Why does the author seem to think that a digital solution is required for control over temperature?
The bimetallic approach can be extended to select temperature. An analogue dial can adjust the switch cutoff position (or alternatively, the starting angle of the bimetallic strip) with a screw. All you'd need to do is calibrate the dial, and the added cost shouldn't be more than 50 US cents on the end retail price.
IMO the reason a "good" kettle costs $90+ is because there isn't enough demand to create enough competition and support enough production to amortize the fixed costs. It's essentially a market broken apart into two of Porter's three fundamental strategies, cost leadership and differentiation. The cheapest kettles would be uncompetitive with a temperature selector even if it added only pennies, while the differentiated kettles have specific features for smaller, less cost sensitive markets. Most people simply don't need temperature selection.
I believe the reason he's going for a PID is the ability to reach the target temperature without overshooting. A PID would gradually reduce the power to the heating element as the temperature get closer to the target.
That said, the adjustable bimetallic solution is certain worth a try, it might just work good enough and be far cheaper and much more reliable in the long run.
I don't think he can gradually control the power, just turn it on and off. A thyristor or something for kilowatts of power would be expensive. This makes a PID controller excessive: just heat until it reaches the target temperature and then turn it off. Turn it on again when the temperature has dropped a couple degrees if you want it to keep a temperature.
I do not know about pricing, but they're widely used for espresso machines. That said, my point was merely that it allows for a much more precise control of the temperature - not whether or not it was economical feasible or provided enough value for the money.
Digital thermostats work on electrical resistance which is more accurate than bimetallic thermostats. Also bimetallic thermostats basically only switch at one temperature whereas most digital thermostats can handle a range. If you intend on using a kettle for both coffee and tea, that range is desired since the ideal temperatures between coffee and tea is quite significant.
Pop-disc bimetallic thermostats switch at only one temperature, but the property of bimetallic expansion is in no way limited to switching at a single cutoff point.
I'd be willing to make a large bet that a majority of people who brew both coffee and tea do so using boiling water or close to it. Black tea (breakfast tea) is best made with freshly boiled water, while I find coffee made with anything below 90C to be too weak - I generally use water made shortly off the boil when I'm using an Aeropress.
I have zero doubt digital thermostats are more accurate. That wasn't the point.
Thermocouple, RTD, Bi-metallic strip/coil, Silicon Bandgap, Thermistor, and I'm sure there are a few common types that have slipped my mind. Your statements are a gross over-simplification.
The bimetallic approach can be extended to select temperature. An analogue dial can adjust the switch cutoff position (or alternatively, the starting angle of the bimetallic strip) with a screw. All you'd need to do is calibrate the dial, and the added cost shouldn't be more than 50 US cents on the end retail price.
IMO the reason a "good" kettle costs $90+ is because there isn't enough demand to create enough competition and support enough production to amortize the fixed costs. It's essentially a market broken apart into two of Porter's three fundamental strategies, cost leadership and differentiation. The cheapest kettles would be uncompetitive with a temperature selector even if it added only pennies, while the differentiated kettles have specific features for smaller, less cost sensitive markets. Most people simply don't need temperature selection.