In this answer, im going to assume that the partners that the partners that the OP, or the communication target, does have some business relation to, is enroute without any tampering, or is a government of that location, is trusted. This means, in the bank example, the following parties is fully trusted:
- The OP's own network and all parties involved in that, for example his family.
- The OP's landlord, which has access to the communication lines in the building.
- The OP's ISP, and all ISPs up to, including the backbone operator.
- Any government agency operating in the OP's location.
- The Banks ISP, and all ISPs up to, including the backbone operator.
- The Banks landlord, which may, or may not, have access to the communication lines.
- The Banks own network and all parties involved in that, for example IT admins.
- Any government agency operating in the Bank's location.
- And finally, any government agency operating in any country that is enroute to the target location, considering that any tampering of any routers enroute to the target location has NOT happened, even if by request by that government.
On security, I also assume that:
- Any equipment owned by OP is secure.
- Any equipment owned by bank is secure.
- Any equipment owned by government is secure.
The reason I make these assumptions, is to isolate the cases where a eavesdropping or modification of traffic poses a security risk, rather than just being a nuisance for the user. If a government intentionally sniffs up OPs credit card number, it does not pose any risk for the OP, because the credit card number will not give the government any greater access, the government could with one single phone call seize all the money from OPs account anyways. Same with the ISP, the ISP who modifies traffic to insert advertising, is not gonna make any use of OPs facebook password anyways.
Another important thing is to assess if a organization is able to keep their equipment safe from external attackers. Here, I assume the OP is knowledgeable and can configure his own routers and firewalls securely, thus I set the OPs equipment to secure. This means any equipment owned by the OP cannot be remotely compromised. The banks equipment is obviously configured securely. And the governments equipment used to store eavesdropped data or similiar info, is of course configured even more securely than the bank.
Considering this trust assessment, this means that there is few attack vectors to attack HTTP over a wired network.
The only attack vectors I can see is the following:
Eavesdropping on publicity accessible wiring. An example would be listening on ADSL signals from a air-hanging telephony wiring.
Breaking and entering into local equipment rooms. For example equipment rooms owned by the ISP or the landlord. Larger equipment rooms serving a larger area, normally have severe alarm protection, guards nearby if not onsite and high security, but local equipment rooms are often weak security, sometimes a locked rack cabinet in a cellar with exposed wiring that can be spliced into and eavesdropped without affecting the lock on the rack cabinet.
Remotely hacking into local equipment. Landlord equipment for smaller landlords are often configured insecurely with bad passwords, in some cases this also apply to ISP equipment that is local. A router that OP hired from the ISP and does not have any administration capabilities by the customer, might also be suspecible to easy hacking.
Eavesdropping on wireless links with weak/no encryption that the OP cannot control over. This can apply to a point-to-point link set up by a ISP to bridge a area that is difficult to route any physical cables over. This can also apply to certain mobile links.
BGP manipulation by annoucing routes for networks that somebody don't own. This is the only risk actually worth considering. Some ISPs may also use methods that prevent unauthorized people from annoucing availability for the ISPs networks. Such security solutions may include, but are not limited, to restricting BGP traffic from any customer owned equipment.