two solitudes: Alan Sawyer's views on the media industry

Saturday, May 27, 2006

On toilets and the internet

When we first moved into our current house, we were on a flat-rate water plan. There was no meter and we paid an amount based on the number of sinks, toilets, bathtubs, washing machines, etc. that we had in the house. What we paid had no bearing on actual consumption. The previous owners were a medium-sized family whereas we were just a couple. The flat-rate plan was probably a great value for them since, at that time, there was only one bathroom serving 5 or 6 people. From a bathroom perspective, they paid based on 1 bathroom sink, 1 toilet and 1 bathtub regardless of how many people used them. When we bought the house (and prior to some badly-needed renovations) we, too, paid for 1 bathroom sink, 1 toilet and 1 bathtub, although our usage was theoretically only 2/5 or 2/6 what the previous owners’ usage had been. Had we stayed on the flat rate plan, when we added the downstairs bathroom, our monthly flat rate would have gone up. I’ve yet to figure out how having two toilets in the house instead of one could possibly increase the water consumption (barring leaks, of course), but that’s the way it worked. We now pay on a metered plan and save money (but others doing the same switch might have ended up paying more).

So why the heck am I rambling on about toilets? Well, the initial high-speed internet business model, formed 10 or so years ago, was based on a single high-speed connection serving a single computer. Some savvy users did connection sharing or had cable/DSL routers, but for the most part it was 1 connection : 1 computer. There may have been many non-concurrent users of that computer, but it was still 1:1. Just like having only one toilet... many potential users but only one at a time (please!). As well, (and I’m speaking in relatives terms here, compared to today) there really wasn’t that much content out there. So even when 2, 3 or even 4 computers in the home were sharing the connection to the internet, the volume of traffic was reasonable. Today, though, that’s changed. There’s much more content out there, some of which leads to very large downloads, and overall leads to much more traffic. And more traffic means greater costs for the ISPs. Traffic shaping / packet shaping is done to throttle back the QoS on some types of traffic (e.g. BitTorrent). While this may seem arbitrary and punitive to some users, the network operator would argue that it’s necessary to prevent overloading of the network. I certainly wouldn’t want my e-mail to slow down because all of my neighbours were watching video streams, downloading torrents and running 7x24 webcams for the benefit of their friends and family in Cardiff.

So, what are the options available to the ISPs? Well, I can think of the following:

#1 Flat rate
#2 Flat rate based on number of connections
#3 Metered
#4 Metered time-of-use

… and various combinations and permutations of the above. For cablecos, #2 is a familiar model. How many of us pay (or for that matter, don’t pay) our cable operator for additional outlets in our homes for our additional TV sets? Would we do the same for the internet – pay an extra fee for each computer we wanted to connect? Not likely, but maybe. Certainly a lot of folks wouldn’t. [Interestingly, though, it would probably be much easier to track this sort of ‘theft’ than it is to track similar ‘theft’ of analog cable signals.]

What about the basic flat-rate model – that’s essentially what most of us have today. Some folks are extreme bandwidth users while others aren’t. The blinking lights on my cable modem (well, actually, the non-blinking lights) tell me I’m in the latter category most of the time. Flat rate models inevitably mean that light users subsidize heavy users. Do I want to see my rates go up to subsidize someone (let’s call him Joe) who’s running full out all the time? Absolutely not. I want to pay a rate that is appropriate for my usage patterns and not subsidize Joe. Joe, on the other hand, does not likely want to pay a rate that actually reflects his usage – he’s got a good thing going now and he knows it.

And what about time-of-use metering. I haven’t heard anyone suggest this, but I think it’s an obvious thing to consider. Like any network (the telephone network, for example), there are peaks and valleys. Operators must build capacity to meet peak demand. If they can incent the customer to help balance the demand by shifting some traffic to off-peak times, it could be a win for the operators and the consumers. Download all the torrents you want at a low, low rate per GB, but only between midnight and 6 AM. Outside those hours – pay a premium to do so. Ironically, this is much like the old long-distance telephone billing model.

I think in the end we'll see a billing model that is similar to that used for airtime in the cell phone world. Depending on your anticipated usage pattern, you will subscribe to a plan that allows you to download as much data as you want up to a pre-defined limit for a fixed price and pay extra for everything beyond that. And, like cell phone plans, where evenings and weekends are cheaper/free, we will proably see something similar for data (although the reduced-rate windows will almost certainly not be the same). Tiered data plans are nothing new -- we see them today in the wireless data world (and no doubt will see combined wired and wireless data plans in the future -- soon, too, I hope! Anyone at Rogers listening by any chance?)

Electricity consumption has peaks and valleys, too, and electric companies are switching to time-of-use metering. We actually used to be on a voluntary time-of use electricity metering plan but it was discontinued. It saved us a lot of money, just by simple things like shifting our laundry to the weekend, running the dishwasher after 11 PM, and so on. Now time-of-use electricity plans are returning and this tine, they won’t be optional.

But I digress… back to the internet. Personally, I don’t see the flat-rate model surviving, and neither does the CFO of Rogers Communications, Bill Linton. At Morgan Stanley's 11th Annual Media & Communications Conference Linton predicted a move to metered billing. “The objective here is not to restrict usage. The objective here is for people to use this service as much as they want. We just want them to pay a reasonable amount for what they use. Usage is going up and it costs a lot of money to produce that capacity.”

So... is metered internet usage a good thing for the consumer? It depends. For some it will mean savings or at least cap rising costs. For others, it will cost more. Is metered internet usage fair thing for the consumer? Absolutely. You should get what you pay for and you should pay for what you get.


  • You are only looking at only one side of the issue. It is obviously the ISP's perogrative to using any pricing model that they see fit.

    I would remind you though that in North American, local telephone service is essentially unmetered and unlimited. It seems to work fine. In Europe one generally pays by the minute.

    The real threat of tiered service is discriminating against content providers by charging them a fee to reach internet users. I find that particularly troublesome. Since I already pay for the bandwidth, that model basically entails throttling non-preferred suppliers. It stinks.

    By all means, charge people for the bandwidth they use but do not act as gatekeepers to the information that we access. That will kill the internet as we know it and lead to its monopolization by big media. Ultimately, we will be poorer culturally and economically if we go that route.

    By Blogger Steve, at 7:44 PM  

Post a Comment

Links to this post:

Create a Link

<< Home