|
>Just charge customers by the gigabyte like a mobile network or a satellite network. That's silly on multiple levels (and I'll note Starlink doesn't do any such thing either). The marginal cost of transmitting a gigabyte or terabyte for that matter is effectively zero. Some minuscule fraction of a fraction of a watt. What costs money is bandwidth, priority bandwidth when there is contention, lowest latency bandwidth, 9s of uptime, and support. What's actually needed is for ISPs to be required by law to spell out the SLA for everyone, same as an SLA you'd get with any normal commercial service, and for all shaping to be perfectly neutral to what data is being carried, and that customers themselves may do so themselves as they wish with standard tools (like using PCP fields per 802.1p in the tag control information) within the raw limits of their service agreement. >It is misleading in the first place to advertise certain speeds on an underutilized network. The phrase you're looking for is probably "over provisioned" not "under utilitized", and there isn't anything misleading about it per se. Most users have extremely bursty workloads, and they can indeed get what they're paying for when they want it in general. It makes lots of sense for them to split the cost with a bunch of neighbors, everyone gets what they need for less. The lack of transparency is the big issue, markets don't work without information symmetry between sellers and buyers. With real leased lines everything is spelled out, and that's what should happen for consumers too albeit naturally at a much lower level of service. So say you sign up for "1 gig service" and it'll spell out you have 1 Gbps maximum bandwidth which is promised on average 67% of the time (16 hours a day), 125 Mbps guaranteed floor bandwidth (ie., 8:1 overprovision), 1 Mbps low latency bandwidth (which you could assign to a VoIP vlan say), maybe in some cases a burst capability (can go to 2.5 Gbps for 2 seconds per minute), and response for broken connections within 72 hours. Or whatever the numbers are for any service, but all spelled out, measurable by the customer, neutral to customer content, and directly comparable between services. Customers don't need to necessarily use all that or care, but if they do they can make sure the ISP is doing as promised, and the ISP can match its costs to what it charges in a clear way. >This is like running an all-you-can-eat buffet and complaining when a fat guy shows up and eats you out of business. It's a flaw in the business model they CHOSE to implement. Physical analogies are often bad when applied to tech. |
There were issues in Australia with the rollout of the NBN where the wholesale charging mechanism was basically designed to give structural kickback to back haul fiver bandwidth providers (including the national incumbent Telstra) as a way to get everyone on the negotiating table. So instead of needing bandwidth to a couple dozen places ISPs needed it to over a hundred, and we ended up with a significant fraction of the ISP market being throttled by ludicrous “contention ratios” I recall some notorious ones being over 100 to 1 but I’m on mobile so it’s a bit hard to go digging for historical links. (Whirlpool.net if you want to learn more)
The end result was a generation of frustrated young consumers did actually learn a little bit about this due to it being a widespread issue that even helped some companies differentiate their product in the market by beginning to advertise their contention ratios and it genuinely did help them get customers, but it likely would have just been “confusing” a decade early had the issue not become prominent enough that people had some idea what this thing meant when they were selecting an ISP.