This is might be a helpful way to get people to care about making small, incremental improvements but the scientific basis is suspect.
There is no evidence that 20 minutes of exercise adds an hour to your life. Taking the aggregate information around years and dividing it into days is not a meaningful operation and I'm sure that Spiegelhalter knows that as he says that this method of displaying the data "seems to resonate with people".
As an aside, I've heard Dr Spiegelhalter regularly contributing to More or Less[1] and he seems to be a well informed scientist.
I'll assume these figures apply to only established daily patterns.
I find exercising lonely, painful, boring, and stupidity inducing - although I completely believe others obtain opposite results. If I had a choice for myself between exercising for 20 minutes, then dying, or dying immediately, I'd choose the latter.
The non-exercising part of my life is spent pleasantly or sleeping, so I see a net gain of 27 minutes ((60 minutes life extension - 20 minutes exercising) * 2/3(awake factor)) for the first 20 minutes/day of exercising, but a net 7 minute loss for the next 40 minutes ((30 minutes life extension - 40 minutes exercising) * 2/3(awake factor)).
Even if you don't like exercise, it isn't necessarily a loss. Have you considered the potential quality-of-life improvements you could enjoy during the time you aren't exercising? E.g. better mood, better health, more energy/vigor.
Having seen some of Dr Spiegelhalter's talks in person, he most certainly knows that the units are not based on a meaningful operation.
He is extremely passionate about explaining uncertainty and risk to ordinary people. As an example, plenty of people think that "you either get cancer or not, so it's 50-50, and might as well leave it to chance". In other cases, politicians ask scientists for the "worst possible outcome" of an event, and then react as if it was a certainty or at least 10% probability, while in reality the risk might be extremely small.
Most people have trouble with the very concept of uncertainty. Microlives, and their opposite, micromorts (1-in-1M chance of dying while doing something) are helpful in comparing improbable-but-terrible events in terms of certain-but-insignificant, easier to understand concepts.
There is no evidence that 20 minutes of exercise adds an hour to your life. Taking the aggregate information around years and dividing it into days is not a meaningful operation and I'm sure that Spiegelhalter knows that as he says that this method of displaying the data "seems to resonate with people".
As an aside, I've heard Dr Spiegelhalter regularly contributing to More or Less[1] and he seems to be a well informed scientist.
[1] http://www.bbc.co.uk/programmes/b006qshd