Resumes, Fact-Checking, and LinkedIn

William Grosso @ July 12, 2008

THE BACKGROUND:

In the bad old days, before everything was on-line and available in structured ways, people used postal mail to send paper copies of their resumes to potential employers.

And, a surprisingly large percentage of the time, their resumes were inaccurate. A somewhat smaller percentage of the time, the resumes were deliberately inaccurate¹.  Depending on the industry you’re in, and the area of the world you live in, and the relative level of economic stagnation currently being experienced, the number of people who lie on resumes is astounding. As the article referenced above puts it:

A recent survey by Morgan & Banks that found that an alarming 17 per cent of male and 7 per cent of female respondents admitted to lying outright on their CVs to secure a position, with those earning $50,000-$100,000 most likely to commit what the Americans call “resume fraud”. If you think those figures are causes for concern, how about this one: an astounding 80 per cent of Silicon Valley employees admit to having lied on their resumes.

This problem lead to lots of moralizing about how lying is bad and people applying for jobs shouldn’t do it².

Which, of course, stopped very few people — people who are lying on their resume probably aren’t going to stop lying simply because someone claims that lying is bad.

Which leads to LinkedIn. In the past 6 years, LinkedIn has turned into one of the web’s busiest sites. And it seems to have three major uses:

• Contact management. More and more people I know, including me, think of LinkedIn as a portable contact manager.
• Online career history. This is a fancy way of saying “online resume”– LinkedIn also gives you the ability to collect endorsements and people tend to include fewer details in their profile on LinkedIn than they do on their resume. But, by and large, LinkedIn is well on its way to replacing resumes.
• Recruiting tool. This is symbiotic with the online career history– LinkedIn has created a knowledge-intensive job market.

THE PROBLEM:

Given the last two bullet points, it’s reasonable to wonder: how good is LinkedIn at detecting errors or fraud?

The answer seems to be: not very good at all.

I can easily, just using the companies I’ve worked at and people in my network, find people who’ve made mistakes³. I can find people who’ve stretched the truth a bit⁴, and I can find people who’ve flat-out lied.

Even more insidiously: LinkedIn recommendations, while rarely factually inaccurate, seem to suffer from two serious problems:

• Permanent and public visibility. Recommendations are readable by the recipient, will be shared with the world, and are often done as part of a “recommendation trade” (e.g. two people agree to recommend each other. Or one person recommends the other and the second person reciprocates). This is not a recipe for nuanced evaluations or factual declarations of strengths-and-weakness. In fact, many recommendations seem to best described by the phrase unctuous, oily, and unseemly flattery.
• Bad timing. People tend to write recommendations in clusters. The way it works is: a company downsizes and everyone who’s been downsized writes a recommendation for everyone else. That is, recommendations tend to be written, en masse, at a time when the people involved are emotional and defensive, and the impact of anything other than a glowing recommendation is going to be magnified.

THE PROBLEM IS BIG:

Right now, LinkedIn seems to be ignoring these issues. And that might seem reasonable– this is the online equivalent of problems that existed with resumes anyway; why should we expect that bringing things online would get rid of it?

But there are some crucial differences here. LinkedIn’s reliability problem is important because:

• It’s bigger than resume fraud. Resume fraud was inherently a local problem: A printed resume has a short shelf live (usually limited to one job search). Online career history is much more generic. More people are seeing your career history on LinkedIn than ever read your resume. And they’re doing so over a much longer period of time.
• It impacts you, even if you’re honest. Many job searches are done on LinkedIn. Recruiters or hiring managers search for, and reach out to people who would otherwise not have heard about the opportunity. Now if someone else has falsified their career history and is more attractive than you as a result, you don’t hear about the opportunity.
• It impacts you even more. Right now, LinkedIn is generally perceived to be reliable. People generally trust, and act on, the information in it. What happens if that trust dimishes or evaporates completely? Your career history in LinkedIn will lose value.

WHAT TO DO:

I’m not sure whether this is a start-up opportunity (an OpenSocial-based widget could solve some of these issues), something that the community (or individual communities) should handle externally to LinkedIn, or something LinkedIn will eventually try to handle, but I do think it’s a serious problem.

The good news is that it’s also a solvable problem. Some obvious thoughts include:

• Errors are mostly detectable. LinkedIn is collecting a mass of career and personal information in a very structured way.  And even the unstructured parts areamenable to semantic analysis: they’re short and written in general-purpose English with a very specific context. Fact checking won’t be perfect, but it should be possible to spot many issues.
• Community policing is possible. The great lesson of Craigslist is that if you put an abuse flag link next to something, people will report abuse. What if you enabled people to disagree with a recommendation, or register their level of agreement?

The only question, to me, is will the problems be recognized as severe enough that fixes become inevitable? Or will we just live with the current state of affairs.

1. e.g. the person who wrote the resume lied↩
2. As well as some very interesting academic research about amoral behavior. A follow-up google search on the Sontag Connector is also illuminating.↩
3. Errors in dates, for example↩
4. For example, claiming to be a team lead when they were an individual contributor or implying sole-creator status on something that was a shared effort↩

Filed under: Business of Software, Dyspeptic, Society | Share on Facebook