The fallacy of the linear value of time

So often I see a cost estimate, or worse still a business case, based on the fallacy of the linear value of time. You know the sort: a five minute outage of the service costs $100k; our tool saves five minutes per employee per day which equates to $76M per annum.

The cost of interruption to a service, or the value of time saved, is generally non-linear. (Heck, everything in our physical world is non-linear - a fact often overlooked when forecasting economic recovery, safe levels of biotoxins, sea-level rises, AIDS penetration, technology adoption...) If a failed service serves office workers then a five minute outage may cost nothing. They stretch, they take a leak, they come back. It is a break they would have taken anyway. It might even boost productivity. A two-hour outage costs much more than 24 times a five-minute outage.

On the other hand if it is a front-office booking system then yes 5 minutes hurts.

Or if it is an automated financial system then you can value 10 seconds of outage.

As Aale Roos said on a LinkedIn discussion about this: My model is to ask the value of different lengths of interruption, for example: 5 min, half an hour, hour, 4 hours, day, several days, month. It is quite unlinear. At one point there can be a big jump.

The same fallacy is used by those who build business cases with an ROI of 5 minutes saved per day for 20,000 employees. Pure B.S.

(Actually safe levels of biotoxins is a real interesting one. Interpolating data from rats force-fed lethal doses of carcinogens to determine a "safe" level for humans is pure scientific silliness and all the scientists know it. The "natural" world is a toxic, carcinogenic, radioactive environment. Our bodies have evolved in it. The one and only shred of truth or sense in homeopathy is the concept that a little of what hurts you does you good. There are experimental indications that low doses of some supposedly harmful substances - or radiation - actually strengthen the body.

Bugs too. Certainly the current obsession with germicidal cleaning agents is pure neurosis. The link between child allergies such as asthma and LACK of exposure to environmental bacteria is now established.

If the safe levels of radioactivity demanded in nuclear facilities applied in the real world, we'd have to evacuate Denver. If California is truly serious about removing carcinogens from the environment, when are they going to ban potatoes, chillies and tomatoes? In fact many foods. Mother Nature is not held to the same exacting standards of trace toxins that our manufacturers are.)



Assuming linear relations is a common fallacy. The reason we do it is that it works for short term in many cases. It can get really harmful long term. Think salt. Make your favorite salted dish without salt. Taste and add a little salt. Taste and add a lot of salt. Keep adding until you see my point (but do not eat too much salt, its not good for you).

I just realized that we are seeing the same thing with ITIL. A simple process model proved to be useful in improving service. Investing a little in training with certification helped. Using a software tool brought savings. So with the linear logic more processes, certifications and tools will improve results. It won't.

Nokia Siemens Network has set up Simplification Office to simplify processes & bureacracy, I wonder were they hit by ITIL V3. I Nokia Siemens case it could be just the clash of Finnish and German business cultures but any V3 house has a danger of needing their Simplification Office soon.

OGC is rolling out ITIL Lite which could mean that they have seen the danger too. Unfortunate choice of name, in Europe only women buy lite products. That's why we have Coca Cola Zero and Pepsi Max. ITIL V4 or ITIL Max would have been a much better name. I have re-read ITIL V3 books lately (took the Art of Service e-learning Managers bridge and just received my Expert certificate) and I can see value in the framework. Just cut the fat away from the books and fix the remaining 20 % and there could be a really great ITIL Zero Fat.


An extra line on the Kano model


Good point. The Kano model picks up some elements of this in the "basic" curve that ceases to deliver benefit after a certain threshold, but I can imagine behaviours and outcomes that fit the pattern you suggest.where satisfaction initially follows the basic line and then swaps to follow the inverse line.


All real-world curves are non-linear - just look at Las Vegas

All real-world curves are non-linear. And in fact very very few real-world curves run away on unstable, accelerating effects as the global-warmers believe (not that it is impossible, just unusual - the IT Skeptic is a cautious accepter of the general theory of anthropological global warming (AGW)). Nearly all of them roll off as they hit some counter effect or boundary that limits the curve.

I have just returned from a classic real-world example of how more is not necessarily better: Las Vegas. I blogged on it

When it Comes to IT Value User Opinion Doesn't Matter

I got this link from @Servicesphere tweet. The articles argument is based on the fact that IT Service Desks may overdeliver service "above and beyond" in ill-advised efforts to "exceed expectations". This is true, overdelivery is not worth the effort. But it is a major mistake to believe that underdelivery doesn't cost. User opinion matters a lot if all users howl to their management about the lousy IT service.

Customer satisfaction should be optimized, not maximized. This is the lower "basic" line in Kano's model on James' blog. The article is an nice example of linear thinking.


Syndicate content