In the E-Mail Relay, Not Every Handoff Is Smooth

[Woodmonkey]

FOR centuries, letter writers have wondered whether their missives safely reach the hands of their intended recipients. “I know not whether this will ever come to your hands or miscarry,” opened a typical one in 1625, from Roger White in the Netherlands, who was writing to friends at Plymouth colony.

Today, of course, we send e-mail messages that travel great distances in seconds, rather than weeks. Occasionally, however, we don’t hear back and wonder whether our message was ever received. Wouldn’t we be grateful if we could know with certainty?

I’m blithely inconsistent, however. When I’m the recipient of an e-mail message, I’m uncomfortable when a sender, seeking reassurance of safe delivery, presents me with a pop-up box requesting that I click to acknowledge its receipt. I routinely decline to do so. Why? I can’t say exactly. Maybe it’s like the unpleasant business of being presented with a certified letter from an unpaid creditor.

Before the advent of a federal postal system, letters passed through the hands of many volunteer carriers on the way to their destinations. William Merrill Decker explains in “Epistolary Practices: Letter Writing in America Before Telecommunications” (1998) that letter writers were willing to “consent to write five letters on the chance that one might reach the addressee.” When a letter was lost or delayed, it was said to be “miscarried.”

So, too, can e-mail be miscarried. A message usually hops several times as it traverses one mail router to the next. (The technical name is M.T.A., for mail transport agent.) Each mail router can see only as far as the next hop. Once it hands off responsibility, it has no way to track the message’s progress.

The basic Internet e-mail standard — S.M.T.P., or simple mail transport protocol — has always provided for the destination server to send back an error message if the original message cannot be delivered. If no error message comes back, however, can the originating server assume that the message arrived, safe and sound? Not necessarily. A misconfigured server anywhere in the path between sender and recipient can miscarry the message.

The problem that leads to a message being lost can also prevent the sender from receiving a report of failed delivery. In such instances, e-mail disappears into the ether.

Features that let the final server tell the originating server “message received, all’s well” were added in the 1990s and benefit everyone. But what happens after that is the recipient’s private business. Requests to allow a confirmation receipt are invasive, like a stranger jumping out of the computer screen and demanding a response.

Acknowledging the irritation with these requests, Dev Balasubramanian, a Microsoft product manager, says that there is a “negative stigma” attached to asking people to participate in a receipts system. As its default setting, Microsoft Outlook has “receipt requests” turned off.

Asked last week whether he availed himself of receipt-request functions, Mr. Balasubramanian said, “I don’t use them at all.”

Neither does Keith Moore, a networking consultant in Knoxville, Tenn., who was a co-author of the software specifications in 1996 that generate a confirmation message when e-mail safely reaches its destination server. Mr. Moore is frustrated because end users’ e-mail software still lacks the design capability to use the server-to-server messages about completed delivery.

Some entrepreneurs have seen that uncertainty and offered senders the ability to obtain receipts that a given message has been read — without the recipient’s knowing that a confirmation has been sent back to the sender. ReadNotify, based in Queensland, Australia, started in 2000 and promises to report not only on whether a message is read, but also on how long it is opened for reading on the recipient’s PC. It can also send the message in “self-destructing” form, preventing forwarding, printing, copying and saving. I admire ReadNotify’s ingenuity in presenting booby-trapped messages as being feature-rich.

Last week, Chris Drake, the head of ReadNotify, defended his company’s service. Some experts have questioned whether such technology is legal under American law, but Mr. Drake says “e-mail tracking is legal because e-mail is ‘owned’ by the author.”

A similar service, MsgTag, based in Wellington, New Zealand, does not want its features to seem overly intrusive. “We’re interested in peace of mind, not spying,” the site says. Its distinction? It does not report on how long the message was viewed.

There are many technical reasons that these services cannot reliably detect when a message has been read. But even when they work, I find their furtive nature offensive. When asked about services that surreptitiously monitor what recipients do with their e-mail messages, Mr. Moore of Knoxville expressed similar revulsion.

“I don’t want people to know I’m reading my e-mail,” he said. “In particular, I don’t want spammers to know I’m reading my e-mail.”

Ah, spam. Thank you, spammers, for making e-mail delivery more uncertain than ever. Not only are legitimate messages trapped in filters, but so, too, are error messages generated when e-mail delivery fails. Overwhelmed e-mail administrators may delete all nondelivery messages as a matter of policy, or the messages accumulate in the spam filter, unread, while we wonder why we haven’t heard back from our correspondents.

WHEN Christopher Columbus mailed the first letter from the New World on his return voyage in 1493 — it was addressed to Ferdinand and Isabella and enclosed in the only envelope handy, a barrel, which he tossed overboard — he did not expect to get a return receipt. (In “Epistolary Practices,” this is called “the first instance of a trans-Atlantic letter lost in transit.”)

Columbus was dependent upon the kindness of strangers, and wrote that he was “earnestly begging whomsoever might find” the message to convey it onward. Today, it is networking protocols that request the same favor. We don’t expect a return receipt, either.