Emergic: July 28, 2003 Archives

Countering Spam

Technology Review has an article by Vipul Ved Prakash, founder of Cloudmark:

Here's a list of three rules (created after the most important features of e-mail) that anti-spam software should strive to follow:

1) Ability to send and receive e-mail from a stranger. (Whitelisting, payment systems, and challenge/response break this rule.)
2) Ability to send and receive pseudo-anonymous e-mail. (Domain-based authentication breaks this rule.)
3) E-mail should be free. (Payment systems break this rule.)

If we can solve the spam problem while maintaining the three features of e-mail, it would be a much sweeter victory.

Related Entries: [All]
Spam Trouble [December 12, 2006]
Spam Crisis - Over!? [September 22, 2004]
Anti-Spam Companies [April 5, 2004]
Mailblocks for Battling Spam [February 20, 2004]
E-Mail Stamps to counter Spam? [February 4, 2004]

Software | PermaLink | Comments (12)

Vipul's rules on anti-spam software sound more like complaints about competitors' products rather than anything based on logic and reasoning.

E-mail is not free. It has never been free. Most users may not directly pay for the e-mail service, but that does not make it free.

Ability to receive e-mails from stranger cannot be a rule. If US Postal Service increases delivery of junk mail to every one's house with growth rate based on a geometric progression (as it has happened to e-mails), every American will write to his/her Congressman to change USPS rules. You may substitute Indian scenario for the same.

Ditto for the "2nd rule."

Posted by Samir Desai

If all the currently proposed solution break the 'rules', then would he care to give the real solution, please!!

I have maintained earlier also that the solution to spam will be an economics-based one and not a technology-based one. Hence we have to wait for a non-geek (most probably someone from economics, psychology, industrial engg background) to come up with a neat solution.

In the meantime, lets all rant about currently proposed solutions!!! LOL!

Posted by AJ

The solution to spam should be based on economics as AJ suggests above but the implementation will be technology-based. An economist will naturally first identify the problem in economic terms. Essentially, spam imposes a cost on the receiver of spam and the sender of spam does not bear that cost. In effect, the problem is one of externality and therefore the solution has to internalize the externality somehow. This can be achieved by forcing the sender to pay for the cost imposed on the receiver.

I guess a little bit of thinking will reveal the technology required to implement the cost internalization. But first, we have to design the mechanism or protocol. I can off the top of my head think of one. Here is the method.

Sender S sends an email to reciever R and a third party T debits amount m from S's account and credits R's account. R reads the email and determines it is spam and keeps amount m; or R notes that it is not spam and refunds amount m to S.

R's mailbox could have two sections: the free section that only certain emails get through -- those whose addresses are recognized. The other section is the not-free section. All emails that originate from addresses not known have to deposit m and only then is the email stored; else it is rejected.

This mechanism allows people to send you email without fearing rejection provided they have paid amount m. And you have the option to refund the deposit if the email was legitimate.

Well, that is my two cents worth written from Incheon (Seoul International Airport.)

Atanu

Posted by Atanu Dey

Atanu, I liked your idea indeed. However thinking through it, there is one issue I foresee. This all will work if the amount 'm' is a deterrent enough ( taking into account the total amount as a result of sending out 'n' number of mails). Sice it has to be linked to e-mails which are so common, it has to be low value (2 cents?! kidding !!). So it will be few cents than in the range of $50 or $100. This system might prove inefffective if the sender's of spam mails send it with a knowledge that they would not get the refund and if they see the lost money as the 'cost' of sending out pointed advertisements which would still be low as compared to other traditional means for the target segment.

A

Posted by Ashutosh Agarwal

Furthering Atanu's thoughts on cost-based solution, there's this new article from NYTimes outlining costing of spam.

Posted by AJ

Following up on Ashutosh's concern about the value of 'm' in my comment above. I think that m has to be considered a 'barrier' and that is set by the user. Suppose you consider your time to be very valuable and each junk mail you read costs you $2. So you set the barrier at that: anyone wishing to send you mail, has to deposit $2 before sending you mail. So mass marketers who are willing to spend only 2 cents per addressee will not be able to reach your mail box. However, if I am willing to read junk mail for only 2 cents, I could get a lot of email in my mailbox and make a pretty penny as well.

In the last 8 days, I got 200 pieces of junk mail. If my willingness to accept junkmail was set at 10 cents and if all the mail could get over that barrier, I would be looking at $20 in my account. Not bad for the 5 minutes it took me to delete them.

Atanu

Posted by Atanu Dey

Simple math and economics can be the most effective tool against spam. Micropayments are not a new concept , but one that ultimately woiuld prove most effective. Legislation is to slow moving, Black lists can be circumvented or their host machines attacked, Proxy approval mail is all effort and no reward, and managing mail filters is a constant battles for missed junk mail and false positives. Forcing ALL messages through a payment system, much like the postal system levels the playing field and could result in people signing up for commercial mail with vendors they want to get mail from.

I suggest that all e-mail should include a sliding scale of micropayments payable to the recipient. Example rates follow:

Simple text message $0.01 per recipient
Message with graphics or file attachments $0.03 to each recipient
Message with recipient listed in BCC $0.05 each
Anonymous messages $0.10 each
Rejected mail $0.005

Micropayment vendor (Paypal, Citibank, US Postal Svc?) would offer an SMTP server plug-in indicate whether an e-mail could get through. All non-funded e-mails would be returned to sender as non-deliverable and would be indistinguishable from mail returned due to an invalid address.

Simple change to address format (user$company.com) replacing @ with $, would enable mail servers to route all paid messages during the transition.

As an incentive, a fraction of the micropayments could go to the mail hosts and micropayment vendors. Hotmail alone probably processes a few billion messages per month. If they got $0.001 per message, they could be paid in excess of $100M per year in pure profit to support the new model.

From the burden of cost perspective, the mail user would likely fall into one of 4 categories;
1) Occasional user with 1-25 messages sent per month (monthly cost less than $0.75)
2) Daily user with 25-100 messages sent to multiple receipients per month (montly cost less than $6.00)
3) Business users with 100-1000 messages sent per month (monthly cost less than $60)
4) Level One Spammers send 1000-1,000,000 messages per month (delivery cost = $60-$30k+)
5) Bulk mailers send 10's of millions per month and would likely need to pay in excess of $50k monthly with all the invalid address harvested and used.

On the flip side (revenue), occasional users would benefit most receiving 20-30 spams per day could generate as much as $100 per month. Daily users would net a bit less, and Business users would probably break even. Spammers on the other hand would have to ensure that their marketing message was compelling enough to justify the cost of the mailing. How much Viagra and Mortgage refinances do they need to sell to cover the cost of their mailings? Simple economics will improve the quality and focus of spam as well as elimiate the vast majority of junk spam with no point whatsoever. If porn mailers focus on known porn subscribers, they could make a profit, but sending offensive mail to millions of people who will never buy their product would be financially disincented.

Each account could hold a balance from which outbound message costs would be deducted and if the balance was transferable, end users could share the wealth with family and friends.

Obviously the numbers used in the calculations above are just samples and could be adjusted accordingly to find the ideal settings. Additionally, there may be some flaws with the basic assumptions above, but I feel it is a course well worth the research if we ever plan to reduce the amount of junk mail and assist legitimate businesses in more effective and targeted marketing campaigns. In fact if they provide an auto-respond button on the body of the e-mail, only the fractional micropayment to the ISP/Bank would apply.

--

Bill Fencken
623-561-6106

Posted by Bill Fencken