TODAY, MANY COMPANIES are in the business of monitoring social networks like Facebook and Twitter. Marketers and brand managers gather data and run “sentiment analysis” to see how people are, say, feeling about Coca Cola today. At NewsWhip, we monitor Twitter and Facebook to spot the fastest spreading news stories.

Now the US Federal Government has launched its own social media monitoring strategy. And on Monday we saw some of the results of this strategy, as a pair of tourists were arrested in the US on terror charges over jokes they made on Twitter.

A quick recap: the Daily Mail reports that two tourists, Leigh Van Bryan (from Ireland), and Emily Banting (a British woman) were detained by armed guards last week in LA Airport just after clearing immigration. The duo were separated and interrogated for five hours, and Leigh was transported, in a cage (where he had a panic attack), to a cell, where he held overnight in a cell in the company of Mexican drug dealers. The duo were refused entry to the US and forced to buy new flights back to Europe.

Their offense? In the weeks before his trip, Leigh had joked about his upcoming trip to LA on Twitter, saying he would be “diggin’ Marilyn Monroe up!” (a reference to a Family Guy joke). He also asked a friend on Twitter if they were “free for a quick gossip/prep before I go to destroy America? x”

After his arrest Leigh and Emily tried to explain how both of the tweets were jokes. Going to “destroy” a place means you get drunk and careless there in British slang. Digging up Marilyn Monroe was a reference to a gag from Family Guy, an American TV show. The explanations were ignored as the duo were arrested, processed, and packed off to Europe.

So how did this happen? Here’s where it really gets interesting. And, well, scary.

Twitter as Big Brother?

First, how did the Department of Homeland Security (the DHS, the guys responsible for the border) find Leigh and Emily? It’s unlikely that either of these guys were on “watch lists” in advance of Leigh’s tweets. I mean, look at them. And the “documents” provided to Leigh after his arrest by the DHS officers mentions only his posts on “tweeter” as the cause for refusal of admission. Though it’s possible they had other reasons, none were mentioned by the DHS officers.

So assuming Leigh was not pre-identified before his fatal tweets, the DHS must have picked him out of the 250 million tweets published each day on Twitter, known in the industry as the “firehose”.

How is this accomplished? Well, automated processes – like search terms – will be at work, spotting phrases like “kill” “shoot” “destroy”, etc., when those words appear in tweets with words like America or the US.

This is the same process that brands monitor for things like “I hate Coca Cola” or “I love Coca Cola”. In Boolean search terms, the query will look like: (“hate” or “despise” or “disappoint!” or “love” or “like” or “yummy”) w/3 (“Coke” or “Cola” or “Coca”). This will spot various emotive words appearing close to Coca Cola keywords and give a fuzzy picture of how people feel about Coke today.

The DHS probably has something similar, with a huge range of word searches that someone decided should trigger security alerts, like attack, kill, destroy, etc. (You can probably think of more, but I don’t want to trigger alarm bells by stuffing them all into this post.) When those searches are triggered, then the process begins. Last week, that process culminated quite smoothly in Leigh and Emily’s arrest.

So it seems likely that all of Twitter (and probably Facebook) is being monitored for potential threats, and goodness knows what else. So if you’ve tweeted the word “ridiculous” or “Osama” or “destroy” within a few words of “America”, you might be on a watch list too.

Think you’re safe? If you’re a photographer, what if you’ve tweet that you’re planning to “shoot the Empire State Building”? Or if you’re hoping your product really takes off at a conference, what if you hope it “explodes”? What happens if you tweet that you hope the “US gets wiped out” in the Hockey finals? Or if Mastercard want to “attack” American’s Express’s dominance of the corporate cards market in New York?

Well tough luck. With any of those phrases you may have triggered the “deportation algorithm” – the unfortunate combination of words that makes you look vaguely like a threat to a computer in Arlington, Virgina, and results in your eventual forced removal from the USA.

Fascinating, no? The number of false positives a system like that could throw up is huge. Imagine how many potential terrorist suspects the DHS will discover if “Romney destroys Obama” in a debate? Will the resulting chatter be interpreted as millions of death threats?

Also, how do you feel when you are at a border about having to answer questions about everything you’ve ever said publicly on Twitter or Facebook, when the questioners refuse to understand your answers and will deliberately fixate on the worst possible interpretation of what you said?

What’s the upshot? If you are an “alien” (the US government term for non-US nationals) and if you ever intend on visiting America, it looks like you need to be very careful what you tweet. Or share on Facebook. Or even write online. A joke, political comment, expression of disagreement with an act by a US politician, or anything else might one day get you caged, put in a cell, and effectively “deported” at your own expense, likely never to be allowed to return.

And “aliens” don’t have constitutional rights against the DHS, so you can forget about those “freedom of expression” arguments bubbling up in your mind. So what will you do? Self-censor in case you ever want to go shopping in New York, attend a conference in Vegas, or ski in Colorado? Parse every word you utter online carefully with a view to having to justify it one day to a stern prosecutor who won’t listen to any contextual explanation?

What happened to the Humans?

The DHS has a very sensitive detection social media detection algorithm. In a sense, that’s fine. The fact that the DHS has a system that triggers some sort of reaction when someone tweets the words “destroy America” is perfectly sensible. (Even if it is a bit much to assume anyone will tweet about that if they intend doing it.)

But how can that phrase trigger a bureaucratic process that results in people being caged and deported, without any possibility of common sense intervention once that algorithm is triggered? It seems that no decision-making human entered the process, other than to verify that Leigh had in fact written the tweet, and as a matter of pure syntax, it could be interpreted as a statement of intent. I once read a science fiction story set in a world like that, where computers decided who was guilty and innocent based on algorithmic probabilities and (mis)interpretations of statements. It was scary, but laughable. Humans would never introduce such a system, right? Right?

As a matter of national security, this doesn’t look good. Making people from other countries afraid of talking about (or even mentioning the word) America will not serve long term security objectives, or the US national interest. Hearts and minds won’t be won when you can’t even discuss your feelings about a place.

Plus, I wonder if policymakers have considered what happens if other countries start following the US lead. What will the US reaction when its citizens are getting detained and deported all over the world for bad jokes, expressions, and political opinions they once expressed on Facebook?

The weird thing is that the story of Leigh and Emily is probably considered a success story for the social media monitors in the DHS. Threat detected. Threat neutralized. High fives.

Anyway, I’ve been careful as I typed this, hopefully careful enough to not get a detour to secondary screening next time I visit my friends in the states. In US First Amendment-speak, I’ve been feeling a “chiling effect.” I hope common sense prevails and the incident with Leigh and Emily is not a picture of our future. Now if you’ll excuse me, I’m off to destroy a bag of Taytos.

Paul Quigley is co-founder of Newswhip.com.