Hot hands in safaris

We entered Serengeti around 12:30 pm on Saturday, having stopped briefly at the entrance gate to have lunch packed for us by our hotel in Karatu. Around 1 pm, our guide asked us to put the roof up, so we could stand and get a 360 degree view. “This is the cheetah region”, he told us.

For the next hour or so we just kept going round and round. We went off the main path towards some rocks. Some other jeeps had done the same. None of us had any luck.

By 2 pm we had seen nothing. Absolutely nothing. For a place like Serengeti, that takes some talent, given the overall density of animals there. We hadn’t even seen a zebra, or a wildebeest. Maybe a few gazelles (I could never figure out how to tell between Thomson’s and Grant’s through the trip, despite seeing tonnes of both on the trip). “This is not even the level of what we saw in Tarangire yesterday”, we were thinking.

And then things started to happen. First there was a herd of zebras. On Friday we had missed an opportunity to take a video of a zebra crossing the road (literally a “zebra crossing”, get it?). And now we had a whole herd of zebras crossing the road in front of us. This time we didn’t miss the opportunity (though there was no Spice Telecom).

Zebra crossing in Serengeti

And then we saw a herd of buffaloes. And then a bunch of hippos in a pool. We asked our guide to take us closer to them, and he said “oh don’t worry about hippos. Tomorrow I’ll take you to a hippo pool with over a fifty hippos”. And sped off in the opposite direction. There was a pack of lions fallen asleep under a tree, with the carcass of a wildebeest they had just eaten next to them (I posted that photo the other day).

This was around 3 pm. By 4 pm, we had seen a large herd of wildebeest and zebra on their great annual migration. And then seen a cheetah sitting on a termite hill, also watching the migration. And yet another pool with some 50 hippos lazing in it. It was absolutely surreal.

It was as if we had had a “hot hand” for an hour, with tremendous sightings after a rather barren first half of the afternoon. We were to have another similar “hot hand” on Monday morning, on our way out from the park. Again in the course of half an hour (when we were driving rather fast, with the roof down, trying to exit the park ASAP) we saw a massive herd of elephants, a mother and baby cheetah, a pack of lions and a single massive male lion right next to the road.

If you are the sort who sees lots of patterns, it is possibly easy to conclude that “hot hands” are a thing in wildlife. That when you have one good sighting, it is likely to be followed by a few other good sightings. However, based on a total of four days of safaris on this trip, I strongly believe that here at least hot hands are a fallacy.

But first a digression. The issue of “hot hands” has been a long-standing one in basketball. First some statisticians found that the hot hand truly exists – that NBA (or was it NCAA?) players who have made a few baskets in succession are more likely to score off their next shot. Then, other statisticians found some holes in the argument and said that it was simply a statistical oddity. And yet again (if i remember correctly) yet another group of statisticians showed that with careful analysis, the hot hand actually exists. This was rationalised as “when someone has scored a few consecutive baskets, their confidence is higher, which improves the chances of scoring off the next attempt”.

So if a hot hand exists, it is more to do with the competence and confidence of the person who is executing the activity.

In wildlife, though, it doesn’t work that way. While it is up to us (and our guides) to spot the animals, that you have spotted something doesn’t make it more likely to spot something else (in fact, false positives in spotting can go up when you are feeling overconfident). Possibly the only correlation between consecutive spottings is that guides of various jeeps are in constant conversation on the radio, and news of spottings get shared. So if a bunch of jeeps have independently spotted stuff close to each other, all the jeeps will get to see all these stuffs (no pun intended), getting a “hot hand”.

That apart, there is no statistical reason in a safari to have a “hot hand”. 

Rather, what is more likely is selection bias. When we see a bunch of spottings close to one another, we think it is because we have a “hot hand”. However, when we are seeing animals only sporadically (like we did on Sunday, not counting the zillions of wildebeest and zebra migrating), we don’t really register that we are “not having a hot hand”.

It is as if you are playing a game of coin tosses, where you register all the heads but simply ignore the tails, and theorise about clumping of heads. When a low probability event happens (multiple sightings in an hour, for example), it registers better in our heads, and we can sometimes tend to overrepresent them in our memories. The higher probability (or “lower information content”) events we simply ignore! And so we assume that events are more impactful on average than they actually are.

Ok now i’m off on a ramble (this took a while to write – including making that gif among other things) – but Nassim Taleb talks about it this in one of his early Incerto books (FBR or Black Swan – that if you only go by newspaper reports, you are likely to think that lower average crime cities are more violent, since more crimes get reported there).

And going off on yet another ramble – hot hands can be a thing where the element of luck is relatively small. Wildlife spotting has a huge amount of luck involved, and so even with the best of skills there is only so much of a hot hand you can produce.

So yeah – there is no hot hand in wildlife safaris.

Human, Animal and Machine Intelligence

Earlier this week I started watching this series on Netflix called “Terrorism Close Calls“. Each episode is about an instance of attempted terrorism that has been foiled in the last 2 decades. For example, there is one example of the plot to bomb a set of transatlantic flights from London to North America in 2006 (a consequence of which is that liquids still aren’t allowed on board flights).

So the first episode of the series involves this Afghani guy who drives all the way from Colorado to New York to place a series of bombs in the latter’s subways (metro train system). He is under surveillance through the length of his journey, and just as he is about to enter New York, he is stopped for what seems like a “routine drugs test”.

As the episode explains, “a set of dogs went around his car sniffing”, but “rather than being trained to sniff drugs” (as is routine in such a stop), “these dogs had been trained to sniff explosives”.

This little snippet got me thinking about how machines are “trained” to “learn”. At the most basic level, machine learning involves showing a large number of “positive cases” and “negative cases” based on which the program “learns” the differences between the positive and negative cases, and thus to identify the positive cases.

So if you want to built a system to identify cats in an image, you feed the machine a large number of images with cats in them, and a large(r) number of images without cats in them, each appropriately “labelled” (“cat” or “no cat”) and based on the differences, the system learns to identify cats.

Similarly, if you want to teach a system to detect cancers based on MRIs, you show it a set of MRIs that show malignant tumours, and another set of MRIs without malignant tumours, and sure enough the machine learns to distinguish between the two sets (you might have come across claims of “AI can cure cancer”. This is how it does it).

However, AI can sometimes go wrong by learning the wrong things. For example, an algorithm trained to recognise sheep started classifying grass as “sheep” (since most of the positive training samples had sheep in meadows). Another system went crazy in its labelling when an unexpected object (an elephant in a drawing room) was present in the picture.

While machines learn through lots of positive and negative examples, that is not how humans learn, as I’ve been observing as my daughter grows up. When she was very little, we got her a book with one photo each of 100 different animals. And we would sit with her every day pointing at each picture and telling her what each was.

Soon enough, she could recognise cats and dogs and elephants and tigers. All by means of being “trained on” one image of each such animal. Soon enough, she could recognise hitherto unseen pictures of cats and dogs (and elephants and tigers). And then recognise dogs (as dogs) as they passed her on the street. What absolutely astounded me was that she managed to correctly recognise a cartoon cat, when all she had seen thus far were “real cats”.

So where do animals stand, in this spectrum of human to machine learning? Do they recognise from positive examples only (like humans do)? Or do they learn from a combination of positive and negative examples (like machines)? One thing that limits the positive-only learning for animals is the limited range of their communication.

What drives my curiosity is that they get trained for specific things – that you have dogs to identify drugs and dogs to identify explosives. You don’t usually have dogs that can recognise both (specialisation is for insects, as they say – or maybe it’s for all non-human animals).

My suspicion (having never had a pet) is that the way animals learn is closer to how humans learn – based on a large number of positive examples, rather than as the difference between positive and negative examples. Just that the limit of the animal’s communication being limited means that it is hard to train them for more than one thing (or maybe there’s something to do with their mental bandwidth as well. I don’t know).

What do you think? Interestingly enough, there is a recent paper that talks about how many machine learning systems have “animal-like abilities” rather than coming close to human intelligence.

For millions of years, mankind lived, just like the animals.
And then something happened that unleashed the power of our imagination. We learned to talk
– Stephen Hawking, in the opening of a Roger Waters-less Pink Floyd’s Keep Talking

Charades of obscurity

Having “played” dumb-charades (DC for short) competitively at a school and college level, I don’t particularly enjoy playing it casually. I’m prone to getting annoyed when people around me (either on a picnic, or a party) exclaim with great enthusiasm that we should play DC. Till recently I used to think it was like chess – where my enthusiasm for the game has been killed purely because I played it competitively, but now I realize there are more reasons.

The challenge with “competitive” DC is that it is a timed game. You are judged based on how fast you can act out a certain name/place/animal/thing/. Because of this the clues need not be too hard, and there is a fair degree of challenge in acting out even simple things. Apart from this, the clues are set by a neutral third party which means they can all be trusted to be of approximately similar standard, so there is some sort of a level playing field there. Then, you have teams that have practiced well together, and have clues for all the trivial stuff, and you have a game!

With casual DC, there are several problems. Firstly, the games are not timed. Secondly, the teams haven’t practiced together at all, so it takes ages to communicate even straightforward stuff (which is why the games aren’t timed). And then the clues are usually given to you by your competitor. And for some reason, casual DC always has to be movies. No books, no places, no animals, no personalities, nothing.

The f act that the games are not timed, combined with the fact that the clues are given by the competitor, means that the game usually gets into a downward spiral of obscurity. You don’t want your competitor to guess the movie easily, so you give a vague movie. And they reply with something vaguer. And so forth, until teams have to check IMDB to find out if the movies actually exist. By which time all the enthusiasm for the game is lost.

On a recent trip (with colleagues, as part of our CSR initiative. more on that in another post) we played casual DC, and after some 10 clues it had gotten so obscure that nothing was guessable. I’d lost interest when someone suggested we do Kannada movies! Now, that’s something few people would’ve played – DC with Kannada movies as clues, because of which we could give clues while not keeping them too obscure (but it was hard. I completely bulbed trying to act out “Kalasipalya”).

Still, my hatred for casual DC remains, and I try as much as possible to not play it. Maybe next time I’ll impose conditions (like timing, choice of subjects, etc.), and refuse to play if they want to do English movies with infinite time.

Search Phrases – February 2009

I don’t plan to make this a monthly feature, but will write this whenever I find enough funny search phrases to make a post on  them worth it. Googlers and google seem to have had a field day this month,

The top search phrase that has led to my blog is of course “noenthuda“. In second place is the fairly boring “” .  Third place is extremely interesting – top reasons marriage engagements break in pakistan. And I’ve got over 50 people who have searched for this phrase in the last month and then landed up at my blog! Now it makes me wonder what the top reasons are for marriage engagements breaking in pakistan.

Here are a few other gems from the month gone by.

  • gay in iimb (17 hits)
  • 3-letter word for pertinent
  • aunties in chickballapur (chickballapur is my father’s native place, for the record; it is famous for its extremely spicy chillies)
  • best english speaking course in north india
  • can we put the shoes and chappals near the entrance of the house
  • cricketers animal names
  • funny message for my cousin who wants to move back to bangalore
  • i am working in singapore what do i need to do to buy a car in delhi
  • i don’t know how to speak english but i know hindi can i work in delhi
  • iimb course to be on your own
  • job interview edition on
  • karwar muslims
  • matha amritha, things she does
  • number of north indians settled in south india
  • societal influence on a bastard child
  • the true story of a man who learnt fluent spoken english
  • which indian breakfast item can be made with bread?

Ok that has been a very long list indeed. Much longer than I intended it to be. But it only reflects the brilliance of googlers and google in the last one month.

Bloomberg Watching

Two weeks back we were all given dual screens at office. A couple of days after that, those of us that had joined recently got Bloomberg logins. It’s a very restricted version of Bloomberg, with most of the strong features having been disabled. One feature that is enabled, though, is to get the graph of the daily price movement of a security, or an index.

It is necessary to have hobbies at work. It is humanly impossible to concentrate solely on the work for all the eight or ten hours that you spend at office. You need distractions. However, in order to prevent yourself from being too distracted, it helps having one or two very strong distractions. Distractions which can crowd out all other distractions. They can be called “office hobbies”.

In the past, my office hobbies haven’t really been constructive. In my first job, I was part of a PJ Club, and we would exchange horrible jokes. By the time I got to my next job, I had been addicted to Orkut, and kept refreshing it to check if I’d gotten any new scraps. Of course, when there is a cricket match on, the Cricinfo screen makes for a good office hobby. In the last ten days, the World Chess Championship has served my evenings well. However, it is important to have a sustainable hobby which could also be constructive. One which might have a small chance of making impact on your work. And most importantly, it would be ideal if the boss doesn’t really disapprove of your office hobby.

For the last week and half at work, my right screen (remember that I have two screens) has been reserved for Bloomberg Watching. A Bloomberg window is open there in full size, and I would’ve usually put the daily movement graph of the Nifty there. And it updates real-time. It’s like a video game. I just sit and watch. And get fascinated by the kind of twists and turns that the markets take.

Twenty years back, I would spend my evenings in the courtyard of my grandfather’s house in Jayanagar watching ants move about. I would be fascinated by their random, yet orderly movements. I would spend hours together watching them.

Around the same time, I used to play another game. I used to splash water on the (red-oxide coated) walls of my loo, and watch the different streams of water flow down as i crapped. I would get fascinated by the patterns that the water droplets would form, the paths that they would take, the way they would suddenly change speed when they intersected, and so forth. I would end up squatting there long after I’d been done with my crap.

So what I’m doing now is not exactly new. I just watch a point move. Orderly from left to right. Wildly fluctuating in the up-down direction. I look at the patterns and try to guess which animal they look like, or which country they look like. I get fascinated by the sudden twists and turns that the curve takes, and wonder about the collective wisdom of all market participants who are faciliating such movement. I occasionally scream out to my colleagues saying stuff like “nifty below 2600!” and they respond with a “behenchod…” or some equivalent of it.

As the day wears on, I realized that some animals I had recognized earlier in the day are hardly visible now. They are but specks in the larger graph that is the day. And then I realize that unless there was something truly special, the movement of the day will also soon be lost. It will be available for download from the same Bloomberg terminal but that will be about it. And so forth.

Occasionally I catch some unsuspecting soul on my GTalk list and spout such philosophy. I tell them about how after a while everything becomes insignificant. About how we will always be just small players in the larger system. The smarter among them will add their own philosophy to mine, and sometimes we come up with a new theory. The not so smart among them – they will ask me about my views on the market. And what would be good picks (this has been a regular question I’ve been asked ever since I got back into the finance industry but more about that later). And then they say something like how terrorists are the reason the stock markets are plunging, and how the government should protect investors’ money and stuff.

Some day I hope all of this will be useful. Some day I hope my eye for recognizing animals and countries where none exist will enable me to come up with some earthshaking strategy, which can make millions for my fund. However, now that doesn’t matter. All that matters is the unbridled joy of watching the ticker move up and down. Rise and fall. Take baby steps, and the occasional giant leap. It’s surreal.