3 ms·
In my comment, I said that asymmetric connections are not a network neutrality issue - by which I mean that proposed network neutrality laws don't / shouldn't c
by A1kmm 16y ago
In my comment, I said that asymmetric connections are not a network neutrality issue - by which I mean that proposed network neutrality laws don't / shouldn't cover them, so the article is bringing up a strawman issue.
Options 4 and 5 are network neutrality issues. If they are allowed, it is likely that they could be more cost effective for ISPs, and become the only options. Competition doesn't work for things like this, because most consumers don't understand the benefits of network neutrality.
However, deviations from network neutrality are bad for everyone in the long run, because it means that ISPs become gatekeepers, and slow innovation on the Internet.
If I were to use Skype, I would pay my ISP for traffic between my computer and Skype's supernode, and the owner of the Skype supernode pays their ISP for the traffic as well. The only time traffic isn't paid for is generally when it gets to the tier 1 networks, who don't charge each other because the exchange of traffic is mutually beneficial.
Given a fixed level of network hardware, bandwidth is a scarce resource, and economics provides a very good way to manage scarce resources - charge for them. This is exactly what happens at present. There is no need for any kind of ISP control over what sites are accessed or what protocols are used - and if too many people are using the bandwidth, ISPs can simply put up the price.
ISPs undoubtedly oversell bandwidth - but that is not an insurmountable problem for ISPs if there are network neutrality laws - they can either oversell bandwidth by less, or just stop overselling.
- grandalf 16y agoAs one of the other comments pointed out, ISPs could just charge for bandwidth. The current (speculative) pricing schemes and hypothetical non-neutral pricing schemes are simply alternatives to pricing per megabyte. Why? Because ISPs believe that their product appears to have moe value when it's sold as x megabits per second for y dollars rather than x megabytes for y dollars. Since most customers can't estimate how many megabytes they use each month, it makes little sense to price the product that way. Whereas many customers know that if they currently have 10 Megabits, if they upgraded to 20 megabits their downloads would be faster. Comcast's boost pricing is evidence of the way ISPs are evolving to meet consumer demands. They give you lots of bandwidth for just long enough to let you download a song or a movie, but don't constantly make it available to you. This creates the perception of faster speeds for most customers.