Christian Rudder and Corporate Ratings

One of the studdest book chapters I’ve read is from Christian Rudder’s Dataclysm. Rudder is a cofounder of OkCupid, now part of the portfolio of matchmakers. In this book, he has taken insights from OkCupid’s own data to draw insights about human life and behaviour.

It is a typical non-fiction book, with a studmax first chapter, and which gets progressively weaker. And it is the first chapter (which I’ve written about before) that I’m going to talk about here. There is a nice write-up and extract in Maria Popova’s website (which used to be called BrainPickings) here.

Quoting Maria Popova:

What Rudder and his team found was that not all averages are created equal in terms of actual romantic opportunities — greater variance means greater opportunity. Based on the data on heterosexual females, women who were rated average overall but arrived there via polarizing rankings — lots of 1’s, lots of 5’s — got exponentially more messages (“the precursor to outcomes like in-depth conversations, the exchange of contact information, and eventually in-person meetings”) than women whom most men rated a 3.

In one-hit markets like love (you only need to love and be loved by one person to be “successful” in this), high volatility is an asset. It is like option pricing if you think about it – higher volatility means greater chance of being in the money, and that is all you care about here. How deep out of the money you are just doesn’t matter.

I was thinking about this in some random context this morning when I was also thinking of the corporate appraisal process. Now, the difference between dating and appraisals is that on OKCupid you might get several ratings on a 5-point scale, but in your office you only get one rating each year on a 5-point scale. However, if you are a manager, and especially if you are managing a large team, you will GIVE out lots of ratings each year.

And so I was wondering – what does the variance of ratings you give out tell about you as a manager? Assume that HR doesn’t impose any “grading on curve” thing, what does it say if you are a manager who gave out an average rating of 3, with standard deviation 0.5, versus a manager who gave an average of 3, with all employees receiving 1s and 5s.

From a corporate perspective, would you rather want a team full of 3s, or a team with a few 5s and a few 1s (who, it is likely, will leave)? Once again, if you think about it, it depends on your Vega (returns to volatility). In some sense, it depends on whether you are running a stud or a fighter team.

If you are running a fighter team, where there is no real “spectacular performance” but you need your people to grind it out, not make mistakes, pay attention to detail and do their jobs, you want a team full of3s. The 5s in this team don’t contribute that much more than a 3. And 1s can seriously hurt your performance.

On the other hand, if you’re running a stud team, you will want high variance. Because by the sheer nature of work, in a stud team, the 5s will add significantly more value than the 1s might cause damage. When you are running a stud team, a team full of 3s doesn’t work – you are running far below potential in that case.

Assuming that your team has delivered, then maybe the distribution of ratings across the team is a function of whether it does more stud or fighter work? Or am I force fitting my pet theory a bit too much here?

The Unpopular People’s Network

Recently I had blogged about how I find it hard to get along with people who are generally “popular”, and find it so much easier to get along with oddballs, people who have a reputation of being “arrogant”. So I’ve been discussing this with this one old friend, who is far from being universally popular, and (back when we had a large common network) had a reputation of being arrogant.

So we were recently talking about a mutual acquaintance and she said “She’s very cool. You’d like her. She’s far from ordinary and normal 🙂 “. Now, I must point out that this conversation was conditioned by our earlier discussion about my blog post, but it is interesting how this friend assumes that I’m going to like this mutual acquaintance because she’s also, like the two of us, an “oddball”!

So I wonder if there’s something about us oddballs that attracts us to each other. If there is some kind of inherent solidarity between us because we are all of the type that don’t make us particularly popular. There is no guarantee of course that a randomly chosen pair of oddballs get along, but I wonder if the probability that two randomly chosen oddballs get along is higher than the probability that one “normal guy” and one oddball getting along!

And coming to the data that Christian Rudder has put in Dataclysm, on people getting 1s and 5s being more likely to get a date than straight 3s, I wonder how it will look if we are going to condition the data on the rating profile of the reviewer – maybe someone who has a lot of 1s and 5s is more likely to give 1 and 5 ratings to others? And 3s give 3s to others? It would be interesting to find out, except that the data is not public!

Getting along with popular people

I’ve been thinking about this for a while now but it all came together a while back. The basic funda is that I find it extremely hard to hang out with people who are generally popular and who everyone wants to hang out with. On the other hand, I find it significantly easier to hang out with other people who generally most people consider as being “arrogant” and hard to hang out with.

I wonder if it is connected with what Christian Rudder writes in Dataclysm on people who have been rated a few 5s and a few 1s being more likely to find a partner than one who is rated a consistent 3 (holding average rating constant). Basically if there is someone who is generally popular, they are something like a consistent 5, and they are perhaps generally popular because they exhibit the kind of behaviour or attributes that most people like. Effectively they cater to what I can uncharitably call the lowest common denominator of popularity among people, and that generally means they spend most of their effort catering to that (being “generally nice” and all such) that there is very little idiosyncrasy that they can offer which makes them interesting!

And with time the fact that they are popular affects them, and they expect that everyone like them to the same (high) extent as everyone else! And when you start asking yourself what the big deal about them is, and start wondering why they’re so popular, there is a “respect mismatch” – the respect you are willing to offer them doesn’t match up to the respect they expect (thanks to being generally popular), and you can’t hang out for long.

With people who are generally not particularly popular and branded as “arrogant” by most people, firstly there is no expectation of respect as they generally know that they are not particularly popular. Secondly, the fact that makes them arrogant also makes them interesting to people who are interested along that axis. The fact that they are not generally popular means that there is an idiosyncrasy about them, and if you happen to like that you can get along very well with them!

Of course, I admit to selection bias here. There definitely exist people who are generally classified as “arrogant” who I also find arrogant and don’t hang out with. But there exist a lot of people who are generally classified as “arrogant” who I get along quite well with!

Going back to Rudder’s ratings, I’m likely to rate people who are generally considered “arrogant” either a 1 or a 5 – the idiosyncrasy sends them to either extreme. Thus there are a few of them who I love hanging out with irrespective of what the world has to say about them. As for the popular guys, I’m very likely to rate them a 3 – basically unspectacular, and going by Rudder’s theory, “meh”. And since they expect the general counterparty to rate them higher than that, there’s a mismatch when I meet them and things fall apart.

Makes sense? What has your experience been of people in relation to how other people rate them?

Good boys don’t get laid

Last night I bought Christian Rudder’s Dataclysm: Who We Are (When We Think No One’s Looking) and started reading it. I’m now about 10% into the book, well past the Kindle sample (I bought the book in entirety after I’d finished the sample). I’m past the first couple of chapters and am now reading a chapter on the contributions of Twitter to linguistics.

Rudder is a co-founder and Chief Data Scientist at the matchmaking website OkCupid, and he draws upon some aggregate data that his website has collected to point out some rather interesting stuff about how people think, view themselves, and the kind of partners they are looking for. One very interesting piece of analysis (which includes a couple of brilliant graphs) shows the preference for the partner’s age among men and women of different ages. So far the book has been absolutely spectacular.

The part of the book that I’ve found most fascinating so far is the one on averages and variances. Rudder looks at the average ratings of a large number of women registered on OkCupid (as rated by men) and tries to correlate their ratings with their success on the site (measured in terms of the number of messages they have received from men on the site wanting to date them). Given the scale of the data that Rudder has access to (rather large), the results are rather stupendous.

What Rudder finds is that for a given level of average rating for a woman, the higher the variance in her rating, the more the number of messages she receives. There are some quirky statistics he quotes (a lot of which has been extracted in this post on Brain Pickings – it was after a friend sent me this post that I got interested and bought the book) which show that women who are consistently rated a 3 (on a scale of 1-5) by men are much less likely to get a message than someone who gets a mix of 1s and 5s.

From this Rudder concludes that negative ratings actually boost a woman’s chances of getting a date – the fact that a number of men have rated someone unattractive means that there is something about her that a lot of men don’t like. This implies that the “competition” for getting her is possibly low, and you might be able to get a “bargain” or an “arbitrage” if you are able to get her.

While this is a plausible and rather palatable thesis, I have an alternate explanation for the same data – I posit that low ratings don’t matter. Some people might have rated you lowly but they don’t matter since they aren’t interested in you. What matters simply is the number of high ratings that you get – people are always on the lookout for spectacular people to date, and by getting a number of 5s, you are showing that you are found rather attractive by a number of people. The ratings of 1 that you have received are an anomaly – messages of rejection from people who don’t want to date you, and all they do is to pull down your average. A better way of comparing women would be to throw away the bottom 20% of all ratings that a woman gets and then calculate the average – and a lot of 3s that Rudder has analysed in his book are likely to come out as something more than that.

Irrespective of the reason for the correlation of variance with attractiveness, though, what is undisputed is that people look for spectacular people to date. If you are a consistent three, irrespective of whether you go by Rudder’s thesis or mine, a large number of men are likely to rate you as being “unspectacular”. When given a choice between dating someone who is a “common minimum program” on most dimensions and someone who has a “spike” (as recruiting management consultants like to put it), you are likely to be more interested in the one that has the spike. What you consider to be a spike might be considered to be a trough by others, which probably leads to an average average rating (but high variance), but it is the spike that attracts you to her.

The problem with the arranged marriage market in India is that it is set up such that people show off their “average” side. As I had argued several years earlier (back when I was in the market), the Indian arranged marriage market is dominated by people who are themselves “common minimum programmes” and who are looking for “common minimum programmes” to marry. Thus, if you want to enter that market yourself, you try to mould yourself as yet another common minimum program and try to hide your spike rather than to enhance it (it is also a result of counterparties sharing notes in the arranged marriage market, something that doesn’t happen in the dating market. If you have a spike that one girl considers to be a trough, her folks are likely to tell people known to them about your trough (which is actually a spike), and that might pull down your average rating).

Most Indian parents bring up  their kids to become good materials in the arranged marriage market, and since it is the unspectacular CMP that succeeds in that market, parents aspire to get their grown up kids to fit such moulds. Any possibly deviant behaviour is quickly dissed, non-standard careers are strongly discouraged, you are encouraged to dress unspectacularly and so on. Taking this together with Rudder’s thesis, what this means is that if you prime yourself for the arranged marriage market, you are losing out on the dating market!

What it takes to be a success in the arranged marriage market (solidity, unspectacularity, CMPness) are directly at odds to what it takes to succeed in the dating market (a spike, quirkiness, character) and so once you have decided to enter one market you automatically become a failure in the other. This thesis also explains why people who break up in their mid/late twenties and who consequently enter the arranged marriage market (possibly since it offers the quickest chance of a rebound) struggle significantly in that market – they have been primed for the dating market which makes them unhot in the arranged marriage market.

One “spike” that I consider to be a part of my character is this blog. Back in the Benjarong conference, I was given sage advice that I do not disclose this blog to prospective brides from the arranged marriage market, thanks to posts like this one and this one. Finally I ended up marrying someone who I met as a consequence of this blog (this post to be precise – she later told me) – who messaged me on Orkut saying she likes my blog, because of which we got talking and so forth. It was the spike – possibly considered abhorrent by many – that was responsible for my finding my wife!

So decide which market you actually want to be in before you prime yourself. “I’ll also casually look at the other market” is never likely to work.