AOL and Yahoo will soon start using Goodmail, a system that lets bulk email senders bypass the companies’ spam filters by paying the companies one-fourth of a cent per message, and promising not to send unsolicited messages, according to a New York Times story by Saul Hansell.
Pay-to-send systems are one standard response to spam. The idea is that raising the cost of sending a message will deter the kind of shot-in-the-dark spamming that sends a pitch to everybody in the hope that somebody, somewhere, will respond. The price should be high enough to deter spamming but low enough that legitimate email won’t be deterred. Or so the theory goes.
What’s different here is that senders aren’t paying for delivery, but for an exemption from the email providers’ spam filters. As Eric Rescorla notes, this system creates interesting incentives for the providers. For instance, the providers will have an incentive to make their spam filters overly stringent – so that legitimate messages will be misclassified as spam, and senders will be more likely to pay for an exemption from the filters.
There’s an interesting similarity here to the network neutrality debate. Net neutrality advocates worry that residential ISPs will discriminate against some network traffic so that they can charge web sites and services a fee in exchange for not discriminating against their traffic. In the email case, the worry is that email providers will discriminate against commercial email, so that they can charge email senders a fee in exchange for not discriminating against their messages.
Is this really the same policy problem? If you advocate neutrality regulations on ISPs, does consistency require you to advocate neutrality regulations on email providers? Considering these questions may shed a little light on both issues.
My tentative reaction to the email case is that this may or may not be a smart move by AOL and Yahoo, but they ought to be free to try it. If customers get fewer of the commercial email messages they want (and don’t get enough reduction in spam to make up for it), they’ll be less happy with AOL and Yahoo, and some will take their business elsewhere. The key point, I think, is that customers have realistic alternatives they can switch to. Competition will protect them.
(You may object that switching email providers is costly for a customer who has been using an aol.com or yahoo.com email address – if he switches email providers, his old email address might not work any more. True enough, but a rational email provider will already be exploiting this lock-in, perhaps by charging the customer a slightly higher fee than he would pay elsewhere.)
Competition is a key issue – perhaps the most important one – in the net neutrality debate too. If commercial ISPs face real competition, so that users have realistic alternatives to an ISP who misbehaves, then ISPs will have to heed their customers’ demand for neutral access to sites and services. But if ISPs have monopoly power, their incentives may drive them to behave badly.
To me, the net neutrality issue hinges largely on whether the residential ISP market will be competitive. I can’t make a clear prediction, but I know that there are people who probably can. I’d love to hear what they have to say.
What does seem clear is that regulatory policy can help or hinder the emergence of competition. Enabling competition should be a primary goal of our future telecom regulation.
I do not think the “principles” of e-mail neutrality and net neutrality can be equated. One big difference is push vs. pull.
Net neutrality (at least the way the debate has been phrased in the public eye) has been about whether the ISP can discriminate against downstream web traffic; this is traffic that the user has deemed important enough for him to pull/download. An ISP’s lack of net neutrality, therefore, has a direct negative impact on the user’s service.
E-mail, on the other hand, is a “push” service where the user often gets bombarded with spam that he does not care about. It can be argued that AOL and Yahoo’s efforts — be it the use of spam filters or GoodMail — are targeted solely at improving the user experience. (It can also be argued that the e-mail service is free — and beggars can’t be choosers.)
However, a separate argument can be made that GoodMail is a particularly poor way to accomplish this, thanks to its targeting of commercial e-mail alone. (If someone tagged a message with GoodMail, it is probably a good indicator to the user that he should send it directly to the Spam folder :-))
A counter-question: If messages from Yahoo are never filtered by Yahoo’s spam filters (and they aren’t), although a majority of Yahoo’s customers consider it to be spam and train their filters by marking them that way, does that violate e-mail neutrality?
I also agree with cm that the spam problem is, conceptually at least, a solved problem. The definition of “spam†is “what I, personally, don’t like to readâ€, […]
Chris,
I would agree that noise is the undesired component of power (or energy), distinguished from signal.
Ian Brown Says:
“Perhaps, in the same way as telco regulators require telephone number portability, we need mandated e-mail address portability.”
Camel-nose-tent… with “government regulation of email” starring as Joe Camel.
Perhaps, in the same way as telco regulators require telephone number portability, we need mandated e-mail address portability. How hard is it for an e-mail server operator to set up a line in their .forward file…
As others have said it does seem irrelevant to anyone bright enough to manage their own email… but let’s be paranoid just for fun: what happens if yahaolmail declares that all their free/low-cost mail accounts must now be filtered through “MediochreMail”, as opposed to being ad-supported… or in addition to being ad-supported?
(Aside from commercial suicide, that is… 🙂 )
Hmm. I had a message before http://freedom-to-tinker.com/?p=971#comment-11198 but it seems to have been destroyed/lost. Well, I thought it was interesting, anyway…
I don’t see how this AOL system will reduce the amount of unwanted mail received by its customers. As I understand it, the status quo system remains in place; postage-paid mail is merely an added option. So the current deluge of spam will continue unabated, augmented by any paid-for mails.
Err, that should read “and any system that does not pay attention to the preferences of individual users as regards the actual content of messages is doomed to failure.” Sorry.
I agree with Edward Kuns that ISP competition is basically nonexistent. It’s a commodity market with absurdly low margins, and the only way to survive is to be the biggest/only provider in the region. In the 90s many small ISPs were eaten up by bigger ones.
I also agree with cm that the spam problem is, conceptually at least, a solved problem. The definition of “spam” is “what I, personally, don’t like to read”, and no system that does not pay attention to the preferences of individual users as regards the actual content of messages is doomed to failure. Half-baked schemes like DomainKeys, SPF, Goodmail, and so on — which try to decide message content goodness on the basis of non-content information — are doomed to fail as spam filters. And have failed. Spammers love SPF…
But cm is even more right: the future of personal information management is searching and “dynamic sorting”, not “static sorting”. Rather than having a million subfolders which we use to categorize our emails by hand, and rather than having dynamic sorting into only two categories (spam, ham), we will have delightful things like Reel Two’s Classification Engine and the free Weka on which it is based.
Different people will need to make different bandwidth/CPU tradeoffs, and different products and configurations will rise to meet those needs.
With my email, the POP3 interface does not filter, and otherwise I can switch off filtering altogether. I’m doing that anyhow as I’m convinced that my own client filtering (Mozilla probabilistic) and homegrown rules are way more effective.
As long as email services provide an unfiltered access method where users can do their own filtering, that’s the way to go. The tools are there, and will continue to improve.
I’m skeptical. Net neutrality is without doubt a vindictive power/money grab by ISPs. We’ll have to wait and see how this turns out. My guess is it will be ignored altogether.
They claim that senders will still be able to do things “the old way,” implying no serious change in the spam filters. Unless a specific company is already have issues with that filter, I doubt that the GoodMail system will have any takers.
The question is, are they trying to provide a “premium delivery service” or rake in a profit at any cost to service.
I think Ed’s optimism about competitive forces is misplaced. Sure, people can change longstanding email addresses to get around the fact that people can no longer send them mail. But it’s just about as easy to stop using email. Go to IM and SMS, or conferencing/bbs/blog systems, or back to the — gasp — telephone plus voicemail. It’s quite possible that what you’ll see is simply a further reduction in the value of email for cheap, timely comunication. (Think of all the post-Axelrod work showing that ecosystems can evolve into states displaying arbitrarily low levels of productivity for arbitrary length of time.)
One thing that this initiative has in common with other network-non-neutrality boondoggles is its implicit redefinition of the internet as a big-player broadcast-style medium. Note that it’s only bulk emailers who have the option of paying to get past spam filters. Individuals, be they the proprietors of small mailing lists, information aggregators or merely prolific correspondents, might be lucky enough to be whitelisted; otherwise they’ll just have to sit in the back with the rest of the spam.
Concerning net neutrality, I just read a report in the Washington Post about a Verizon Executive publicly stating that Google (and other internet content providers) should not have a “free lunch” by obtaining “free” access to Verizon’s and other telecommunications companies broadband pipes. The executive from Verizon is quoted as saying that the companies have to find a way to recover their investment in broadband pipe.
Google and the other content companies aren’t pushing content to end-users, it’s the end-users who request the content (by typing in the web address, for example or clicking on the link). From my view, the person seeking access to the content and requesting that it be delivered to him / her, should be the person payhing for access to the broadband pipes. User pays. In most cases, that person is the individual subscriber.
So, from whom should Verizon and the other telecom companies recover their costs? From the end-users, the individual subscriber to their services.
Any proposal to charge the content companies for access to the telecoms pipes to deliver content requested by an end-user, strikes me as akin to pigs trying to feed from both ends of the through at the same time.
Ed,
A few more technical differences:
TCP has flow control built-in. Further, it interprets packet loss and implements congestion control. While UDP has neither flow control nor congestion control, any protocol built on top of it MUST cope with forseeable (inter)network and host conditions.
People are not hosts, and in comparison with machines have drastically limited amounts of processing power for certain tasks.
As far as the policy analysis goes, in 2005, I very much doubt that either AOL or Yahoo, or both in combination can convince people to “shut up and eat your spam.” A number of years ago, I used to worry that the vast majority of new internet email users would (as McLuhan predicted) interpret email in terms of snail mail, and just accept spam until the entire email system suffered congestion collapse.
Ned,
Clearly the two technologies differ. The question is whether any of the technical differences affect the incentives and strategic options of the parties enough to change the basic policy analysis. Thus far I don’t see how they do, but perhaps you can convince me otherwise.
To me, the net neutrality issue hinges largely on whether the residential ISP market will be competitive.
Just as a data point, I have the following options:
1) Dial up
2) ISDN
3) IDSL (DSL over ISDN … $100/mo for 144kbit symmetric)
4) cable modem (finally) from Comcast
5) DSL from SBC
When you exclude options that are dial-up speed and slower, and you exclude options that are > $100/mo for 144kbit or slower, you end up with two choices: cable modem from exactly one provider and DSL from exactly one provider.
Why DSL from exactly one provider? Well, I live 21000 feet from the CO, but SBC blew fiber to a remote terminal less than one mile away from my house. They are required by law to share COs with other DSL providers, but are under no such requirement to share remote terminals. I see them building fewer COs and more remote terminals.
For high speed access, I have a choice between exactly two ISPs. That is not enough to ensure a market that is competitive. Evidence seems to show that you need multiple cable providers or multiple DSL providers before effects of competition show up.
In the short term, I am not hopeful about a competitive ISP market. In the long term, it will eventually have to happen. Market forces will inevitably push in that direction. But it may be ten or twenty years down the line.
Is this really the same policy problem?
No.
Because the underlying technical problems are sufficiently different.
The internet email architecture was designed to provide a universal, asynchronous, store-and-forward text messaging system for person-to-person communication across a heterogenous environment of network and host email systems. The adaptability of internet email to bulk or broadcast messaging is a useful artifact—a side-effect stemming from the automated copying capability of computers. The internet email system could have been designed for one-to-many messaging. If it had been, the architecture would probably more closely resemble Usenet.
Otoh, IP was designed to provide a universal, non-reliable, routed datagram service for host-to-host communication across a heterogenous environment of networks. IP was specifically designed as a layer underneath TCP and UDP.
The policy framework that regulates a system primarily designed for person-to-person messaging should not be expected to be the same as the policy framework regulating a host-to-host datagram service.