How stupid can an e-mail program be?

Eudora's MoodWatch feature presumes to judge the offensiveness of my language.

By Peter Y. Sussman

Published October 24, 2000 7:30PM (EDT)

I am hot, hot, hot. I am indebted to Qualcomm Inc., the purveyor of Eudora Pro, for alerting me to the appalling depravity of my racy life.

I recently downloaded version 5.0 of the popular e-mail program -- paying, as usual, for added features that I never knew I needed. (Such is the guilt of the techno-challenged.)

Among the expensive new features in version 5.0 is something called MoodWatch -- available, as luck would have it, only in "Sponsored and Paid Modes." The latter is a category designating fools like me. As the online manual describes this feature, "MoodWatch [whatever happened to the concept of spaces between words?] is a new Eudora feature that monitors incoming and outgoing messages for offensive text. A new settings dialog has been added for MoodWatch so you can determine how you wish to use MoodWatch," which, one is informed, "works similarly to a spell-checker," scanning words and phrases in e-mail messages and headers and flagging those that "may be offensive."

The feature bears some graphic resemblance to a slot machine, with iconic drawings of vegetables instead of fruit. A new column appears in all e-mail logs alerting the user to "the level of offensiveness" of messages by the appearance of "one, two, or three red chili peppers." One chili, says the manual, indicates that there "may be a slight bit of offensive language in the message." I'm not certain what constitutes a slight bit of offensive language. Perhaps it's the salutation "Hi, honey" in a letter to a female friend. In any case, it's a tamer category than two chilies, which is still conditional but somewhat more emphatic: "There may be offensive language [not just a slight bit of it] in the message."

Three chilies and you've stepped way over the line, buddy. It means "there is offensive language in the message," no "may be" about it. This kind of outgoing message triggers a "three chili warning dialog box," says the manual.

The three-chili dialogue box is a pseudo-hip yet quaintly old-fashioned reminder that I have succumbed to the temper of these licentious times. It reads: "Your message to [recipient] regarding [subject line] is the sort of thing that might get your keyboard washed out with soap, if you get my drift. You might consider toning it down." I do relish the anthropomorphism of software that refers to itself in the first person, but the message sets off a higher level of paranoia: There's someone hovering above my e-mail messages hinting with his or her "drift" that I'm getting out of line again. To send that message despite its patently offensive text, I'd have to click on "send anyway," which, I confess, I am tempted to do with a wicked chuckle. Mind you, these are messages we ourselves were dumb enough to compose and send before Qualcomm Inc., with its innate good taste, intervened in the interests of decency.

As a writer, I have a silly and sentimental attachment to my own words, offensive or otherwise. Fortunately, there's a software fix for those who remain attached to their own offensiveness. I immediately clicked on the box in "Options" that disables MoodWatch, and went about my business. But Qualcomm did not so easily relinquish its vigil over the language of my correspondents. No dialogue box popped up when I sent messages, but incoming messages continued to arrive labeled with chilies. I tried eliminating the MoodWatch columns individually in each of my mailboxes by dragging the column lines together, leaving no room for the chilies, but the next time I opened a mailbox, the column would be resized to its original width to accommodate the chilies, which had also returned.

What were my naughty friends saying that so offended the refined taste of my corporate guardian angel? I checked a few of the many messages in my mailboxes that were conspicuously labeled with those bright red chilies (even after I had turned off MoodWatch).

Three chilies: I think I get the drift of this one. An elderly friend sent me a message with background information for a book I'm writing that contained a reference to the so-called Filthy Speech Movement at the University of California at Berkeley in the mid-'60s. His message included the words "fuck, fuck, fuck." Perhaps Qualcomm Inc. gave it one chili for each use of the filthy word.

I subscribe to one of those "word a day" e-mail services. A message from this service also merited the coveted three-chili rating. The word that day was "coxcomb (KOKS-kom), noun," and the service gave its derivation as from the Middle English "cokkes comb, crest of a cock: cokkes, genitive of cok." The second definition listed was "A jester's cap; a cockscomb." Qualcomm Inc. saw right through that transparent attempt by lexicographers to disguise a penis by adding the "comb" suffix.

Two chilies: A friend sent me a column from the New York Times about capital punishment. Actually, here I side with Qualcomm Inc. I wish others were as aware as this corporation is of the offensiveness of the death penalty. But I do wonder what specifically it was about the columnist's discussion of that barbaric practice that set Qualcomm's sensitive antennae aquiver.

One chili: Here my analytical powers are not up to the challenge of recognizing offensiveness when I see it on my own, without the aid of chili icons. One-chili messages included one from a friend alerting me to an upcoming documentary on healthcare, another from a neighbor discussing an error in our property taxes and a mass mailing from CNet's I haven't a clue why these messages were considered potentially offensive enough to merit a chili, except possibly for the last one, which did tout an article on "giant hard drives, and whiz-bang gizmos." I will read that article with new enthusiasm to learn some of the latest salacious news about these monstrous hard drives and gizmos. For shame, CNet!

I decided to do a few tests. I turned on (so to speak) the MoodWatch option and sent myself messages in which I called myself a "shit head" and a "shithead." Sure enough, they rated three chilies -- though sometimes, inexplicably, the "shithead" spelling sneaked through undetected. Then I turned off MoodWatch and resent the same messages. They arrived with no chilies whatsoever! How come my friends can send me messages about property taxes and TV shows that merit a chili even after I've disabled MoodWatch, but I can't set off this sensitive barometer by calling myself a shithead? What does this say about the twisted mind that controls my software and my every exchange of deceptively innocent messages?

I tried another experiment. What if I quoted the words of our family-values Republican nominee for president? I sent myself a message with the following text: "There's Adam Clymer, major league asshole from the New York Times." With MoodWatch enabled, the words of the Republican nominee were three-chili offensive. With Moodwatch turned off, the ever-quotable George W. Bush was only a one-chili offender.

I also tried testing various racial epithets. Some slurs drew three-chili ratings. Others sailed through without the company's detecting that "there may be a slight bit of offensive language." Tell that to a person on the receiving end of the slur -- especially if the word is delivered as a slur.

And isn't that the point? There is no way to program into a machine the complex sensibilities or robust creativity of the human heart and mind -- with this or any other software program. So much depends on context and individual personality and tone of voice. It's not a matter of any particular word that should be added to or subtracted from the program's database. The very act of attaching chili icons to e-mail messages has as much cultural utility as pinning paper tails on paper donkeys. Moreover, it's dangerous. If library patrons are prevented by software from reading anything containing the word "breast," they will learn nothing about breast cancer, and if adults in their own homes and offices are cautioned by their software every time they write or read a word like "fuck," they too will be diminished by the tools they rely on to learn and communicate.

Words become offensive by the nature of the attention that is paid to them. When a corporation tacks a chili onto this or that word in an e-mail message or builds a software barrier around a word on a Web site, it invites writers and readers to consider the word one-dimensionally, with only the meaning and intent that the corporation has interpreted as offensive. It's Qualcomm Inc., not I, that is pointing leeringly at those words in my e-mail. Qualcomm's single-minded focus invests those words with the prurience that it then claims to find objectionable. Now, that is offensive.

Yearning to get rid of those damned chilies, I e-mailed Eudora tech support, presuming on my elite "Paid Mode" status. But the money I paid apparently didn't qualify me for a reply -- certainly not a rapid reply. I've heard nothing in the two weeks since I sent my plaintive appeal to tech support.

I do wish I could find some way to escape the heavy breathing of my prurient e-mail program. If you ask me, Qualcomm Inc. has a software bug up its [three chilies].

Peter Y. Sussman

Peter Y. Sussman, a San Francisco Bay Area writer and editor, is the co-author of "Committing Journalism" (W.W. Norton) and is currently working on a book of the letters of Jessica Mitford.

MORE FROM Peter Y. Sussman

Related Topics ------------------------------------------

First Amendment