Very happy to hear that idle is better. Very unhappy to hear that usage is worse.
I would prefer not to think about my usage patterns and just sit back knowing my new laptop (2016 mbp) is always better than my old (2015 mbp) one in all cases.
The first generation "Retina" machines were not without their drawbacks either, but those were ironed out quickly enough. If you update every single year you'll be intimately aware of these problems. If, like most, you're on a 3-5 year cycle it's largely irrelevant.
People used to buy cars every two to three years as well, but now it's common to retain them for a lot longer unless you're leasing.
If you are on a 3–5 year cycle it is relevant because you would be stuck with it for 3–5 years. Unless you like the hassle of migrating to your new MacBook Pro, test it for a week, and then return it. If you are unlucky, you can’t migrate back because the newly upgraded macOS, iTunes library or Photos library is not compatible with your old macOS, iTunes library or Photos library. So you won’t be able to migrate back or open your libraries.
You have 14 days to decide you don't like it and take it back, no penalty, which is nice. Plus if you're on a 3-5 year cycle skipping a year is no big deal.
It's also unlikely you're OS locked on a 3-year old machine. My 2012 MacBook Pro still runs current versions of macOS.
The article defines "usage" as "high load". That could still mean that any fixed workload is handled better on the new MBP than the old version.
I'm pretty sure that the new CPUs are more efficient – in terms of power/workload – than the previous generation.
That would mean the new MBP is <= 25% worse for a given workload. Maybe somebody knows the numbers for the CPUs, but previous generations improved efficiency by 10-30%.
I would prefer not to think about my usage patterns and just sit back knowing my new laptop (2016 mbp) is always better than my old (2015 mbp) one in all cases.