Fair Use Policy

Owen DeLong owen at delong.com
Thu Aug 23 01:17:37 UTC 2012

On Aug 22, 2012, at 17:41 , Jimmy Hess <mysidia at gmail.com> wrote:

> On 8/22/12, Bacon Zombie <baconzombie at gmail.com> wrote:
>> I how you are talking about 3G or there is a typo.
>> An ISP with a 5GB cap that is charging the end user more then 5$ total
>> {including line rental} a month should not be allow to operate.
> I don't believe $5 even covers an ISP's typical cost of having a line,
> let alone getting data through it, maintaining, supporting it, and
> providing upstream networking.  Last I checked you can't even buy
> dial-up services from national ISPs for that low a price,  before the
> per-Hour usage charges,  and those require simpler less-costly
> infrastructure to maintain for the ISP.

I agree that $5 is low. Should be more like $30.

> With residential broadband, if there is not a heavy degree of
> oversubscription,  the ISP will either go broke, or the cost of
> residential service will be so high that the average person would not
> buy it.   "I want my line speed 24x7"  is a technical argument,  it is
> a numbers game, and  the average subscriber does not make that
> argument,  or at least,  rather, the
> average res. subscriber is not willing to bear the actual cost
> required to actually pay
> what it would cost their ISP to satisfy that  for every user trying to
> utilize so much.


> Why should the end users who transfer less than 1GB a month, with only
> basic web surfing, have to suffer periods of less-than-excellent
> network performance  or pay increasing costs to subsidize the purchase
> of additional capacity for users at the same service level expecting
> to use 100GB a month?

Here we disagree -- somewhat.

If you are selling an unlimited product, then you take on the obligation of
making it effectively unlimited. If you are selling the same unlimited product
to the 1GB/Month user and the 50GB/day user, then you are not doing a
good job of pricing your products in a manner that is fair to your subscribers
and you will lose in one way or another:

	+	Your product is too expensive and your lower-end customers
		depart and you're left with only the few high-end customers.
	+	Your product is not expensive enough and everyone gets poor
		service because you do not have adequate upstream bandwidth
		to service all of your customers.

The way you achieve fairness is to provide different products at different
price points to meet different customer needs.

For example:

	Metered products for low-tier users:

	1GB/month			$x
	2GB/month			$2x
	5GB/month			$4x
	10GB/month			$8x
	Over				$0.9x/GB

	Flat rate products for power users:

	10Mbps/1Mbps		$y
	30Mbps/5Mbps		$2y
	50Mbps/10Mbps		$3y
	100Mbps/20Mbps	$5y


Note that there's no unlimited product in the list, but, the flat rate
products present limitations in terms of wire speed while the metered
product allows low-utilization customers to take advantage of high
wire speeds for short bursts of transfer.

That way the customer can choose the pricing model and service
that best meets their needs and all you have to do is make sure that
the values of x and y are sufficient to allow you to upgrade your
capacities as needed to meet the demands of increasing customers
and/or customers moving up the usage tiers.

> There is a certain degree of fairness there.

Not really. It assumes a single one-size-fits-all pricing model which
is bound to be inherently unfair.

> Even if   the metric is wrong --  the idea of metering bytes
> transferred is broken,
> because it does not positively reinforce the good behavior.

While there is truth to this, the bottom line is that selling bandwidth
based on time of use billing to residential customers results in a
product that the consumer simply can't/won't understand and you
can't make any money.

> It's like trying to reduce congestion during rush hour on the freeway
> by imposing a  "40 miles of travel per day"  limit  on each vehicle
> owner.
> That gives no benefit for those effected by the limit to adjust what
> time of day they travel those 40 miles,  however.
> A  "X=10 gigabyte per  4 hours"   rolling average  limit  would make more sense.
> Where "X"  is varied,  based on the actual congestion of the network between
> other users of the same service level.

Try explaining that to the average residential user... It's just not going to work.
The incentive won't affect the behavior because they just won't understand.

>> And if your infrastructure and handle 25% at a minimum maxing out their
>> connect them don't advertise " unlimited " since you can't provide it and
>> it is false advertising.
> There's no such thing as unlimited, period.    Even if the provider wanted to,
> there will be some physical limits.

True... I object to the use of the term unlimited as well.

> I agree the use of the word is confusing... when they say unlimited
> what they are
> often indicating is  "You are not  limited  by the provider in the
> number of hours a day you can be connected to the service".

Which is meaningless if being connected doesn't mean you can transfer
data at some reasonable speed throughout those hours as well.

>> The world would be a better place if ISPs that either throttled, cut off or
>> added on extra charges to the end users bill were fined to hell for false
>> advertising and repeat offenders were named and shamed on a public website.
> [snip]
> There might be no residential ISPs left

I doubt it. I think that if they had to, residential ISPs would, in fact, find a way
to develop pricing models (like the one I mentioned above) that fit that criteria,
still allow them to profit, and also pull in enough revenue to build out capacity
as needed to meet demand.


More information about the NANOG mailing list