So if we are going to account for all of the things making my smartphone possible as consuming electricity, I think we should account for all of the things associated with running my refrigerator such as the energy it takes to make the food I want to store and the refrigerated trucks it took to get to me.

The power utilisation of a refrigerator scales at best poorly with the quantity of food moved through it, unlike a smartphone (door openings is a far greater predictor), and human food consumption is generally constrained (1800-3200 kcal/day per person, more or less), whilst data appetites both vary and have grown markedly.

You do raise a good point that behind msny. products is a long chain of either embedded manufacturing or complementary products costs. But the avoidable costs of food vs. data (refrigerated or not, people need food to eat; we'd survived 200,000 years before iOS) argues against your comparison.

That's a false equality. Counting food would be analogous to counting the energy used to build the apps. The fact that a fridge has less features that require additional power draw don't detract from the point.

Going from that link to the atkearney report, I tend to interpret that the 19.1 kW/h per GB of data as data centre cost. If that’s the case, it would not be correct to use it to calculate per phone consumption.

The 1GB of data cost is sunk cost, whether it’s accessed by one or a million. On the other hand, if we only had one smartphone in this world, we wouldn’t be storing zettabytes for it to access :)

More suitable numbers were 3.5 kW/h for the phone, and 23.4 kW/h for each connection. So, phone consumption should more accurately be 23.4 + 3.5 + (19.1 * total world storage / total world smartphone) !

You mean kWh, not kW/h. Watt per hour would measure a change in power consumption over time.

Still an absurdly high number. fixed line internet has 0.06 kWh/GB according to this 2018 study:

https://onlinelibrary.wiley.com/doi/pdf/10.1111/jiec.12630

The majority of data heavy stuff also tends to happen in WiFi environments. So I dont know where this 300x overhead increase is supposed to come from. A typical 3G base station is burning 500W/h no matter what data usage is. So unless you share you wireless cell with less than 5 people this should be magnitudes below a fridge. The wired part of the infrastructure is much more efficient and aggregated in general.

I suppose the biggest culprut are not even the smarthones, but their cloud storage services, which are completely optional and not used to a large extent by the majority of smartphone users.

Most of all ISP profits and subscription costs would also be decisively lower/higher if energy use and therefore cost was supposedly that significant.

> 500W/h

Again, "W/h" is almost always wrong. You mean just "W", which already contains the "per time" part because W = J/s.