Unfriendly Algorithms

I have a friend. He is quite eccentric. Doesn’t talk too much, but is not necessarily a good listener. I have another friend. He is quite rich and equally humble. He is also very funny. He can easily get anxious about small things, but tries to hide it. One friend from many years ago has a big heart. A really BIG heart. He isn’t rich, neither is he eccentric. He talks and listens in equal measure. And there’s another one. He is intelligent. Sometimes too much for his own good. Yet, you could say that he is the epitome of being human (I don’t mean it in the way they declare on T-shirts) – in the true sense of that word. Be-ing Human.

I don’t know if any of these friends would lay down their life for me. I don’t believe any one should lay down their life for any one. If you have to do something with a life, you have to convert it to joy. It doesn’t matter if it’s yours or someone else’s.

Like my friend—another one—who is a sun. I call him that because of his energy. He radiates it, when he is around and everybody breaks out in a summer dance. He doesn’t read much. But, I have a friend who reads a lot. You could call him eccentric – not because he is – but because he likes to think so. My friend who is rich reads little, but understands a lot. Unlike another friend of mine who doesn’t attempt too much of understanding, but is the sweetest person on earth. There is no way you could feel unhappy when you are around her. She is not the sun, so to speak, but you can feel earthly warmth around her, if you know what I mean. I have many friends. Let me tell you about another one. He is, as far as I know, my oldest friend. Not by age. (Perhaps I should say, he has been my friend for the longest time). If you are talking of age, my “oldest” friend will be 70 years next month. My oldest longest friend, I have known since 1980. That’s 33 years. We lost touch for a couple of decades in between. It didn’t matter. When we met, we managed to cover up the couple of decades in a couple of hours. There are other friends. I can tell you all about them. Right now a few are upset with me, a few are away, some busy and the rest – I don’t know what they are up to.

These are my friends from primary school, secondary school, college, and various workplaces. I know some of them because of this blog and one because of Facebook. I’ve met these friends in unique circumstances, at different times, in varying contexts. School, college, and workplace friendships happen through association. What’s important, is to note what happens when you are no longer associated with the institution. What’s even more important, is that you usually cannot decide who your friends will be or should be. It takes two hands to clap and all that jazz.


Brooke Bloom plays Emily Morton in Person of Interest S02-E14. She is creator and founder of a matchmaking site called Alchamentary.  According to the PoI Wikia:

Emily’s social networking program [Alchamentary] not only helped someone find their perfect match regarding relationships but also business and other fields.

Let’s not bother ourselves with what happens in the rest of the episode.


Instead, let’s bother ourselves with this tweet from Mahendra Palsule:

Yday, after reading more about Netflix’s recommendation algo, I was imagining of an equivalent in life for friends.

It was followed by:

Our friends do perform the algo’s functions, but automating it would be so much more efficient.

I was quick to disagree, but then my disagreement was made redundant by two tweets; one from Mahendra:

OTOH, the human factor would be lost. There is a joy when you connect via someone, which wouldn’t be there anymore.

And the other from Bala (who was a part of this conversation, all through)

Efficient, yes. Effective? Not sure. For some things, I would still trust the human element. Old fashioned 😀

My disagreement was accompanied by my wish to respond in long-form. (This post is the response). To my mind, recommendation algorithms work well with fixed and limited attributes. It’s impossible very difficult to do that with humans. Just because you and I are friends does not mean that he and you will become friends. Actually, you might. But no algorithm can say for sure. That unfriendly algorithm will work on available data – and in these times when we struggle to know our selves – what data can we offer to the algorithm? In spite of all the attributes, we never know how we become friends. Why we become friends. This is one aspect – the human factor or the human element – that’s still a domain that has (thankfully) not yet been ensnared by machines. I hope that it doesn’t happen in my lifetime. I’d detest a machine telling me who I can be friends with.

The first rule of being friends is that we should not want to be friends, with anyone. There is no “making.” We should allow whatever happens with the people we meet. Even after (and if) we become friends. That’s also the last rule of being friends.

The only thing you can do, is keep the faith.

Keep the Faith

Keep the Faith

Afterthought: This post is obviously not about describing my friends, though it does sound like that, I realise. And I should tell you this: some of those friends I mentioned at the start, are über-curious. They’ll try to decipher who I am talking about. But, I am smart. If they ask me, I’ll just say, I wasn’t talking about you; it’s someone else you haven’t met.


6 thoughts on “Unfriendly Algorithms

  1. Interesting post! (I watched that PoI episode too, and though I haven’t been on twitter much, hat-tip to Mahendra for opening out a fascinating subject for discussion.)

    My view: It is not that such algorithms can never exist for human relationships, but that human relationships may be too complex to be meaningfully modeled. I don’t mean too complex to be computed (I believe we do have adequate computing power today, at a fairly affordable cost). I mean too complex to be modeled – accurately and scientifically. And even then, they are likely to be probabilistic rather than deterministic models. Why do I say this? Because I suspect there’s a ‘method to the madness’ of human relationships. And where there’s a method, there can be an algorithm.


  2. Human friendships are largely beyond algorithmic understanding today. As Hemant said, they are too complex to be modeled. It is like predicting the weather. Scientific weather models have, for decades, failed to accurately predict thunderstorms and the monsoon in India. But even the ability to model such complex systems as the weather are improving day by day. Then why not human relationships?

    The xenophobic reaction to algorithms taking over our lives is not new and has existed for centuries. Scientific advance outpaces human acceptance. Technological advances outpace human adaptation.

    Once upon a time, folks would have “detested” a machine telling them which restaurant they should visit. Today, this is *expected* from any *smart* phone device/algorithm. We *expect* algorithms like Netflix’s to tell us which movie we should watch next; in earlier eras, such a recommendation from an algorithm would have been detested.

    An algorithm to tell us who we should be friends with next? Not so far into the future. It is our choice if we choose to detest such an unfriendly algorithm, or expect the next generation to expect it. As a parent, one can choose to detest technological advances or to accept and adapt to them. As a human, one can choose to treat algorithms as unfriendly, or make friends with them to discover new friends.


  3. Pingback: It’s Not About Photographs -III | Gaizabonts
  4. This is a fantastic post.

    I’ll not ask which one I am, because I know I’m not one of them ;-). But, like you said, that’s not the point of the post.

    Two points:

    I agree with your assessment that with multiple — and more importantly non-quantifiable — attributes, it’s very hard to solve that ‘problem’ algorithmically. It might be plain intractable. The only hope would be fuzzy/probabilistic algorithms. But note that our decision systems are also precisely those fuzzy/probabilistic algorithms. So right now, some very intelligent machines are making those decisions! Whether those machines can make something as complex as them is the central problem in AI.

    Second point, if that were to happen, I won’t resent it. I don’t hope it will, neither do I hope it won’t. Even if it does, I will not use it (or so I think). And the reason is, as you say it: one doesn’t decide to become friends. Being friends, for long term, is being able to grow in directions where you still find a point where you can meet.

    And this also brings back memory of what I had written, in albeit different context —

    From http://asuph.wordpress.com/2010/05/12/edge-question-and-random-ruminations

    The thing is this: soon, social media will perfect the art of suggestion. I’m not sure that it is such a good thing. Of course, you will like what it suggests. That’s not the point. It is this: you’ll stop growing, because you’ll only consume what you’ve already liked, in different bottle. When friends suggested, they had the audacity to tell you: “trust me, you’ll like this”, when that it was something very different from what you had consumed before. It was not perfect. I’m not sure that was so bad. I fear that this social media oracles will slot you, just like actors are slotted because they were good at one specific role. Like an indulgent mom who feeds a child just sweets, because he likes it, social media will indulge us by feeding more and more of what we are sure to like. We’ll settle down into comfort zones. I guess that’s nothing new. People have always slotted themselves into ideologies, religions, cultures, and so on. But web 2.0 was supposed to be our answer to that (or was it?) — freeing us from the tyranny of mainstream, letting us define and redefine ourselves, ad infinitum. Instead, there is a risk of hardening our identities. And it will be so much earlier, so much … well serendipitous!

    That ‘serendipity’ could be entirely outside the scope of any algorithm.


Use your Twitter, Facebook or your WordPress account to comment

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.