ietf
[Top] [All Lists]

Re: on UCE: Possible Interest (fwd)

2002-08-20 00:39:17
This thread is missing a key element to the way we all pay for our use of the Internet.

This latest proposal to set up a clearing system to settle payments among senders and receivers would put the Internet back in the Telecom era of billing settlements, where we learned that the customer's cost of phone service was loaded with 70% allocated to billing and settlement system operations.

Going back there suggests that our ISP services will cost everyone (repeat: EVERYONE) 100/30 = 3.33333 times as much as now. I suggest it is not worth this much as a spam cure. Especially since all this extra expense goes to our green eye shade friends in the billing departments.

For one, we all pay for all the mail we send, and receive, from some mystical midpoint between the sender and the receiver. It is inherent in the business models of ISPs, which charge me (and you , and you , and you, and all) for all traffic of all kinds from me to the mystical middle of every connection caused by me and every connection accepted by me.

The "mystical middle" is really not a mystery to understand; It is just impossible to always know where it is. First, think in terms of local phone calls:

If I call someone on a "local" number, we both pay for half of the call with our flat rate charge for local service. If I make a lot more calls that you make, my rate/call is obviously less, but we are both paying our share of every call, none the less.

But, if I make a non-local call I get to pay time and charges on my end, and the phone company has to share those charges with the delivering phone system, which is the root cause of settlement payments, just because all the charges are collected on one end, or the other, but not split between sender and receiver as in the Internet.

So, In the Internet, what I am calling the mystical middle is that point between us where our different ISPs hand off to each other as peers, which tends for many connections to be somewhere in the realm of the primary level backbone IP networks, which service lower level ISP on some traffic volume basis (other then peering).

Thus, the logic of peering is that the backbone IP carriers are getting paid for all traffic in both directions by their customers, and thus there is no such thing as a need to "settle".

And there is no directional volume differential to argue about. There is no easy way to balance traffic in both directions in any case, because most protocols call for unbalanced traffic. Email is mostly outbound from senders, and incoming for receivers. WEB surfing is just the reverse.

The only people hurt by spam then are those who pay by the minute for their receiving traffic, so these people would like to have an ISP that offers them a way to filter incoming traffic for spam, and other unwanted mail.

It has taken many of the IP backbone carriers a long long time to figure all this "no settlement stuff" out, though it has been clear to many of us for many years now. The Old Bell Heads took a long time to figure it out.

The people who care most about spam are those who pay some fee by the minute for downloading mail they do not want to receive. I expect they would like their ISPs to offer them POP and IMAP accounts that can filter out the spam with Black and White filter lists. But, the black and white lists need to be controlled by the ISP customer, not by the ISP.

Those of us with DSL or better service do not so much care about spam wasting bandwidth, but we are bothered by needing to sort through the cruddy spam. So, an arrangement where I get to bill a bunch of unknown spammers 5 cents per message is not worth using because they are not going to pay those bills in any case, and I am not going to spend money on suing them in China or Russia to get paid.

So, here is what I do.   Building the filters is manual as I do it now.
I have invested in about 700 filters, many of which are now obsolete, so I figured out a better way to do it with black and white filter sets.

I have a white list and a black list. My white list has specific accepting EMail addresses in it, and the filters do me the favor of filing all mail into a more or less correct folders. The few that land in the wrong folder get moved without a lot of problems. Beats dumping it all into one INBOX for manual sorting, since I only need to deal with the misses, which are much fewer than dumping it all in the inbox.

My black list has mostly domain names that match domain names, or something rather general that catches huge chunks of the Internet. YAHOO.COM, MSN.NET, CN.NET, NET.CN, COM.RS, EXCITE.NET, EARTHLINK.NET, MINDSPRING.NET, etc, ad nauseum.

So, my inbox stays rather clean of spam. I don't remember the last time spam got to my inbox.

The downside is that I also need to scan the trash folder for the rejected stuff in case I have missed someone in my white list, or someone changes their address, or something. But, it is interesting how obvious the good stuff is in a list of mostly spam. I hope that someone one day builds a Mail User Agent system with these kinds of tools to help users build such a system.

I do not strongly recommend that everyone do what I do, but with a little thought, I am sure you can think of something like it that will be vastly less bother than trying to collect nickels from spammers;-)...

I know of only one guy that actually collected $50.00 from some unfortunate spammer that used a valid From address, and became convinced that paying was better than getting a summons for a lot more money based on the Washington State Spam Law, and possibly being put out of his lucrative spamming business. At his hourly rate, the $50.00 of found money fell short of covering his time;-)...

So, forget all this business of solving the spam problems with 5 cent charges.
At $10.00/message, you still cannot make such a scheme work, so filtering is the only answer.

Let's hope that our Mail User Agents and our POP Mail Servers can find ways to implement some really useful black & white list filters that we can directly control through desktop of web server interfaces with good security components.

Among other things, they should provide scannable logs with rich search tools that identify what filter caused which message to land somewhere.

There might be some room for also doing some IETF work on defining some standard headers to support smarter filtering, but I have no suggestions for this. The problem is to get the spammers to use them correctly;-)...

Cheers...\Stef


At 9:20 PM -0500 8/19/02, Eric A. Hall wrote:
on 8/19/2002 8:23 PM Larry Smith wrote:

 > How very true.  Until it becomes "economically" un-profitable to send
 > spam, it will continue to both be a problem - and a growing one...

Unfortunately, experience with Usenet has already proven this to be false.
Instead, new idiots are born everyday, each of whom think that there are
millions of potential customers out there just dying to hear about their
pyramid schemes, web sites and lotions.

It's a social problem, not an economic problem.

--
Eric A. Hall                                        http://www.ehsco.com/
Internet Core Protocols          http://www.oreilly.com/catalog/coreprot/



<Prev in Thread] Current Thread [Next in Thread>