[Mailman-Users] Goodmail spells doom for mailing lists?

Dave Crocker dhc2 at dcrocker.net
Mon Mar 6 22:32:57 CET 2006


>     Dave> AOL creates a specialized, rather expensive process that it
>     Dave> provides for free, to ensure delivery of a class of mail.
> [If you want to know why Brad asks if you're an "intentional shill for
> the advertising industry," there you have it.  The purpose of the
> "process" from the point of view of the AOL subscriber is to ensure
> _non_-delivery of a class of mail.  

Do you expect your postal service to screen your mail, or your phone company to
screen your calls?

In other words, this demand that we place on the receive-side email service is 
really quite extraordinary.  That does not make it unreasonable, but it does 
mean we need to be careful about what we expect and how it is valued (and paid for.)

Given the arms-race nature of the filtering world, the cost of doing it well
increases without bound.  That's not such a good business model, when it
represents pure overhead, rather than resulting in direct revenue.  (I am 
assuming that serious discussion about these issues views it as acceptable for a 
business to make a profit.)

More generally, a serious challenge in these kinds of discussions is the failure 
to see it as requiring a 'negotiation' among the participants, including 
authors, recipients, and all the operators.  Each has different goals and 
constraints, and some of the constraints compete with each other.

So we need to approach the topic with an understanding that there are no perfect 

Equally, every set of participants has some good actors and some bad actors. Not
everyone who sends bulk mail is doing a bad thing.  Not everyone who complains
about mail they receive is doing so carefully and in good faith.

So this negotiation towards an acceptable-but-imperfect solution must worry
about misbehaviors from any direction.

And the third challenge is that this all involves a bit of very large-scale,
very important infrastructure operation.  No one -- and I mean no one -- knows 
how to do that perfectly. The absolute rule when making any change to a social 
system -- and email definitely is one -- is that will have unintended 
consequences, most of which will be undesirable.

What we have today is an undifferentiated mass of email, with a very poor signal
to noise ratio.  What the services that focus on good reputation are trying to 
do is to identify a class of senders who will have a very *good* signal to noise 
ratio, and thereby handle this safer set of mail in a way that gives better 
service, namely better delivery assurances.

Once we agree that this is a reasonable goal, then we are debating methodology.

My own view is that any new methodology for a difficult problem is an 
experiment. Some experiments look more plausible than others, but ultimately, 
none of them has a certain outcome.

The only thing that is important is to make sure that it is possible to conduct 
multiple experiments and that the evaluation of them can be done by the market.

The hyperbole and polarization, present in the current public environment, makes 
sure that serious market evaluation is not possible.

> I doubt most subscribers really
> care whether they get their "opt-in" mailings from "specially selected
> advertisers"---but those advertisers care, and care enough to pay!

This represents a good example of missing the point.  The better signal-to-noise 
efforts are not focusing on delivering "advertising" but on servicing existing 
customer relationships.  The pristine example of this is transaction mail -- and 
that has nothing to do with advertising.

Of course, companies with whom one does transactions also tend to send 
advertising, but there is a world of difference between getting that sort of 
bulk mail -- that I might not want -- from getting the masses of fraudulent and 
deceptive mail from people with whom I have no relationship at all.

At least when it is a company with which I have a history, there is a reasonable 
chance of fixing the problem.  Hence, debates about how it gets fixed are a 
matter of quibbling, in comparison to the larger and strategic issues of 
controlling spam from the really bad actors.

> Most public-service MLs, on the other hand, will not be able to pay.

There are two different lines of response to this:

1. Of course we all want to be nicer to organizations that have an altruistic 
quality about them.  Unfortunately, we need to be careful about defining this 
special class and defining their special benefits, lest we wind up giving 
benefits to people and activities that are not really altruistic.

2. These altruistic activities pay for other utilities just like the rest of us. 
  Why must they be given special pricing (e.g., free service) in this case?

These are competing points.  So, once again, we are faced with inherent 
difficulty in any effort to resolve them.

Again, it might give one a nicely warm feeling to claim that things are much 
simpler than I am describing and that they is easy to resolve, but that style of 
thinking, for the world of social behavior, is pure folly.

> Except those which are mailing solicitations for donations!

The list of exceptions is far, far longer.

And having a list of exceptions then creates a very real problem with compliance 
and enforcement.

> Who is being served here?  Advertisers, not AOL or mailing list
> subscribers.]

Except when it is not a matter of "advertising".

>     Dave> So along comes a few companies who are trying to find ways
>     Dave> to let receive-side ISPs outsource the job of assuring that
>     Dave> trustable bulk mail is, in fact, trusted.  (That is, the
>     Dave> receiver wants this stuff and these services are provding
>     Dave> ways to assure that they get it.)
> Note: trusted _by the ISP_.  The ISP should be a reliable
> representative of their subscribers, or the whole scheme is
> suspicious. 

A long time ago, I was taught that all of a receive-side provider's actions 
about spam are driven entirely by their customers' complaints.

It is difficult to get more representative than that.

> If a Goodmail token bypasses any user-defined filters, that's spam.

I recall seeing that it bypasses the AOL filters.

I do not recall seeing that it bypasses a user's filters.

> If AOL doesn't provide a way for users to define such filters, they're
> aiding and abetting, no?

I see a non-sequitor here but I am not sure I can describe it well.

1. Providing user controls for filtering is a nice feature, but not 
automatically required of an email operator.  But, yes, it's great when they do. 
At any rate, declaring that the absence of such filters is aiding and abetting 
makes no sense, especially when there is no precise statement of what is being 
aided and abetted.

2. This presumes that if user filters exist, they are are being bypassed.

   Ie, they will help the advertiser/Goodmail
> to push things as close as possible to what the subscriber(s) consider
> unacceptable. 

All of the good-actor reputation services are focused on getting mail that the 
user wants delivered.  As with any group of people, I'm sure that some of these 
services will have higher integrity than others, but their stated goal is the 
delivery of *mail wanted by the recipient*.

Hence, your premise appears to be entirely off the mark.

  In particular, they are likely to adopt a "voting
> criterion" for "unwanted," or a "wanted until specifically refused"
> criterion.  All in the name of maximizing information flow.

Most of the rep services do claim that complaint rate is the ultimate king.  I 
assume that is what you mean by voting.  But that is why I distinguish rules for 
gaining membership with the rep service, versus on-going monitoring.

Rules for gaining membership need to be visible, clear and have reasonable 
community support.  But this presumes reasoned community discussion, rather than 
the current tone of paranoia, misrepresentation and unwarranted certitude.

>     Dave> So one of these services lands some strategic relationships
> [Ie, attempts to restrict trade. :-)  That's all that "strategic
> relationship" means, has ever meant, or can ever mean.

"restricts trade" is a rather specific term of art.  You appear to be applying 
it far beyond its usual meaning.

You buy your food from a particular store, rather than from all the stores that 
sell food.  By your logic, you are restricting trade.

   If it meant
> something else, you'd call it by a more precise name: "customer-
> supplier", "competitor", etc.]

No.  I'm happy to use these terms, too.  I chose the term "strategic" because 
the relationship is necessarily deep and long-term.

>     Dave> and makes a splash announcing them. Somehow, this
>     Dave> value-added service is heralded as subversive,
> Tut, tut.  I certainly wouldn't call this service "revolutionary",
> although I wouldn't be surprised if AOL/Goodmail do!

Where did I say it was revolutionary?

(Actually, I've been trying to be pretty careful to make comments about this 
class of service, rather than about any particular company seeking to provide 
it. Further, I've been trying to talk in terms of what it takes to provide, its 
intended benefits, and other analytic points, rather than use any marketing hype 
about it.)

>     Dave> in spite of the fact that pretty much all other
>     Dave> communication services have levels of service.
>     Dave> I must be missing something, here.
> Yes.  First, you may have missed the fact that "special interest"
> above refers to "bulk emailers who have been getting best QoS for
> free".

Huh?  What bulk emailers have been getting *best* qos for free?

> Second, the key point is that, up until "spam", where carriers provide
> QoS to third parties,

I simply do not understand what you mean by "where carriers provide QoS to third 
parties" but I'm pretty sure I don't agree with it.

> Third, the fact that AOL/Goodmail are certifying mail in bulk as an
> exclusive relationship[1] rather than competitively offering various
> ways to help users filter at the mailbox level means that they are

News flash:  companies have limited resources.

For the benefit of competition, yes, it is important to have multiple players. 
But it is not unusual to have a company start with one.  (I'm not commenting on 
whether I think the one they chose is the better or worse one.)

In any event, we've been overtaken by events, since AOL ha snow announced that 
it will add some other rep services.

> (more or less deliberately) setting up a situation where they can turn
> the communication connection to a very large block of users in one
> click of a GUI.  Stretching the point to make a point, under the
> Sherman Act that's prima facie an illegal combination in restraint of
> trade.[2]

1. I don't know what "turn the communication connection to a very large block of 
users" means.

2. You've jumped to a legal assessment that seems to be missing both lots of 
steps and lots of justification.  Is this really a good forum, and are we really 
the right people, to attempt legal assessments of something like anti-trust?

> Put it this way: what is wrong with a competitive solution where AOL
> allows the _users_ to choose _one or more_ filtering service(s)?
> Besides reducing revenue to Goodmail and AOL (and maybe raising
> overhead a bit)?

1.  It's probably worth trying to design the details of such a capability, to 
make sure that path is feasible and scales well, before requiring it.

2. There is a difference between what is desired for the longer-term and what is 
possible in the near-term.  Any company that tries to do everything at once 
usually winds up doing nothing.  (Analogy:  Do you know what happens when you 
try to juggle too many balls at the same time?  You do not drop just one.)

> A couple more points.  "Other" communications services that provide
> levels of service often negotiate them with subscribers, not with
> third parties.  

Huh?  You mean the service that hosts dcrocker.net consulted me before 
contracting with their co-location service?  I don't think so.

And when is the last time a telephone company negotiated levels of service it 
would offer with subscribers? (Hint: "never".)

"Other" communications services are often classified
> as "common carriers", or outright nationalized.

and, more recently, privatized and not regulated.  these things vary over time.


Dave Crocker
Brandenburg InternetWorking

More information about the Mailman-Users mailing list