Context matters even at Goldman Sachs

A couple of weeks ago, I wrote that when it comes to email, context matters.  My point was that context in which content is found influences whether or not it is considered spam or not.  At around the same time, the Wall Street Journal came out with an interesting story indicating that using swear words in email at Goldman Sachs has now been banned:

There will never be another s— deal at Goldman Sachs Group Inc.

The New York company is telling employees that they will no longer be able to get away with profanity in electronic messages. That means all 34,000 traders, investment bankers and other Goldman employees must restrain themselves from using a vast vocabulary of oft-used dirty words on Wall Street, including the six-letter expletive that came back to haunt the company at a Senate hearing in April.

"[B]oy, that timberwo[l]f was one s— deal," Thomas Montag, who helped run Goldman's securities business, wrote in a June 2007 email that was repeatedly referred to at the hearing.

Mr. Montag, who couldn't be reached for comment, wouldn't be allowed to send that email under Goldman's sanitized communications policy, which is being enforced by screening software. Even swear words spelled with asterisks are out.

A Goldman spokeswoman said: "Of course we have policies about the use of appropriate language and we are always looking for ways to ensure that they are enforced."  The new edict—delivered verbally, of course—has left some employees wondering if the rule also applies to shorthand for expletives such as "WTF" or legitimate terms that sound similar to curses.  In the spirit of the times, there is no written directive specifying which curses now are officially cursed. But screening tools being used by the firm would detect common swear words and acronyms.

New York-based media company Bloomberg LP says it has monitored emails for more than 10 years, using an application that scans messages for 70 words and phrases—in English and several other languages— considered profane.

But using software to screen for obscenities doesn't always work. Satellite-services provider Intelsat SA, of Luxembourg, a few years ago began screening email for profanities. But it found the level of emails that were wrongly captured was too high, and so it dropped the screening. It now relies on a policy of reminding employees of "appropriate language for external use," a spokeswoman said.

The last couple of paragraphs are the ones I want to highlight.  Key words used to filter on sensitive words are something that always sounds good at the time but when you actually implement it in real life, it doesn’t work as well as you thought it would.  If you read through the article, some commentators argue that the lack of professionalism on Wall Street is rampant, at least with regards to language.  When people think about banning swear words in email when it comes to spam, they are usually thinking about it in terms of sexual innuendo.  This is not always true but is frequently true. 

Unfortunately, swear words are also used as slang terms.  And slang terms can be used in legitimate contexts even if it does appear unprofessional.  I have never worked at a job where I didn’t hear that type of language from time to time, sometimes regularly when I did landscaping.  Indeed, I have been known to utter a few of these words myself when I can’t get my wonderful desktop phone to work properly, or can’t figure out how to use the printer/scanner combination.  Of course, I would never type up such things in email, but in the wild west of Wall Street, such language isn’t that uncommon.

This illustrates the point banning mail on certain words will be rife with false positives.  People use slang to convey thoughts, feelings, emotion and context.  They are used as adjectives, verbs and adverbs.  The fact is that slang has multiple uses and that is why it has made its way into our language.  Using technology to filter out these phrases is going to be difficult because technology has to understand language.  Without that it will never work as a filtering methodology unless people are willing to tolerate a high level of false positives.

Goldman Sachs, and others, attempt to get around this by appealing to the nature of people.  I have doubts upon whether or not this will work since it would require a culture change within the investment firm.  Corporations don’t change cultures quickly, they change them slowly (if at all).  It remains to be seen whether or not this requirement will stick.

Comments (1)
  1. Nathan says:

    Back when I was in high school. I was typing an e-mail in Hotmail and I typed the words "finish it". The lovely censorware they had running turned it into "fini*****". And I'm sure most people who've been on the Internet for any length of time have used some forum software that doesn't let you type Charlie Peacock or Reese Witherspoon for reasons completely unrelated to the quality of their performances.

    However, I am curious…since you don't think this is a problem which can be solved with technology and you're skeptical of Goldman Sachs' ability to solve it through culture change, how do *you* think they should approach the problem?

Comments are closed.

Skip to main content