increase proportionately with the size of the organisation, but there is a lot more room for strategic machinations and more time for people to contemplate how to grab a larger piece of the pie when they occupy roles or hold titles that don't _really_ need to exist, strictly speaking.
If a company people that would likely be trimmed in a downturn purely on the basis of general "uncertainty," rather than an objectively steep and protracted decline in sales, it's gone too far off the staffing cliff. This is particularly true in technology. Really, don't hire more people than you need just to feel like you're running something big.
This element tends to be well-considered when a company is very small, but usually forgotten once it gets to "mid-size" (say, 75+ people -- I know that's still "small" from many people's point of view, whatever).
2) One of the most persistent and agonising problems I have seen in my experience is small companies (say, 25 people or less) that--led by executives that come from mid-size or large-business management cultures--reproduce, in a fractal manner, the same kinds of pathologies that afflict large organisations.
Systems and processes that can be replicated at decreasing marginal cost are an essential and unavoidable aspect of significant business growth, but can be far more harmful than they are beneficial if enacted prematurely. I've seen companies of ~20 people with ~5 VPs, I've seen complicated and overblown and overly abstract departmentalisation methodologies imparted on companies of a dozen people, and, my favourite, the pervasive belief that if you blow big dollars on certain kinds of investments that are conceivably beneficial to a large company (six-figure accounting systems or project management systems), you will make big-company revenue.
I cannot believe how many times I have seen it claimed that a company of 1-2 dozen people should buy an enterprise-level project management suite because "that's what the big guys use." Yeah, well, that has no imaginable evidentiary relationship to what you should be using.
This idea that problems can be solved with ambitious throwing of fistfuls cash at them is naive and misplaced, and tends to warm the plate for a lot of political dynamics to play out.
Once people see that major strategy and/or purchasing decisions are made on the basis of bling or something read a half-hour ago on Delta in-flight magazine on return from Acapulco, they see the entire decisionmaking process of upper management as wide-open for gaming, as it is clear from that point that the primary objective of anyone seeking anything is to define and nurture a perception in the manner of textbook marketing.
Even worse, it even incentivises otherwise benign people to more cynical behaviour by setting undesirable common denominators; if you don't game the system, somebody else will. There are some people that have never had a cynical thought, and there are some people that always have cynical ulterior motives, no matter how benign and transparent the organisational culture, but it's the middle 80% of the bell curve -- the ordinary human being, who is pragmatic and given to occasional, circumstantial opportunism -- that you're really moving in the wrong direction by showing them that government-by-vague-impression, or worse yet, cargo cultism, does in fact drive most of the significant decisions.
3) In general, organisations with engineering-driven cultures (in the sense that engineering defines the primary discursive space and the central agenda) tend to do better with politics than sales-driven ones.
However, this must be balanced with the reality that to get high sales performance, you need to harness a lot of the kind of ambition and energy that only sales sharks have. If they feel marginalised or undervalued in an overly nerdy company that, from their point of view, spends much of its time on technical navel-gazing and twiddling its thumbs, they won't stick around long.