Author: SK

On multitasking, queues and call centres

Queues and call centres, with linear processing, are inefficient as they result in low utilisation. 

I recently read this excellent article by Tim Harford about multitasking.  In this, he talks about research which says that multitasking makes you ineffective because of high cost of context switching, something that I’ve come to learn over the last three years. He also has this nice piece on ADHD here:

“You’re letting more information into your cognitive workspace, and that information can be consciously or unconsciously combined,” says Carson. Two other psychologists, Holly White and Priti Shah, found a similar pattern for people suffering from attention deficit hyperactivity disorder (ADHD).

It would be wrong to romanticise potentially disabling conditions such as ADHD. All these studies were conducted on university students, people who had already demonstrated an ability to function well. But their conditions weren’t necessarily trivial — to participate in the White/Shah experiment, students had to have a clinical diagnosis of ADHD, meaning that their condition was troubling enough to prompt them to seek professional help.

This piece, however, is not about multitasking at the personal level. It is not about ADHD, either. It is about Adigas, and Citibank, and call centres.

As I had mentioned in a blog post yesterday, I visited Vasudev Adigas in Jayanagar 8th block on Sunday, after a really long gap. They have completely revamped and redesigned the restaurant, changing the place of the cash counter, food counters, kitchens and what not. Most of the design is good, and speeds up processes. Except for the cash counter.

A “feature” of cash counters at South Indian fast food restaurants is that there is no queueing. Counters are placed in a way that people can crowd around it from all directions. While this leads to some confusion and encourages bad behaviour, it also ensures that the person at the cash counter is always productive. If one customer is dillydallying about her order, the cashier can simply process another order before the first customer has made up her mind.

The new cash counter here, however, have very restricted access which makes it hard for the guy at the counter to multitask. As a consequence, his utilisation is low (customer take time to make up their minds), and the average wait is longer. And there is no queueing either, so there is no reduction in bad behaviour also.

For a similar reason, call centres are ineffective – they result in low utilisation on the part of both the customer and the call centre “executive”. It is unlikely that you spend all your time talking, and there is significant amount of time wasted in being put on hold or listening to boilerplate messages.

For example, I need to talk to Citibank because they’ve issued me a new debit card but haven’t sent me a PIN. When I call them, I’ll have to enter my account number multiple times, waste time listening to their options read out in a linear fashion and simply wait listening to random music when they inevitably put me on hold. On the “executive”‘s side, there will be time taken to verify stuff – when they are stuck with me while they might be serving another customer instead.

Call centres are a vestige of the 1990s (or earlier outside India), when phones were plentiful and internet not so. A significantly superior mechanism is to replace the call centre with chat – either through a web interface or through an app. Chat allows both the customer and the executive to multitask, and not waste time in meaningless tasks. Authentication can be superior to that on the phone, no time is lost navigating (since nothing needs to be read out linearly), and utilisation of both the customer and the executive is really high.

Yet, Citibank doesn’t offer this option. Neither do a lot of other supposedly progressive organisations. And that is disrespect to the time of both their customers and their “executives”. Hopefully, they’ll offer a chat option soon.

As for Adigas, the redesign has been after their new PE money came in. I’m less bullish about their changing their billing counter design.

Characterising network effects

Met a bunch of people for drinks this evening. Most of the conversation was just okay. But there was this little bit about network effects. Where I figured out how to calculate whether network effects are present in an industry. It all came out of Kingsley claiming that the age of network effects is over, and there are no more network effects left.

The discussion presently moved to how you discover whether there exist network effects in different industries. Does the fact that  Amazon’s marketshare is nowhere close to that of a monopoly mean that there are no network effects in e-commerce marketplaces? Doesn’t Google have network effects in that given the larger number of people searching on the platform, there are more clicks and more opportunity for learning (for Google) and hence better results?

At a point of time in the conversation, I made the statement that Google (in particular and search engines in general) has “partial network effects”, in that more users means more learning and hence more results. And that for this reason Bing or any other competitor can’t match up.

So how can we characterise whether an industry has network effects, and if so, to what degree? Thinking about it, it’s rather simple. In a “normal” non-networked industry, the value of the user base is directly proportional to the number of users. Going by Metcalfe’s Law, in a fully networked industry, the value of the user base is directly proportional to the square of the number of users. An industry with “partial network effects” should surely have its value a power between 1 and 2 of the number of users?

Here’s how we figure out how networked an industry is. Take all the players in the industry and tabulate the size of the user base and the value of each of the players (excluding very small players). Plot them on a log-log plot, and measure the slope. If the slope of this log-log plot is close to 1, it means that the industry is not networked at all. If the slope is close to 2, it means it has “full network effects”. And the numbers in between represent the spectrum of possible values.

Rather simple, isn’t it? This is why I love drinking sessions, for they allow you to unleash such thoughts. Oh, and I “recorded” this thought by sending a WhatsApp voice message with the gist of the above content to Hariba. He replied with “keep them coming” or some such thing, but this was all it was for this evening.

Coffee, sugar and cream

A couple of weeks back, my wife and I had a long discussion on the operations of the coffee counter at Maiya’s in Jayanagar. It was an interesting discussion since while I was extremely familiar with the operations there (having gone there almost every other day for the last year), the wife was seeing them for the first time.

My hypothesis was that it was the structure of the coffee+tiffin combo and not accounting for multiple orders in one ticket that caused the congestion. The wife’s diagnosis was rather different – she recognised the sugar counter as the bottleneck.

Most South Indian restaurants have ready two kinds of boiling milk – one with sugar and one without, and your choice of milk (or a linear combination) can be added to decoction to make coffee of the required sweetness for you. Maiya’s does it differently. They only have unsweetened milk, and you need to add the sugar yourself.

So there is sugar placed in a bowl beyond the coffee counter where you add the sugar, get yourself a spoon (inconveniently placed before the coffee counter which means you stretch across) and go on while stirring the coffee. For non-regular customers (my untested hypothesis is that most Maiya’s customers are regulars), this is a novelty and leads to inefficiency of the full queue.

The wife argued that if Maiya’s were to keep both sweetened and unsweetened milk (like other restaurants), sugar could come pre-mixed in the coffee and the bottleneck could be eliminated. Since the barista doesn’t multitask (he fills exactly one cup at a time), there is no problem in miscommunication, etc.

The problem is that turnover of the unsweetened milk in other establishments is not high enough to maintain quality. The thing with the milk is that it needs to be constantly stirred, or at least poured from, for cream to not form in it (such cream can make the coffee gross). When the demand for a particular kind of milk (usually unsweetened) is low, it is not stirred enough, and cream forms. And then when you ask for coffee without sugar (or “less sugar” – remember linear combinations of the milks are possible) you end up with cream in your coffee.

This happened to me twice in the last three days. On Saturday I was having coffee at Hatti (opposite Maiya’s), asked for “strong, less sugar”, which meant I got some of the unsweetened milk, which means there was cream in my coffee. I had to spit out some to make it palatable. And the story repeated itself at the Vasudev Adigas in Jayanagar 8th Block on Sunday. Nice tasting coffee made gross by the cream.

It is to solve this problem that Maiya’s perhaps has only one kind of milk – it is constantly boiling away and being poured from, and there is no cream. And you get superior quality coffee. For which I’m willing to pay a premium.

Optimal risk sharing

The wife moved to Ann Arbor over the weekend, where she will be spending three months. She took an Air France flight (AF191) in the wee hours of Sunday morning, and then switched to a Delta flight at the legendary Charles de Gaulle. I must mention upfront that she seems to have had a peaceful journey.

Except that people following the same schedule exactly twenty four hours earlier would not have. AF191 that departed from Bangalore i n the wee hours of Saturday morning returned to Bangalore after a bomb scare. The flight was subsequently cancelled.

There are many risks to flying. Schedules nowadays are packed so closely that your flight might be delayed. Occasionally it might be cancelled even, sometimes without a good reason. A delay might sometimes mean that you miss your connecting flight.

The question is who bears the risk on this one. If I’m booked on a flight that gets cancelled or delayed (because of which I miss my connection), whose responsibility is it that I’m transported to my destination? There are three possibilities – the passenger himself, the airline and an external insurer. The question is which of these is most optimal.

The traditional model in aviation as I understand it is that it is the airline’s responsibility. While this makes sense because a large number of delays/cancellations are on account of faults on account of the airline, even when the delay is not due to the airline’s fault, the airline is best placed in terms of mitigating the risk.

Leaving the risk on the passenger has the advantage that he can choose his own risk profile. If you are flexible about your trip, you might choose to go without insurance, and take the hit yourself. If you’re a frequent flyer, then the “insurance cost” thus saved will compensate for the occasional delay. Yet, the problem with this kind of a model is that people tend to underestimate the risks, and will more often than not not insure, and get hit badly when the delay happens.

Which brings us to the final absorber of risk – the insurance company. I’d purchased “travel insurance” for a recent trip, and there was a component on account of delayed or missed flights. If my flight was delayed by a certain amount of time, my insurer would pay me a fixed amount of money.

While this financial hedging is good, it may not adequately represent the costs of making a new booking (including the hassles) when my flight is delayed or cancelled. So this is not a workable solution at scale.

Another solution is for the insurer to guarantee that you will reach your destination by a certain time in case your flight gets delayed or cancelled. This might work out to be more expensive than a fixed cash payout but this removes the cost and hassle of figuring out the next best alternative on the part of the customer. The problem, however, is correlation. Insurance works when people’s risks are uncorrelated or negatively correlated. Here they are positively correlated – all passengers on Saturday’s AF191 to Paris were affected similarly, and this pushes up the cost for the insurer to rebook people.

Unless they tie up with the airline itself! If they reach an agreement with the airline such that the airline commits to transport the stranded passengers, then this “positive correlation” I mentioned earlier will be taken care of. Seems workable, right? Except that what is being insured here is the risk that the airline abandoned in favour of the passenger, who insured against it from an insurer, who reinsured it with the airliner! Can we just cut out the middle men?

From this rather unscientific argument above, it looks like airlines are best placed to insure passengers against disrupted flight schedules. Back in the days of regulated air fares where competition had to be “on service”, airlines would take responsibility. This might have disappeared with the move towards unbundling over the last 2-3 decades. For good reason – insuring a schedule results in an additional (albeit hidden) cost, and getting rid of it can result in cheaper (base) fares.

Yet, given that airlines are best placed to insure schedules, we need a solution. Maybe they can charge a premium for insuring schedules apart from the base fares? Or would they argue that the current “unrestricted fares” are such insured fares (implying the premium is rather high)?

Short of  government mandated regulation, what is the best way for allocating the risk of disrupted flight schedules, and pricing it appropriately?

Tailpiece: A decade ago, our valuation professor (at IIM Bangalore) had told us that “risk cannot be eliminated. It can only be mitigaged by selling it to someone who can handle it better”.

Getting BRT to work

Dedicated bus lanes are neither a necessary nor a sufficient condition for BRT

After significant success in Ahmedabad and spectacular failure in Delhi, Pune is the latest city in India to embark on a “Bus Rapid Transport” (BRT) project. As the name suggests, the point of a BRT is to provide fast and convenient transport to people on buses that ply on existing roads, with some sections of some roads being reserved for buses.

However, in popular imagination, BRT has become synonymous with bus lanes (a lane of road reserved for buses), and the whole controversy in Delhi (which caused the project to be shelved) was about a lane of an arterial road being reserved for buses. In fact, however, a dedicated bus lane is neither a necessary nor a sufficient condition for implementation of BRT.

The attraction of BRT is that it comes with low infrastructure cost – unlike a train or monorail (or even a tram) line, there is not much investment required in terms of physical infrastructure. The challenge with BRT, however, is that its buses are liable to get stuck in traffic (just like every other vehicle) which might prevent it from living up to its middle name.

For this reason, certain changes are made to traffic patterns so that BRT indeed remains rapid. For example, traffic signals on arterial bus routes might be designed to give priority to the directions where buses travel. You might have bus stops in the middle of the road for people to get on to buses. And you might reserve lanes on roads for buses. Once again note that the last named is not a necessary condition for BRT.

What BRT should deliver is a dense and reliable network of buses. On arterial and other key roads, frequency of buses should be extremely high. Our current model of point-to-point and hub-and-spoke based bus routes need to be given up in favour of a more dense network, where it might be quicker for people to change multiple buses to get to their destination. This also warrants a change in the ticketing system, using a zone-based ticket than the current point-to-point ticket, and moving ticketing offline.

 

The fashion so far in India (with Ahmedabad being a possible exception) is to announce arterial roads as “BRT corridors” and start off the BRT services by reserving lanes on these roads for buses, without bothering about linkages and networks at either end. The problem with this is that the losers of the road space “pay” immediately, but the benefits of BRT are not immediately forthcoming.

A better method of implementation would be to make reservation of bus lanes the last step in BRT. The first should be to increase the density of buses and creation of networks. The problem with this is that it requires investment and the expanded (and densified) network might run far below capacity for a while. Yet, as the network expands (even without dedicated lanes), people will begin to see the benefits and convenience offered, and demand for BRT will increase.

Two things will happen – firstly, the expanded and densified network of buses will start crowding out (literally) private vehicles on the road. Secondly, people will see the relative benefits of taking these buses and these buses will start filling up. As these two effects take place, there will come a point when lanes can be reserved for buses without slowing down any of the rest of the traffic.

What we need, in other words, is “system thinking“, and to look at BRT as a solution to move people to their destination in a more efficient manner. Once policymakers recognise that bus lanes are only a means to this end, we can expect BRT to implemented in a proper fashion.

Tasting Gods’ food

The norm during festivals and other “happy occasions” when food is “offered to the Gods” is that the food is not tasted during preparation. For tasting thus would contaminate it, and make it impure for the God. Thus, the first time a human will taste such food is when it is offered as “God’s offering” after the rituals are over.

While there might be good reason for doing so (food thus prepared is distributed to a lot of people and you don’t want to contaminate it and so on), the problem is that if the food is not accurately prepared, it cannot be corrected. By the time someone figures out something is not right, “the God would have tasted it”, and if the food hasn’t been accurately prepared, you would have ended up serving the Gods bad food! Which can only bring ruin upon you.

Let me draw an analogy. Instead of food, let’s assume that you’re offering God a computer program that you’ve prepared. You’ve got the best team of programmers in the world and written a kickass algorithm and got these programmers to code it, and you offer the program to God. And what happens when he tries to “consume” it by running it? Most likely, a stack overflow or some such error.

Would you let that happen? Even when you’ve got a kickass algorithm and a kickass team of coders to code it in, it’s not guaranteed that the code will perform as it should on its first running. Irrespective of how good the code is, it needs to be tested, to make sure it is doing what it’s doing before the user sees it. Especially if it’s an “all-important” user such as a God.

If you were to do that for code, why should food be different? Why would you want to “cook blindfold” by not ta(/e)sting it adequately, and making sure that it’s as perfect as you want it to be? After all, you’re offering it to a God!

Bah, these silly rituals!

Analysing the BBMP Elections

The Bruhat Bengaluru Mahanagara Palike (BBMP) went to polls on Saturday, and votes were counted today. The BJP has retained its majority in the house, winning 100 out of 198 available seats. While this is a downer compared to the 113 seats they had won in the previous elections in 2010, the fact that they have won despite being in opposition in the state is a significant achievement.

Based on data put out by Citizen Matters, here is some rudimentary analysis. The first is a choropleth of where each party has performed. Note that this is likely to be misleading since constituencies with large areas are over-represented, but this can give you a good picture.

blrelec
Red: BJP Green: Congress Cyan: JD(S) Blue: Independents Black: Others

Notice that there are a few “bands” where the BJP has performed really well. There is the south-western part of the city that it has literally swept, and it has done well in the south-eastern and northern suburbs, too. And the party hasn’t done too well in the north-west, the traditional “cantonment” area.

We can get a better picture of this by looking at the choropleths by assembly constituency. These shapes might be familiar to regular readers of the blog since I’d done one post on gerrymandering.

blrelec2

This tells you where each party has done well. As was evident from the first figure, the BJP has done rather well in Basavanagudi, Jayanagar and Padmanabhanagar in “traditional South Bangalore” and blanked out in Pulakeshi Nagar in the cantonment. In fact, if you try to correlate these results with that of the last Assembly elections, the correlation is rather strong. Most constituencies have gone the way of the assembly segments they are part of.

Then there is the issue of reservation – there were a lot of murmurs that the Congress party which is in power in the state changed the reservations to suit itself. Yet, there are a few interesting factoids that indicate that the new set of reservations were rather logical.

The next two graphs show the distribution of SC and ST populations respectively as a function of the reservations of the constituencies (population data from http://openbangalore.org).

Rplot03 Rplot02Notice that the constituencies reserved for SCs and STs are among those that have the highest SC and ST populations respectively. The trick in gerrymandering was in terms of distributions between general and OBC constituencies, and among women.

I could put the performance of different parties by reservation categories, and on whether the reservation of a constituency has changed has any effect on results, but (un)fortunately, there aren’t any trends, and consequently there is little information content. Hence I won’t bother putting them in.

Nevertheless, these have been extremely interesting elections. All the postponement, all the drama and court case, and finally the underdog (based on previous trends) winning. Yet, given the structure of the corporation, there is little hope that much good will come of the city in the coming years. And there is nothing in the election results that can alter this.

 

Uppi2 Review

Uppi2 is easily the best movie I’ve watched in recent times. Across languages. And I’m not joking. It’s a bloody good movie, and breaks all kind of stereotypes. Having watched the movie, I have half a mind to log on to IMDB and add one more perfect 10 rating.

I have an unusual way of rating movies – essentially, any movie that manages to hold my attention through its length is a great movie by my definition. This is a consequence of my extreme attention deficit, but in general I find it hard to sit through movies. The story needs to be tight right from the beginning, else I’m extremely likely to switch off. The number of half-watched movies on my Tata Sky recording box is not funny, for example.

From this perspective, the best compliment that I can pay to Uppi2 is that never once did I start wondering when the movie would end. There have been several otherwise great movies which have dragged a bit as it has gone along and I’ve found myself checking my watch to see how much longer it might go on for. Not with Uppi2. The movie keeps you fully engaged right till the end and doesn’t drag one bit. And I wouldn’t write more here since that would be giving too much away.

There is only one jarring thing about the movie, and that is the songs. All songs are little better than “extra fittings” and none adds to the story. Most songs are fairly atrocious, but there is one truly outstanding song. And you might be surprised that I’m actually recommending a hip-hop song. Enjoy off:

The story, as you might expect from a Upendra directed movie, is rather complex and has lots of twists and turns. There is a fair bit of self-reference, and you might do well to keep in mind the concept of recursion while you watch it. There are references to several other movies, including Upendra’s earlier movies.

If you think you’re a “buddhivanta” (intellectual) this movie will surely tingle your intellectual nerves. There are a lot of stupid jokes also, like the play on the lead character’s name “Neenu” (“you”), and the pun in the title of the movie itself. And stuff like one guy scolding another guy (whose name is Bala) as “LK Bala”. So even if you aren’t a buddhivanta, there is much for you in the movie.

As this excellent review by Jogi in Udayavani describes it, go expecting to see an Upendra movie, or to eat “Uppit”, and you’ll not come back disappointed. If you go expecting to eat “Obbatt”, on the other hand, you’ll surely be disappointed!

I thought Upendra had done a wonderful job with his last offering Super, but I must confess he has outdone himself with Uppi2. Please go and watch! And contribute to the movie’s rating on IMDB.

On Sub-nationalism

Pramit Bhattacharya has a nice piece in MintOnSunday about the positives of sub-nationalism, which fosters provision of public and common goods. He cites academic research to contrast Kerala and Uttar Pradesh, which were similar in the mid-19th century, but now have significantly differing levels of public goods.

The research Bhattacharya cites argues that the linguistic sub-nationalism that was formed in Kerala in the mid 19th century was responsible for the state’s high levels of public goods and development. The absence of such sub-nationalism has resulted in weak institutions and weak development in UP, he says.

He ends the piece saying that sub-nationalism is not always a good thing and can lead to secessionist tendencies. He cites the example of Assam, where sub-nationalism has actually hampered development rather than fostering it.

This discussion reminds me about last year’s “unofficial” referendum in Catalunya about whether to secede from Spain. The vote was unofficial since the Spanish Parliament didn’t authorise it, but there were strong signs of Catalan nationalism when I visited Barcelona last October. The Yellow, Red and Blue flag of Catalan nationalism hung from several windows. There was a clock in one of the main squares counting down to the referendum (which finally didn’t matter).

And while there were several emotional reasons for the demand for secessions, including repression at the hands of the “Castilians”, one of the main reasons was economic – the share of national spending on Catalunya was far less than the proportion of Catalunya’s contribution to the Spanish National Budget. The feeling of “why should we subsidise the rest of the country?” was rampant.

This little story illustrates both the positive and negative aspects of sub-nationalism. The negative is easy to see from the above – strong sub-nationalism leads to a strong “us and them” sentiment towards the rest of the country, and the region begins to resent the rest of the country, especially if the latter gets a larger share of the national pie. And this can lead to secessionist tendencies as is evident in Catalunya.

The positive thing about sub-nationalism, on the other hand, is that it subsumes groupism at smaller levels. A strong sub-nationalist feeling means that people think of themselves as members of that sub-nationalist group, and solidarity to any “lower level” groups weakens.

The problem with high solidarity among small groups is that it may lead to provision of private goods at the expense of public goods. When a place is strongly divided by caste, for example, each caste group wants to maximise the interest of the particular caste, and thus invests in a way that the caste gets a bigger share of the seemingly fixed pie.

When the solidarity is at the level of a state or region, on the other hand, the best way to develop the region or state is to provide for public goods or welfare schemes that span the entire state or region, and this leads to an expansion of the pie and the overall development of the region. In other words, when the “us” is a largish geographical area, it is more likely that investments happen in terms of public goods for the area rather than private goods.

Coming back to the example of Kerala, the strong Malayali subnationalism of the mid 19th century had the effect of pushing down casteism. Consequently, the groupism happened at a level (“Malayalis”) that was larger and more diverse than the caste-level groupism that happened elsewhere (like in UP) where there was no strong sub-nationlist movement. The lack of sub-nationalism in a place like UP has meant that casteist divisions in the region have remained strong, and solidarity at that level doesn’t lead to public goods or development.

Think of the nation as a hierarchy, of sub-nations and sub-sub-nations and so forth. And each person’s loyalty is divided in different extents up and down the person’s “chain”. And among these different layers, it is a zero sum game. Thus, strong loyalties at a particular level are resented both by levels higher and lower, and justifiably so. But the higher the level at which the loyalty remains, the better it is for the provision of public goods and development. Chew on it.

Tinder and Arranged Scissors

For those of you who have been following my blog for a while know, I was in the arranged marriage market for a brief period in 2009, before Priyanka magically materialised (from the comments section of this blog) and bailed me out. I may not have covered this in any of the Arranged Scissors posts that I wrote back then (ok I alluded to this but not really), but I had what I can now call a “Tinder moment” during the course of my time in the market.

So on this fine day in Bangalore, I was taken to this Marriage Exchange called Aseema. The name of the exchange is quite apt, since based on two data points (my own and one acquaintance’s), if you go there your search for a spouse is literally endless.

My uncle, who took me there and who was acting as my broker-dealer for that brief period, told me that they literally had binders full of women (note that this was three years before Romney), and that I could search leisurely if I accompanied him there on Saturday morning.

My uncle didn’t lie. This place did have several binders full of women (and men – I too ended up in one such binder after I signed up) and four binders that said “Smartha (my subcaste) Girls” were pulled out and handed over to me. My uncle probably expected me to spend a few hours ruminating through the binders and coming up with a shortlist.

It was nothing like it. Each profile in the binder followed a standard format. There was this 4 by 6 full-length photo. You knew where to look for educational qualifications. And professional summary. It was like LinkedIn meets Facebook profile picture. And that was it.

I remember having some criteria, which I don’t remember now. But once I had gone through the first few pages, it became mechanical. I knew exactly where to look in a particular profile page. And quickly come to a judgment if I should express interest.

Thinking back, I might have just been swiping (mostly left – I came up with a grand shortlist of one after the exercise) on Tinder. The amount of time I spent on each profile wasn’t much more than what the average user spends on Tinder. Except that rather than looking only at the photo, I was also looking at a few profile parameters (though of course whether I would want to sleep with her was one of the axes on which I evaluated the profiles). But it was just the same – leafing through a large number of profiles in a short amount of time and either swiping left or right instantly. Talking to a few other friends (some of it at the now legendary Benjarong conference) about this, my experience seemed representative (note that I’m still in anecdata territory).

Maybe there is a lesson in this for all those people who are designing apps for arranged marriage (including the venerable Shaadi.com and BharatMatrimony.com). That even though the stated intent is a long-term relationship, the initial process through which people shortlist is no different from what people follow on Tinder. Maybe there surely is a market for a Tinder-like arranged marriage application!