Five years ago, I became an anonymous guru, answering the world’s questions sent in via text.
My specialty was looking up Google Maps directions. Sitting in my pajamas with a laptop in my lap, I would quickly type in two locations and respond right away with a time estimate. In each message, I tacked on a friendly “ChaCha On!”
On the other end, an equally anonymous person with a cell phone would get my text message and know just how far away the movie theatre was from their restaurant. (Hopefully they wouldn’t immediately respond asking if there was a later showing of Avatar.)
In 2016, this sounds absurd. Today, Google Now will have already told them they’re running too late. Tomorrow, Facebook’s M would have re-booked the tickets for a later showing.
But in 2008, months after the iPhone launched, the easiest way to get information to your phone was to ask ChaCha.
The world’s information at your (texting) finger tips
When ChaCha first launched in 2006, it was a different way to search the internet. Instead of typing into Google, you could just have a normal chat conversation in a box that looked a lot like AIM.
The startup raised $10 million because it was “a game-changing disruptive equation, even in the context of giants like Google, Yahoo, and Microsoft,” its investor, Morton Meyerson, was quoted. Investors like Jeff Bezos latched onto it, pouring more money into the Indiana-based company.
By 2008, the service launched a mobile version, which is when it went viral with my friends. As a high school student, I played around with it after school, trying to come up with the craziest questions I could ask. Unless you had a smartphone with a fast hook-up to Wikipedia and parents who didn’t care about data, texting ChaCha was your info line and your entertainment.
However, ChaCha’s popularity amongst teens like me was a big problem for the site.
“The whole goal was to break the system. You have guides who come in and try to do foul play, and then you have users who come back in over and over again,” said ChaCha’s CEO Scott A. Jones in an interview with Business Insider.
Half of the company was devoted to managing unruly guides — contractors like me who would answer questions — and the unruly users — teens like me who just wanted to ask it crazy questions to see if ChaCha could answer.
“It was only 1-2% of the users or the guides, but they could do damage quickly because they were intent on it. They would spend hours a day just pummelling the system or answering the system with bad answers,” Jones said.
Using humans to fall back on
One way around the abuse is to rely only on algorithms and computers.
When you search Google, there isn’t a human behind it frantically searching. When you ask Siri a question on your iPhone, she speaks back, but in a jilted computer voice. It’s more amazing when she can find things for you, and less frustrating when she can’t. She’s clearly not human, after all.
Early on, ChaCha started cataloging questions and the answers written by guides. Nine months after launch, ChaCha added “Expediters” who could select an answer from the catalogue to send in response. Basic questions like “How many planets are there?” or “Who is on the dime?” didn’t require someone to search it every time.
As a user, I didn’t notice whether it was a human or the AI answering me. Since the answers had been written by humans before, each one generally read like someone had personally written it out, down to the same “ChaCha Again!” or “ChaCha On!” sign-offs.
Jones estimates that about 70 per cent of the questions were answered first by AI, another 20 per cent had an answer selected by the database, and the remaining went to guides who would answer by hand.
“We were 10 million questions and answers,” Jones said. “But it didn’t really matter because we had this army of Guides and they could always answer a question that way.”
Becoming a guide
By the time I was a college student with time and an internet connection to waste, I decided to try my hand at becoming a “Guide” in December 2009.
At that point, the questions that ChaCha did toss to guides like me were either geographical or time-based. Answering “What was the score of the game last night?” required knowing the day and what the big game was.
But about a month into my tenure as a guide, I received an email. My user rating was too low at 84 per cent. ChaCha told me I should make sure I’m still signing my messages with “ChaCha!” or my most-used “ChaCha on!” Otherwise it was a list of helpful hints and tips.
This is what it must feel like to be an Uber driver today, and maybe an M employee tomorrow.
As a contractor, I don’t remember the day I stopped. I think I just stopped logging in once school got back in swing, and my dead time evaporated with a new semester.
My month as a human Siri faded to the back of my mind (to this day, I don’t think I ever earned the $100 minimum). ChaCha, too, began to fade just as quickly.
Lesson for Facebook: People are messier than computers
ChaCha now is more of a “low boil” than the roaring viral sensation it once was, Jones says.
The company was never able to monetise its service. The site followed the trends in mobile, moving away from costly texts to building out an app. The 12 employees who are still with ChaCha work remotely, while Jones has decamped to Hawaii, away from his Indiana headquarters. It makes its money as texting service for Veterans Affairs and licensing out its data.
But now Facebook is following in ChaCha’s footsteps with its own hybrid human-computer assistant: Facebook M.
M is supposed to be more transactional — it’s more about helping somebody book restaurant reservations or buy flowers — but it’s using a lot of the same tricks as ChaCha.
For instance, ChaCha gave guides a full chat history so they could see when a pronoun like “it” referred to a movie theatre’s location versus a movie itself. Many of the apps, like Siri, can’t even follow a chain of conversation.
For Facebook’s vision of M, humans monitor “every” communication from “start to finish” to help the computer. Hopefully, that will cut down on bad answers on behalf of the computers, but it could open it up to bad actors on behalf of the humans.
Facebook claims it has a few dozen employees now, but its chief Messenger boss envisions a network of “thousands” of M Trainers over time. Whether they will behave better (or perform at a better satisfaction than 84%) is hard to know.
But Jones cautioned that people react differently to computers and humans, as we’ve already seen in the early M reviews.
Those reviews tried to push the limits of what the “M” service was supposed to be able to. Computers normally can’t hand-draw photos, so the little human puppeteer of M in the background peeked its head out to do it for you. With ChaCha, part of the magic was knowing that a human was on the other side, looking up things for you — or answering the most insane question you can think of.
Either idea, once it reaches millions of users (or Facebook’s billions) starts to face problems. More humans trying to challenge the computer, more humans introducing error into the computer’s work.
Given Jones’ experience with the unruly guides, he’s sceptical about the privacy, especially since a lot of what M is doing is facilitating transactions, like booking plane tickets or lowering a Comcast bill. It’s an area he intentionally avoided.
“We didn’t do things like make a restaurant reservation on your behalf,” Jones said.
Facebook may have the advantage of having linked accounts (and an extra decade of technological advancement), but ChaCha’s troll trouble is a warning shot.
“I’ll be interested to see how they pull it off because it’s rife with places where people’s privacy could become a risk,” Jones said.