@madsstigborg - traveling around Europe so please excuse the delays in responding.
While I have to respect everything John Nanci has done, I have to ask you the question why you'd want to roast like everyone else who follows his advice? When you're using the same bean selection and the same equipment and the same basic roast profile and same basic recipes - how do you differentiate your product?
In my experience, "exciting" roasts do not necessarily lead to good-tasting chocolate. Not all "interesting" flavors are "good" flavors that appeal to a broad customer base. Are you making chocolate for yourself or are you making chocolate that you can sell enough of to make a good business?
A short (1-2 minutes) high-temp roast to help release the shell is good, especially if you have steam. But steam forms above 100C so any temp above 100C will do the microbial kill step and puff the shells. You don't need 170C to do that. Keep in mind that you're interested in the bean temp not the oven temp and you are NOT interested in the surface temperature of the bean, you're interested in the temperature of the center. An IR thermometer won't tell you this - you need a thermometer in the mass of beans. And - something not discussed often - the temp of the beans going into the roaster needs to be known. A 5 degree +/- swing in bean temp can have a huge impact on roast.
Another thing to consider is when you start timing the roast. When you have an instrumented roaster and the right software you can understand how quickly the roaster comes to temperature after adding the beans. Do you start timing the moment you add the beans? Or do you start timing when the temperature in the mass of beans reaches the desired temperature? And 20F is too wide a window for consistent results. You might want to take a look at the roasting protocols for the Cocoa of Excellence Awards as an alternative to what John Nanci suggests.
Also - what beans? A genetic criollo or high-percentage white bean will require a different time/temp then a 100% genetic Forastero. Any general advice is just that - general. You need to arrive at something you like.
Finally - why ONE roast? There is no such thing as an "optimum" roast. The optimum roast in cocoa is a compromise. If you like a flavor at one time/temp combination you can blend it with another time/temp combination that you like. I am a fan of bring a robot when it comes to figuring out the roast for a new bean. Create a matrix of time/temp combinations and take a sample at each combination and then let everything cool down and taste all the results. There will be ones that are under-roasted and ones that are over-roasted and flavor development is non-linear. Your tasting sense is different when you are smelling the roast - which your customers do not.