The internet company I didn’t start

Felix Salmon writes this morining in exasperation:

Let Me Pay to Send Email!

What I’d love is some mechanism whereby I could pay “postage” of a few cents, maybe to charity, on selected emails I send. That would be a very strong indication my message was not spam, and should be let through. But I fear coming up with a universal standard for such a mechanism is practically impossible.

A few years ago, a good friend and I were toying with the idea of a tech start-up. (This was post-crash, but we’re both really unreconstructed new economy types.) I offered a proposal that I thought had a very short shelf life. Someone surely would do something like this, soon. Hah!

We would start an e-mail service that would require mail to be “stamped”. You could sign up for it like any e-mail service, but if anyone sent mail to your address and it didn’t have a stamp, it would bounce with an explanation of how to supply postage. No new standards or internet infrastructure would be required.

The receiver of the mail would set the postage rate and get the money. That is, you do not pay a postal service for delivering mail (that’s free in the internet age), you pay the recipient for the burden your correspondence places upon her attention. When mail is delivered, it would include a convenient link whereby you could refund the postage to the sender with a single click. Among frequent correspondents, the postage would be refunded as a matter of course. It would serve as a guarantee of nonabusiveness, but would rarely be paid. Therefore, people could set their postage rates fairly high without losing mail they care about. (One could set-up whitelists of senders whose postage would reimburse automatically, and blacklists of senders whose mail and postage are both refused. Importantly, postage need not vouch for the identity of a sender, but need only associate a letter with an account from which postage has been properly drawn.)

Mail from strangers would not ordinarily be reimbursed, and would serve as a potential source of revenue for people whose attention is in great demand. Although senders would have no guarantee that mail with postage would actually be read, recipients would be encouraged to set a rate at which they’d be willing to give a quick read to a one or two page letter. Famous people could give out real e-mail addresses freely, and use the price mechanism to control the quantity of mail they actually have to sort through. Ordinary blokes would have a means to reach the powerful and famous, at a price. (Of course, senders might often plead poverty and beg that the postage be reimbursed. But mailing would be at senders’ risk. If a recipient is unimpressed with a letter, by default they pocket the cash.)

Also, various kinds of businesses might sprout up using incoming mail as a payments mechanism. These might range from informal consulting gigs (experts on various topics who generally respond to paid questions, that’d be very nice for open-source software support for example) to the email equivalent of 1-900-SEX-LINEs. Whatever. Let a thousand flowers bloom.

A scheme like this could be implemented incrementally, without changing any internet standards. Mail service providers would offer POP, IMAP, or web e-mail access, but would require that incoming mail to have an encrypted header or attachment indicating an account from which postage could be drawn (up to some amount, for a limited period of time, for a specific letter). If the token is present and payment succeeds, the mail is delivered to the user. If not, it is bounced. Mail service providers would earn revenue by taking a fraction of unreimbursed postage payments. Because a monopoly mail service provider is implausible and undesirable, the business model here would be to build software and a payments infrastructure, à la MasterCard, something aspirant competitors would prefer to use than to reconstruct. The goal would be postage interoperability. On the software side, the hardest part would be making postage-payment convenient. One could start decently but imperfectly (e.g. how e-mail-to-fax services bill), but hope to get embedded in popular e-mail clients. Initially, one would market the service as a free, secondary, spam-free e-mail account, and try to get some prominent people to offer publish addresses and postage rates. No one need abandon their traditional, free e-mail accounts, but if managing spam is as costly to others as it is to me, there’d be plenty of incentive to encourage correspondents to migrate in order to reduce the need to monitor open spools.

Legitimate commercial correspondents and opt-in mailing lists would expect to have postage reimbursed, but could not enforce that via the mail system. They would have to establish a contractual relationship prior to mailing, stating that unreimbursed postage payments from certain addresses could be billed back to the recipient. That’s a feature, not a bug. Attention is too scarce a resource to commit as lightly as with a webform checkbox set by default.

Ideas like this have been bouncing around the internet for a long time. I don’t pretend this is particularly novel. I don’t know that it could work. It astonishes me, though, that, as far as I can tell, no one has ever made a serious attempt to do this kind of thing.

Update: In the comments, Richard Serlin points to a very nice, similar proposal by Marshall Van Alstyne. (I know there are more out there, somewhere in the archives of Slashdot.) In Alstyne’s proposal the recipient opts to take, rather than opting to reimburse, which might be a better “choice architecture“. That way you don’t piss your friends off just by having forgetten to click a link.

Update History:
  • 18-June-2008, 7:45 p.m. EDT: Added update re Serlin’s pointer to Van Alstyne’s proposal.

11 Responses to “The internet company I didn’t start”

  1. Felix writes:

    The problem is with the micropayments, which to many peoples’ astonishment have never taken off online. And LOTS of people tried. If a viable micropayments system existed, I’m sure this would have been tried by now.

  2. Felix — It’s true that no general purpose micropayments service has ever caught on. But lots of bill-in-$10-increments, charge-in-10¢-increments have done okay. As for payments, Amazon Associates makes a lot of micropayments. The key is always that you stay within the system until the amounts in question are worth the transaction costs of a conversion to more general currency. It’s hard to know how to evaluate arguments of the form “could be done, no it couldn’t” without trying, and even then if it don’t work the “could be” side can always blame operational failures. But I don’t think the history of specific services tied to micropayment-like exchanges is as uniformly pessimistic as you suggest. (Second life is full of convertible micropayments!)

    Also, one aspect of this proposal is that some of the payments are not so micro. That is, if you could get some people to offer desirable services funded by “postage”, you could piggy back on their clients, by giving all payers a postage-funded e-mail address. If everyone who phoned a 1-900 number automatically got their own, we might have them for more than porn lines.

    (It’s not a bad idea, btw, with phones to have an automatic waivable tariff payable by telephone callers to those they disturb. Perhaps I am overly solicitous of my peace, but I do think generally that those who compel my attention when I have not offered it of my own accord ought compensate me for that.)

    In the end, this is a kind of micropayments scheme, it ends up looking like a prepaid PayPal with e-mail attached.

    I’m not sure what the “right” postage level would be for nonfamous peoples’ ordinary correspondence. As you say, a few cents might be enough to block spam. But maybe not. My “snail mail” box is filled to the gills with marketing crap, costing between a dime and 50 cents per in production and postage to whoever was dumb enough to send it. I’d earn $15 to $25 a month if I could charge a quarter for all the credit card offers and advertising flyers I get, and I’d rather that money be spent on mere transfers (to me!) than on environmentally unfriendly physical mail. Spam is preventing the disappearance of snailmail, since with all the filters and scanners and overflowing mailboxes, e-mail is not at all a reliable medium. Even if mail is delivered, it might be lost in the crowd. Banks can no longer communicate at all by e-mail, given the phishing spam.

    Nonfamous of desirable demographics could probably charge a buck or two and earn an even better living from well targeted ads.

    I’m probably naive — after all, as the economists always say, if that were really a $20 bill on the ground someone would have picked it up by now! But I think this is worth a try. E-mail is now, frankly, a failed experiment.

  3. BTW, I should have noted that postage would be for threads, that is replies to the original sender would be accepted based on the original payment, and within a few day window after a response from the original recipient, follow up e-mails would be accepted.

    (There’s enough information in ordinary e-mail headers to make free followups work from any e-mail address, I think, but if there were desirable services — famous people, experts, etc. — a good way to get people to use a pay-to addresses would be a stronger implementation, that is by sending from a pay-to to a pay-to, it’s less likely that follow-up correspondence would be deemed new, requiring postage whose reimbursal would not be guaranteed.)

  4. Boston University economist Marshall Van Alstyne has an article on this exact thing in the March 2007 issue of the Economists’ Voice

    A quote:

    …No stranger can reach a communications

    inbox unless he will promise not to

    waste the recipient’s time. This promise binds

    because the inbox returns all messages from

    strangers that do not provide a sender guarantee,

    literally a tiny bond of say 2–5¢. Recipients can

    set the amount arbitrarily small or large depending

    on their opportunity costs.

    In response, Jamus Jerome Lim, points out a problem with a system of this kind:

    …there is a significant amount of spam that is

    currently being sent via “zombie” computers:

    Internet-connected computers that have

    been compromised by viruses or trojans that

    perform the devilish deed of sending out

    spam, often without the owner’s knowledge.

    Should the owners of these zombie bots

    then be made liable for their contribution to

    the worldwide spam problem?

  5. Richard — Nice catch!

    Re the zombie computer issue, that can be designed around. Postage should be applied per-e-mail, not committed to as a general bond based on originating computer or sender. Of course, e-mail clients would become convenient about applying postage, we wouldn’t want to replace wasting time on spam with wasting time buying postage. Convenience means automated, and automated means accessible to viruses and trojans. But designing postage as prepaid and purchased in limited fixed increments helps mitigate the possibility of disaster. That is, my infected computer might be able to take my $50 “roll of stamps” to send malicious mail. But it could not take more, without getting into my post office account (which is as hard as getting into my Amazon account). Infected computers might send postage to their own addresses, basically steal the money, but paying postage out needn’t be anonymous or untraceable.

    However, it would be important to emphasize that postage authenticates payment, not identity. That is, if you get postage-paid mail from Bank Of America, you can be certain somebody paid (perhaps with stolen money) to get it to you, but you can’t be certain it actually came from Bank Of America (and should not click its link and enter your password). Authenticating identity is a harder problem than this proposal is trying to solve.

  6. Chris writes:

    Or, alternatively, you could use gmail or any other e-mail service with a half decent spam blockers.

  7. br writes:

    Steve –

    Made the same, and similar, proposal several times myself. Alwyas got the “Aw, shaddap and use an e-mail service with a better spam blocker” response from the group, whichever group it was at the time. In the event I use two e-mail services (gmail and gis) both of which have “good” spam blockers. The later posts in the comments section here notwithstanding, it is not the same, for at least five reasons:

    1. Failure. Lotsa spam gets through. Too many ways to mis-spell relevant words ( “de*t consolidation” etc.); too many ways to “fake” acquiantance and purpose (“love and hugs and kisses … cindee”).

    2. Mechanicalness. One wants the mail system to be “free” (in the sense of not bound up) in transport, and it is annoying enough in this world to be refused smtp becuase you don’t have a domain ip address from where you happen to be, etc. Many of the anti-spam measures are notably over-mechanical. Mechanical-ness will always be a prblem is this sort of thing (same issue with attempting to obscenity-rate content by scanning for use of a word, etc. – medical info gets confused with porn and vice versa…)

    3. Nothing about blocking “pushes back.” Most of my proposals to date have similarly had to do with deposit payments and “mini-bond” payments and the like, refundable and cancellable in the case of genuine e-mail. For many reasons (probably too involved to go into here), one wants to “displace” the invader, not just have it bounce off the walls. One wants to push, if gently, as well as block.

    4. Wrong slices across the problem. The zombie problem actually is complicated and cuts several ways. Blocking, too, invites zombie strategeies, even as a way to corrupt or “stain” the zombie victim computer or system. If you can send out something that then gets all traffic to be blocked….

    5. Lack of informational depth / structure than can’t be readily deepened. Blocking, as a category of strategy, is inflexible. This is the mechanicalness issue again, but it is also the probalem not just of making lists, but controlling (blocking??) changes to them. If one looks at the whole system, one doesn’t see, for example, any thread through it that is verificatory, that measures that we’re doing it right, etc. In short, we aren’t developing good information about our informaiton handling.

    For whatever it might be worth, my thinking on the subject at the moment is to toss it in with a general problem of information managment that is looming — large I think. As a small example, one want’s one’s medical records to exist in a computer network, and be read and written readily, but one wants something more than the traditionally “mechanical” password control with respect, for example, to records duplication. (One copy using the password can lead to an infinite number of copies without it…none of which the recordee has any control over) I think systems that do better are possible, and I think they are more than a twenty on the pavement….

  8. br — I agree with most of what you have to say (not so surprisingly). The point about the corrosiveness of blocking — DoS attacks, the cost of mail important mail not getting, the loss of the sense that sending e-mail is a reliable means of communication — is important and subtle.

    I agree that spam is part of an ongoing, very large, information problem. Our capacity to produce information always outstrips our capacity to manage or even reliably store it (let alone secure it). But I don’t think we’ll find overarching solutions, just lots of domain-specific solutions, each one borrowing good ideas from earlier work in other domains…

  9. br writes:

    Steve –

    “Overarching’ solutions to the extent they are over-arching indeed can be a problem.

    But, for example, I spent a little time tryhing to get somebody interested in a system for general wireless net access, a kind of Mastercard-meets-wireless hookup. It was intended for use especially in Europe, which has not had as much fiber backbone capacity. In pmany places in Europe one went through the PITA of going to the desk at the hotel and paying a few euros for a “user-id” and “password” that would be good for an hour of “access” to get through a squeaky door in a badly made walled garden to the establishment’s sort of cranky connection. The connection would go down, and you’d lose your hour in nothing, the password and “user-id” were nonsense strings made up on the spot, and the fat old guy from Belgium would get 6 for G and I for 1 and so forth, be squabbling with the concierge while you were trying to get something done.

    The idea was to have accounts (whcih, in the end are essentially a long, modifiable, bit string) contained on the use computer (your or my laptop as we travel along) which were the analogy to the “card” part of Mastercard interact with a network that has a cetral database structure, does the acounting and charging. Also not so unlike the toll transpoders (FastLane here in the Northeast, etc.) Accounts in the instance, of course, to be established over he internet using the hardware that the user has in his hands, no “transponders,” but the idea is that the user walks into a hotel (that maybe has a little decal on the door — like the Mastercard accepted here) opens the laptop, and his client offeres him service at a stated price (which could easily be a bits price rather than a minutes price), clicks on the icon that says connect and there you go. Gets a bill at the end of the month, can pay by credit card, or can pre-pay, etc.

    Beats the hell out of Boingo. Does well allocating the resouce of backbone capacity. Gives the user control and access without fussing with the Concierge, etc. And, I would argue, beats the hell out of Fon.

    Talekd to a few people. People seemed not to get it mostly becuase they conceived the network as a something that you “accesed” by getting a something else. Sorta you get a ticket and then you go to a concert. There are tickets and then there are concerts. Didn’t understand that struggling with the desk clerk at the hotel for the userid and password wasn’t really the lined up way to go.

  10. apelsopsy writes:




  11. Fatbraliari writes: