How Do You Spell Mememez

Would You Take Research Advice From This Page

Last weekend, I came across a tweet. One of those things that pinballed from its original circle of followers to the wider world. One person’s mix of amusement, frustration, surprise and outrage over what happened during their pre-employment check became fodder for the greater Twitterverse. It had become, what we call in my family, a mem-mem-ez (three syllables, hard zee at the finish).

My daughter and I have this running battle over memes. Not over who can send the other the cutest animal video (I’m especially susceptible to racoon posts), but who knew about memes first. The concept, I explain as a student of the social sciences many years ago, with a Masters to show, has long been discussed. In fact, as I argue, memes existed back when the Internet was two defense contractors sending dot matrix pictures of their butts to each other (no racoons). She insists millennials like her came up with memes as a way of passing along pictures of kids in car seats. What we agree on is, Ok Boomer, Mom, hers, my Wife, has no idea.

What’s a mem-mem-ez she once asked as a cute puppy gif was flying around the family text line where someone called it a meme. Now it’s a family in joke. We don’t say memes, we say mememez. And the open source research community got a hell of a mememez, say it how you want, this weekend. A person Kate Le Franc, with the Twitter handle, @Bruise Almighty, posted about the results of their background check. The retweet that came up in my feed referenced them coming home to 351 pages of background results. My first thought, knowing background reports quite well, was it was a TLO or similar report, with 10 pages of findings and 341 pages of neighbors; something that could happen if you lived in a giant high rise. I then started following the full thread of tweets. It turns out it was a “300+ page pdf of every single tweet I’ve ever liked with the word fuck in it.” They helpfully posted pictures of the report, so I could learn it came from a company called Fama, whose website says right up front about the company, “The smartest way to screen toxic workplace behavior.” More reading of Bruise’s Twitter posts revealed that the Fama work was part of a bigger screen by a company called Sterling Talent Solutions, a company I know and one that has a good rep. Still, if you see all the mockery this meme sent forth over the weekend, you’d start wondering if this was the smartest way to do anything.

There’s so many things this episode makes me want to comment on, starting with, the first rule of background checks, the check should cause you less embarrassment than what is found in said check. I want to skip the whole idea though, of how much risk can be managed by examining which Twitter posts one likes. Instead, I land on two things that transcend the genesis of this meme. It’s not about what specifically they were doing here but the constant desire that background researchers can be replaced with machines and then, who can ever handle a 350 page background report. In most ways, these are two sides of the same coin.

I’ve been doing background research for 30 years. For at least 26 of those years, someone has been telling me that they have a machine, a program, an algorithm, an aggregation, that would replace me, I remember clearly, one thing developed called Hoover because it supposedly sucked up all the information. All these systems are driven by the fact that there are too many backgrounds to run; with too much information to digest, and too many things to consider. Companies longed for a Bat Computer, which we all knew, spit out simple answers. As the Twitter/Fama/351 page episode showed, simple solutions often produce the most pages.

There are over 3,000 emoji’s. Yes, I Googled it. Because when you say fuck, what did you really mean. If we cannot see the sparkle in your eye, feel the vibrations of your vocal timbre, how do we know which fuck you meant. As we’ve dropped our land lines and picked up our text devices, we’ve had to create over 3,000 little pictures to help us know what we mean. Your algorithm may not read emoji. If they did would they still understand the meaning. In my family of mememez, the smiling face with hands is “jazz hands”; the official definition is hugging face. Would your AI know that? Just a nibble at how complicated results can be to interpret.

Listen, everyone needs a way to digest hundreds of pages of results. Ostensibly, the machine does it faster. All search is based on some basic query. Is this word, these words, in the document or public record. Everything else is gravy. Where in the document is it, what other words are there, near, not there. The more likely the word or words appear in documents, the more ways we have to figure out how to find less. And the harder and longer it takes to do such work. That’s the real question, the real answer. How do you look at everything you need to look at without looking at more than you need to look at. It’s tempting to think a machine can do it better. 350 pages of results says not yet.

I remain unconvinced that a computer program can do it right. Like I say, any program that sends you 350 pages of results is not doing it right. You need a program to read the 350 pages and tell me what it means. Or you can have me go through those 350 pages of likes and tell you if there was anything that really mattered. As a non-machine, it is easier for me to not to really care much about how many fucks you liked, because I can tell if liking said tweets really says something about you. I can look at your full social media profile, what, in general is it saying. I can measure that against other things that arise. Do the public records reveal areas of concern. What makes sense, what seems off. Are you really toxic for liking a tweet?

In real life, background results do not come with emoji’s. Next to the bankruptcy, there is no winky face. Sorry just kidding, did not mean to go broke. We take things as they appear. What they mean. Except we know how to make the story. If we cannot say it in less than 20 pages, we are not doing our job.