Thursday, February 20, 2014

Knowledge Portfolio Risk Management



Considering the time spent learning new technology as an investment is not new. It's been common for years. In The Pragmatic Programmer (Hunt/Thomas), the metaphor is extended to include:
  • regular, habitual investment
  • diversification
  • risk management
  • valuation
  • rebalancing
TheHackerCIO wants to focus today on risk-management.

Having worked on Wall Street, I can assure the gentle reader that serious Financial Investors don't operate on "gut-feel" when it comes to risk-management. Banks are required to model their risk (typically measures of "Value at Risk," or VaR) and limit trading and total portfolio exposure based on that quantity.

But there is no way for us as technologists to numerically measure the risk of our portfolio. It would be helpful if there were such a methodology.

Consider that as we invest in our "knowledge portfolio," we're not generally putting our financial capital at risk (most of us don't invest heavily in training programs). At least that is not the bulk of the problem. We are putting something even more valuable at risk! Our time. And our time is really our life. We're investing part of our life in the technology we study, learn, and play around with. And a life is a far worse thing to mis-invest or waste, than almost anything.

Or, to riff on the well-known commercial: if a mind is a terrible thing to waste -- and who would disagree -- then, one can waste it both by not employing it at all, or by employing it on things which provide no return, or little return, for the effort.

With that point established, it's worth noting that quantitative approaches to risk management in finance portfolios is a relatively new phenomenon. That it has been mandated universally for financial institutions is indicative of its success: what is measured can be monitored and controlled. So it bears consideration whether there are any steps we can similarly take to establish quantitative measures of technology portfolio risk management.

This is new-trail material here, so I don't know that I can offer anything earth-shaking. But let's give it try, in accordance with the al fresco approach of blogging, and see where it leads us. ;-)

How could we measure our knowledge portfolio risk exposure?

Posing it as an explicit question helps focus our energies in the right direction.

First, we would need to have our portfolio expressed in a written form -- not as a nebulous, woozy metaphor. This brings us back to our technology journals, doesn't it? Ideally, a journal should be dedicated to our "Knowledge Portfolio." Here is one possible approach:

Base:

  • an inventory of technology skills as they exist at present.
  • estimates of the time invested learning each item.
  • estimates of the time spent using each item.
  • estimates of the time saved by using each item [e.g., how long would the application have taken to write without using the Grails framework?]  
    • Time-saved = time-it-would-have-taken-without-the-technology - (time-spent-learning + time-spent-using) 
  • calculations of the return-on-investment = time-saved/time-spent
  • assessments of levels of risk for each item: 1-10, where 1 is very conservative and 10 is cutting-edge.
  • calculation of amount of time-investment in each risk-level. 
  • calculation of total portfolio risk: TimeAtRisk, or TaR? Which could be defined as the risk-adjusted expected return-on-investment of time-spent on each item in the portfolio. 

New Investment Assessment and Planning:

  • a list of potential new technologies to invest time in.
  • an assessment of the learning-curve associated with each item. The steeper the learning-curve, the higher the risk for the investment. Perhaps a 1-10 scale from conservative to cutting-edge. 
  • Determination of amount of time to regularly invest on a weekly basis.
  • Calculation of the expected time-return, risk-adjusted, for each possible investment item.
This is all just so much speculative fun. But it would be an interesting project to actually attempt to work it out. After all, we as developers, all know how many "slips 'twickst the cup and the lip" there are when working out a new approach to anything. Still, I think the overall idea is useful and it might even be possible to gain some better insight and control by attempting to quantify it. It certainly couldn't hurt. I'd say that the discipline of reviewing time-spent weekly on learning new technologies, and ensuring that a regular investment was made each and every week -- that alone would be worth a fortune, for most. 

It also occurs to me that a set of metrics -- even subjectively reported as evaluations -- would be immensely helpful in looking at new technologies. A web-site, for instance, which categorized and measured a variety of criteria helpful in estimating the slope of the learning curve would be a wonderful resource. 

Of course you could really get fancy and do Monte-Carlo simulations of the valuations, and feed those into a model, the way the Financial Institutions do, but it seems a reasonable first step to just attempt to quantify the actual results we achieve from the time we study. 

Any takers?

I Remain,

TheHackerCIO