Endorsing only similar skills? Really?<p>What if I know I'm bad at something and decide to hire someone who's a professional in this area? And this person does an amazing job. I want to endorse this person to acknowledge the obviously great skill he possesses. But I can't. My skills are not similar.
The problem's even worse than the article says.<p>Several people from my company's cloud services division have endorsed me as an expert in cloud computing (about which I know very little). It appears that simply my putative expertise, combined with my relationship with them, is designed to make potential customers feel better disposed to them when those customers looks to see how much expertise our cloud people are surrounded with.<p>So there are incentives all over the place that cause this thing to be misused.
"...I’ve also been recommended as an expert in a whole bunch of things I know nothing about."<p>Is this even possible? I've only recommended or been recommended for skills I or others have <i>claimed</i> to have in the LinkedIn profile. Nobody's endorsed my Real Estate expertise or deep immersion in String Theory, for instance.
There is also the possibility of leaving "positive" endorsements that, uh, aren't: <a href="http://i.imgur.com/mcgek.png" rel="nofollow">http://i.imgur.com/mcgek.png</a>
I was endorsed for a skill by someone who I haven't seen or talked to for ten years. The last project we worked together on was in 1997.<p>It would be all well and good, except it is a skill in which I would not claim professional expertise. Ten years ago, I was a one-eyed man in the land of the blind. But not today. The area has expanded and credentials have been formalized, and I have made little effort to keep up beyond what I encounter from time to time.
I don't think any recruiter or HR person looking on LinkedIn places any weight on the recommendations whatsoever. It seems to be an overt attempt to get people to engage with the site more, thus making LinkedIn's active user numbers look better. LinkedIn is one of those networks that is only really useful on an occasional basis when you need it to find a particular person or connection at a particular company.
I thought the real reason for their recommend system was to solve the high friction of writing a recommendation.<p>It's a quantitative solution with a qualitative result. I.e. the more people who endorse your claimed skills the more you know there might be something about it.<p>I have both designers, developers and clients endorse me and I think it's a straw man to claim that it's simply because people want other to endorse them.
I'm mostly curious why no one on LinkedIn (a) saw this one coming, or (b) implemented any form of an easy fix. I put two in this article, but there are probably dozens when you consider more advanced machine learning algorithms designed to really match users together. This little feature seems to have so much potential, I'm baffled why they haven't given it the attention it deserves.
LinkedIn's recommendations mostly measure social standing of an individual. I have got lots of them from my buddies, I have got many in reciprocation to ones given and well, I have also reciprocated on occasions.<p>Maybe making them anonymous could alleviate the reciprocation factor but the social standing still remains. I guess the latter is also a problem with the traditional CV vetting.<p>Bottom line? Do your homework on CVs but never skip the actual skill test, unless you want a likeable but not very competent worker.
I am pretty sure there is a huge potential to run machine-learning algorytms on the recommendation data to detect anomalies.<p>It must be fun to be a computer science guy at Linkedin.
I agree with the people calling the solution as bogus. I'm not a pro python developer, but I can sure vouch for the skills of many developers I've worked with based on my experience in working with them. And that's a fairly accurate recommendation.<p>Sorry the similar skill set limit is bogus.
The LinkedIn Skills Recommender is one of the most impressive features LinkedIn has released in a long time, both from a UI perspective and a data science perspective. Here's what they are trying to do (long term)<p><a href="http://news.cnet.com/8301-1023_3-57554673-93/linkedin-ceo-site-will-become-economic-graph-over-next-decade/" rel="nofollow">http://news.cnet.com/8301-1023_3-57554673-93/linkedin-ceo-si...</a><p>Calling this "broken" shows a misunderstanding of what LinkedIn is trying to accomplish. LinkedIn has a huge amount of data on you and your professional relationships, starting from how you know the other person to how many times they have viewed your profile.<p>They are trying to increase the depth and value of your LinkedIn profile by strengthening the graph of data connections, and encouraging users to enter more skills data into your own profile. They are also trying to increase engagement amongst their user base.<p>The author suggests that if Linked either:<p>a) Limit skills recommendations to people with similar skills.<p>If you did this, you will collect an order of magnitude less recommendation data. How many people in your LinkedIn list actually know the same skills as you? I have a lot of connections in adjacent industries. Every time I see that someone has recommended me, I have two incentives. First is to ensure that my skills profile on LinkedIn is complete (so that I may get the appropriate recommendations). Second is to recommend others effectively.<p>If someone perceives you to be good at a particular skill, even though they don't know that skill themselves, isn't that valuable? LinkedIn is about managing your reputation - not just among people with the same skills, but amongst your entire professional network.<p>For example, if you knew someone was good at MongoDB, even though you didn't know anything about MongoDB, couldn't that be something that is important on your resume? Don't you think it could help you get a job?<p>b) Create a new feature to rank the "strength" of the recommendation.<p>FYI - Here's LinkedIn's skills page for "C#": <a href="http://www.linkedin.com/skills/skill/C%23?trk=skills-pg-search" rel="nofollow">http://www.linkedin.com/skills/skill/C%23?trk=skills-pg-sear...</a><p>- They've got some clustering algorithms to find similar skills
- They know who are the most "influential" people with that skill (wow)<p>So, this is a good idea, but I think that LinkedIn is already doing it. I would not be surprised if they are using it, but perhaps not in the way you think. It could be used on their recruiting tool to find the best "C#" developers out there.<p>It often doesn't make sense to publish these kinds of "reputation" scores publicly; that will just encourage you to game it. For example, if you knew that the strength of your recommendations impacted how often you showed up in candidate search results, you might start encouraging your friends to create bogus recommendations, just to move up the rankings.
I think there's a much simpler way to do fix it - force users to anonymously unendorse in such a way as to maintain a fixed ratio of endorsements to unendorsements. Don't show unendorsements or a negative endorsement level.<p>Suddenly incentives are much better aligned.
Maybe a [1-5] rating system would be best... although it'd need some kind of scale.<p>I agree that the current is flawed - my Dad has recommended me for everything from Python to MySQL, and he's an artist!
Here is a related post I was about to make but thought it would be relevant as a comment.<p>"Why I disagree with LinkedIn endorsements"<p>I was having a discussion with a VP of Lending last week about a new company I am working on to supplement financial information for loan purposes. The topic of LinkedIn, credibility, and endorsements came up and he suggested that perhaps someday in the future loans could be based off of your endorsements. I disagreed, stating that I didn’t think endorsements were very accurate but couldn’t fully explain why. This was my analysis.<p>1) A break-down by LinkedIn contacts:
I have been endorsed by 26 people of the 845 (3%) I’m connected to on LinkedIn. I began realizing I didn’t know some of the people endorsing me so I created some criteria for my endorsers:
-Never met: 19% of endorsements
-Never done business with: 50% of endorsements
-Haven’t seen in 5+ years: 46% of endorsements
-Didn’t recognize their name and had to pull up their profile: 19%<p>2) I decided to Social Graph my profile and found that the largest categories of people I know are: 1) From the financial industry 2) CEOs or executives 3) Entrepreneurs 4) From Oregon/hometown/school alumni 5) Team members from my previous company. These all make sense because the last company I founded works with the financial industry, so my social circle has evolved to primarily entrepreneurs and people in the financial industry.<p><a href="http://ow.ly/i/1h9IB" rel="nofollow">http://ow.ly/i/1h9IB</a><p>3) Of these groups, who arguably know me the best of anyone I had:<p>-0 active CEO’s or C level executives (non-entrepreneurs)
-0 co-workers or team members
-0 investors from my previous company (none of which are not on LinkedIn)<p>4) The first listed 2 endorsements of the initial 13 I selected became my #1 and #2 skillset.<p>Some thoughts on Endorsements<p>1) The quality of the endorsements has a ways to go.
-Endorsers often have a small impact on one’s real life, but can majorly impact one’s professional perception.
-It seems that a fairly high % of people who endorse have some type of vested interest be it getting business from LinkedIn, networking for jobs, gaining topic expertise or driving traffic.<p>2) Your mind is tricking you.
-People choose what is listed first: My #1 and #2 skills and endorsements were the first two skills listed.
-Visual Deception: People love graphs, they remember them. They also only remember a couple things about you, like your #1 skill. This is a terrible thing because by nature we automatically assume this person is not as good at the other things and we begin typecasting. Would you agree that your #1 endorsement is your strongest skill? Would you say you are twice as good as your #1 endorsement as your #2 endorsement? These are some of the impressions this is building.
-Group Think: If someone else thinks you’re good at X, you must be, so I’ll endorse that also. This creates an echo chamber.<p>3) The dynamic of endorsing people can get awkward, which may prevent accurate endorsements. Would you endorse your professor while taking a class from them? How about your manager while working for them? The owner of the company where you work?<p>4) Higher ups aren’t on LinkedIn. The people who arguably have the most “influence” on me as an entrepreneur are not actively on LinkedIn and certainly are not out endorsing people.<p>The bottom line is that a vocal minority may influence the majority so be careful what impressions your endorsements may give of you. The people who spend the most time on LinkedIn are probably not the ones you want determining who you are.<p>Disclaimer: Thank you to those who have endorsed me. I know you have done so with the intention of helping me and it is appreciated. This post is about the social flaws with LinkedIn’s Endorsements and also about the behavioral economics behind online social influence.
I've been gradually pulling away from LinkedIn. Right now, I have for my profile:<p>"I don't reveal history without a reason, so my past jobs summary is blank.<p>I'm a New York-based software engineer who specializes in functional programming, machine learning, and language design."<p>I really don't think it's a good idea for people in their 20s to reveal their career histories to the public. A lot can go wrong, and there's questionable upside. All kinds of shit can happen, and who's to say that a person won't need creative repair years down the road (e.g. one's worst enemy takes over a prior employer, years later, and gives a bad reference, and the only way to overcome it is to depart from the truth)? I don't think most people are age will get burned by a need to lie overtly (because that's rare) but more by accidental consistency risk and benign forgetting of details.<p>My issue with LinkedIn is that, although it's 21st-century technology, and it has some stellar people working on it. it seems to me to be still a tool for evaluating technical people by 20th-century means: prestige of company, title, dates of employment. Then there is the game aspect of trolling other people for recommendations to build a more impressive profile, which seems like an enormous source of tension and pissed-off-ness ("you recommended him, why can't you recommend me?") I realize that it has to embrace the 20th-century mentality to get employers to take it seriously, but it's sad that it has to accept those kinds of compromises.