It's an interesting review, however I'm a bit puzzled regarding the city-state paragraphs. He talks about how powerful things are like the maritime ally food bonus, vs. making granaries and watermills yet neglects to mention the amount of gold it takes to ally a city-state. So he should at least acknowledge the trade-off. Maybe he did crunch the numbers and find that gold is best spent allying the maritime city-state but it isn't clear to me *shrug*
1000 gold puts you anywhere from 30 to 60 points above "Allied" with a city-state depending on your initial standing, social policies and the game speed, I think. Let's assume 45 for the sake of argument, and let's also assume that it decays at 1 point per turn even though in practice, if you're bribing city-states you will likely have reduced this to 0.75 or less through policies and/or civ traits.
So, 45 turns of Allied bonuses from city-states, and 30 points of Friend bonuses, for 1000 gold. Fair?
Consider Allied Maritime city-states, since they're widely considered to be the "best". "Allied" means, if I recall rightly, +2 food to all cities, +4 to the capital. "Friend" means +1/+2. This is equivalent to a free Granary for 45 turns (+Watermill in the capital, but never mind), since every city can in principle build one. It is therefore equivalent to 1 gold/turn per city. There are no +1 food structures (unless you run a Lighthouse in cities only working one Sea tile, which is kind of a waste unless you've got some prime real estate that can't be worked otherwise), so let's just call the "Friend" portion of that 1000-gold deal as extending the 1 gold/turn per city by an additional 15 turns.
Effectively then, 1000 gold buys you the equivalent of 1 gold/turn per city for 60 turns. By this metric, you recoup your cost if you have 17 or more cities, which is typically a ridiculous number without careful city micro-management and policy selection. Teching Patronage gives 15 more turns of Allied status, effectively reducing this to 13 cities, which is still kind of bad if you consider things in terms of Granary cost.
However.
Unless you suck at managing happiness, Granaries are a profitable investment. You will want one in each city. It's more reasonable to place the Maritime food bonus on par with Watermills, which can only be built in riverside cities and which are less palatable on account of costing 2 gold per turn to maintain. In this case, straightforwardly a mere 8 or 9-city empire (6-7 with Patronage) begins to profit from having invested in an alliance with one Maritime city-state. All the more since not every luxury resource you want is going to be nicely located within range of a river tile, and so not every city can build a Watermill.
Add in the fact that it's good practice to throw another 1000 gold at a city-state *before* you fall out of Allied status, which buys another say 90-120 turns of the best goodies, after the initial investment, you really do begin to see profitable returns from the arrangement. And, since this effectively amounts to a "free" citizen in every city for every Maritime city-state you are allied with, if you can utilize those free citizens in a productive way in most of your cities, all the better.
IMO the only real issue with the city-state system is that it's too easy to exploit. Throwing gold their way shouldn't be sufficient. There should be more "quests" to gain favor with city-states and it should severely damage relationships with one or the other city-state if you have dealings with both and they are calling for their mutual destruction.
But it is strange that , after civ IV being quite effective in quenching ICS, we have a civ V that brings it back while advertizing the exact oposite 
And I'm pretty sure that pretty much no one would pay what we paid for a civ III with leaders that curse you in their mother language 
