Downsides to harbors:
1. They take a long time to build. In addition to whatever you aren't building during that time, this means that you're losing the trade route income for all of these turns.
2. If they require ships to protect the route, the upkeep costs of the ships will be far higher than the upkeep cost of the road.
3. They provide no mobility bonus.
4. They can't be upgraded to railroads in the late game.
5. You have to build one in the capital too, which is an extra 3 gpt that you haven't mentioned.
Trade route income is approx 1.25 gold per population point in the non-capital city, so a size 8 city could take a road up to 12 tiles long and still break even. Granted, making a profit is preferable, but it's not like roads are going to destroy your economy. With a bit of parsimony, you should always at least break even on (trade income)-(road maintenance).
I really never found it a problem, but in order...
1) This is true, but the same could be said of the worker who is building them...if he's building the road network then he isn't building improvements for your land...it's difficult to quantify, so I can't claim for sure that it evens out, but it is true to say that for ever worker building roads, you have lost out on some food, some production and some gold.
2) Again true, but a small navy has other advantages as well, such as protecting those sea resources that could be providing you with 10 happiness (pearls and whales), not to mention the food and gold from the fishes. Again, difficult to quantify...as an attack by pirate ship (of which I'll agree there seem to be fewer in this version) may or may not happen, and I have no idea how you would accurately calculate your losses if it were to happen. I just think that you should be building at least a small navy.
3) True...but then
I think that the extra money I had by not building a huge road network payed for the extra troops that kept me safe early on. Because my early strategy resulted in a strong economy when I did start to expand, and tech took me to the midgame I had the money to pay for the road network for the larger empire. My first game taught me that if you connect everything up with roads, the good economy is harder to achieve. [I'll be the first to accept that this may not be a fair test, as there probably were other factors that resulted in my strong economy, but I'd have to have played more games to comment on those.]
5) ? Are you sure? I probably built one in my capital, but I never noticed that I did.
4) In my most recent game I didn't need to get that far. Once I had got electricity I drove a straight line for Globalization in order to build the UN and go for the win. Had I not chosen to go for a diplomatic win...well, I probably would have joined up my cities with rail for the end game...but, by then I had most of my tiles improved and the income per turn to pay for the rail network.
My original point was this...being conservative with your road building in the early game may, and I repeat,
may (with no strong empirical evidence presented) be a factor that contributes to a strong economy, and thus a better mid and late game. Harbours, IMO, are part of the process that makes this happen.
It worked for me in the game I completed last night. I'm sure other people have a different point of view on this. It seemed right though.
The other thing I benefited from in my just completed game was to have a corp of workers standing by to pull up roads and rails as my military forces swept across the Aztec continent. Reason: I was razing everything to the ground, but this takes time...and so you are technically the last owner, and thus the person who pays for the roads, even, I think, after the city itself has been razed, certainly whilst that process is occuring. If I'm right about you still paying for the roads after the city has been razed there is a definate advantage to doing this in terms of cost per turn for roads that are perfectly pointless. Don't allow the razed continent to cripple your economy.
Anyhow...that was just my two pence worth.