3 ms·
In this debate, a lot of people - especially those discussing "paying per GB" - that seem to be missing a key issue. Sometimes this is intentional, but I suspec
by pdkl95 12y ago
In this debate, a lot of people - especially those discussing "paying per GB" - that seem to be missing a key issue. Sometimes this is intentional, but I suspect that is a fairly small minority.
The central problem: trying to apply the concept of money and economics to packets doesn't work, because money and economics are tools for handling scarce resources, while the major benefit of the modern "digital age" is that you can copy those bits without a per-copy cost[1]. As bits are not scarce, "pay per GB" becomes a case of applying to wrong tool or trying to solve the wrong problem.
Observing that the infinite scalability of bits isn't leading to free high-speed internet for everybody, a better question is: which resources really are scarce? These are the costs that billing should be based upon.
With networking, those costs are things like creating the network hardware, installing it, and maintaining that hardware. The first two are a one-time cost. We usually distribute that kind of high-variance, unpredictable cost into "monthly payments" in a variety of industries, and tthe maintenance costs are inherently a "per-unit-time" cost as well. This includes the expensive routers that you need to have anyway to be able to tolerate variations in usage[2]. At no point does the contents ("number of packets") enter into this as a cost.
A common rebuttal to the idea of usage not being a cost is to point out the problem of saturation and how an oversubscribed line would the ISP to make expensive upgrades. This problem does exist, and should be part of the billing... because it is a problem about bandwidth, not bits. You don't get to keep billing someone indefinitely for a one-time cost. Attempting to do make up costs like that would be rent-seeking[3].
Various ISPs have shown that billing for a given amount of bandwidth (at some defined defined guarantee of service) can easily be profitable.
[1] Electricity costs are not relevant, and are often not relevant anyway in physical layers where you have to send an empty carrier if there is no data. Most of the rest of the "costs" are associated with using those bits (see: CMOS), which is addressed above.
[2] The specific amount of tolerance required being defined by the guarantees (sometimes with SLA) offered by the ISP.
[3] http://en.wikipedia.org/wiki/Rent_seeking http://en.wikipedia.org/wiki/Rent_seeking
- rlpb 12y agoTo add to this, 95 percentile billing seems to work very well in the UK. But since this isn't useful to an individual end user, this requires the last-mile provider to be required to resell wholesale access under 95 percentile billing, and for multiple ISPs to compete buy under this cost structure and resell to consumers under price structures that work better for them. This works very well in the UK. We have no "net neutrality" problem as a result. [1] http://en.wikipedia.org/wiki/Burstable_billing#95th_percentile http://en.wikipedia.org/wiki/Burstable_billing#95th_percenti...