That's similar to what a recent study on the reliability of e-mail found. As I reported in this space last week, more than 1.5% of e-mails never arrived at their destination in an international experiment conducted by a group at the University of New South Wales in Australia.
This statistic has nothing to do with e-mail blocked by spam filters. It represents only the number of e-mails that didn't make it from Point A to Point B through the Internet's gaggle of servers, clients and routers.
Could the basic deliverability of the e-mail system we've come to rely upon be so bad as to lose more than 1 percent of all the messages we send? I decided to talk with some experts to find out.
George Bilbrey is a principal of Return Path, a New York-based firm that monitors the effect of errant spam filters on the deliverability of its customer's opt-in e-mail newsletters.
To establish a baseline for his client's delivery ratios, Bilbrey says his company sends out test e-mails every 15 minutes. "We look at 18 ISPs [Internet service providers] and 20 to 30 accounts per ISP," he explains.
Return Path's system then determines whether the messages were actually received by those e-mail accounts. "It's proprietary software we've written that retrieves mail from these accounts," Bilbrey says.
"I would say a 1.5% failure rate is not too far off what we find," he concludes. This tends to support the university's research.
Asked to explain the cause of the missing e-mails, Bilbrey can only speculate.
"It's never the same mailbox and never the same ISP" that loses e-mails consistently, he says. "The message isn't being filtered out, it's not being blocked, it's just never arriving.
"It's due to the general messiness of the Internet," he opines.
On The Trail Of The Missing Mail
Somehow, "general messiness" doesn't sound to me like a diagnosis that Internet doctors can do anything about. In fairness, it isn't Return Path's job to solve the overall flakiness of servers and routers. Minimizing the effects of faulty spam filters on legitimate, requested e-mail newsletters is Return Path's forte.
So I turned to Josh Baer, CEO of Skylist, a major e-newsletter service group. Skylist powers the sending of billions of e-mails every year for numerous clients, including such well-known companies as Sybase and the Ad Council.
Baer is a longtime participant in Internet standards bodies. He helped draft RFC 2369, a document that proposed a standardized way for recipients to unsubscribe from or receive help on e-mail lists.
"In general, e-mail is a reliable protocol," Baer says. "But we see inconsistent behavior. If we're testing something, it works 19 out of 20 times, and we just consider it a fluke."
Baer notes that the university's experiment involved only two major ISPs, two educational institutions and 12 free e-mail account providers.
"They primarily tested free e-mail services," Baer points out. "It would be my guess and my experience that corporate e-mail servers are almost 100% reliable." He feels many free services face heavy server-processing loads because they must support a large number of users without necessarily gleaning revenue from them all.
For example, Baer says, "Hotmail constantly has problems keeping up with the e-mails and spams that are being sent to them, because they're such a big target."
When an e-mail gets lost in that situation, Baer concludes, "it's not in the Internet between you and Hotmail, it's been accepted by Hotmail, and then it's getting lost."
A spokesperson for Hotmail did not respond to a request for comment by press time for this article.
Until we have better statistics on the scale of the problem, e-mails that just "vanish" will remain a mystery.
Tim Moors, the senior lecturer in charge of the university's research project, is soliciting volunteers for a much larger test than the preliminary experiment allowed. To participate in the larger test, see the university's volunteer page.
In the meantime, when it comes to signing up for free e-mail accounts, it's probably a good idea to remember that you get what you pay for.