 recognized as one of India's top 10 data scientists and leads a team of beautiful visual data storytellers. Please welcome Mr. Anand with a huge round of applause. The question I guess is, what's a data scientist doing in a design conference? Well, a bit about myself and Kramanabhi. You've probably seen our work. If you've seen any of the election coverages, then all of the visualizations that you see on TV, those are usually ours. But this isn't really a story about what I do professionally. This is really my personal story. See, the thing is, I just love comics. Have you seen this, Calvin and Hobbes, where Calvin asks, how do bank machines work? And Calvin's dad says, well, let's say you want $25, punch in the amount, and there's this guy behind the machine with the printing press that makes the money and sticks it out from the slot. Calvin says, oh, you mean just like the guy who lives up the garage and opens our door? Yeah, exactly. And I've always wanted to take these and use them in many of my examples. The thing is I'm a business consultant and the bulk of my presentations, I want to communicate a point. And it was just so apt to pick a Dilbert or Calvin and Hobbes and put it in there, except there was only one problem. How do I find it? Dilbert eventually had a search engine. Calvin and Hobbes still does not. So what I did was what, effectively, I consider my greatest achievement in life, which is typed out every single Calvin and Hobbes strip. From 2001 to 2007, on the Mumbai suburban trains, I would make the commute half an hour in the morning, half an hour in the evening, bandra to church gate to bandra, and just typed it out. It is complete. It was live. Comics are serious business, right? Just like humor, entertainment, movies, these are all serious business. It's a pretty serious business. And while it's certainly a passion, it's something that I was also using professionally quite a bit. So as Grammar started, we were founded about eight years ago, we assembled a design team. And this design team over the years, as Grammar grew, got pretty crazy too. One of the things they did was, this is literally a wall in our Hyderabad office. They said, we're gonna take sketches, we're gonna paint notes, create sketch notes. And these characters were named D and D. The Tamilians among you will get the joke. The Non-Tamilians among you probably won't, and I'm not gonna bother explaining it. But those are the characters that we have. And that's when I realized something about what Scott MacLeod meant. See, when you see this face, you see someone. When you see this face, you see yourself. And when that face smiles, you smile. When that face frowns, you frown. And it's an extraordinarily powerful way of transferring emotion from one to the other. And we mimic the emotions that we see. Also, that we can simplify comics using patterns, and this was my big discovery. And I'm gonna tell you about this, that you can actually create comics using a systematic structure. We're still discovering the nuances of the structure, but the core principle of this, we embodied into a library called Comic Gen. Comic Gen is a programmer's library. It allows you to create comics. It's at grammar.com slash comic gen. And the way it works is it has a set of characters. So there's D, for instance, who can have a variety of different angles. So D can be straight or to the side or sitting down. And against each of these angles, there are a series of emotions. She can be angry and frustrated, angry and shouting, annoyed, blushing, crying, sad, shocked, smiling, happy, whatever, against which there can be a series of different poses, hands on her hip, holding a book, holding a laptop, pointing right, pointing up, and so on and of course, there's no pointing right, or there is pointing, okay, there's no pointing left, but you can actually mirror and therefore get the equivalent of a pointing left, which you can expose as SVG or as PNG, and either export it or directly embed it as code. And I'm gonna talk about the code part of it a little later. This was one character that we built. The other character that we built was Day with very similar features. So it has exactly the same set of feature expressions. Now with this, we had a very new and different capability and we started exposing this to people both within the team and outside the team. There are a lot of requests, both from within and from the community. People said, we now need a corporate character. D and Day looked like Ruffians. Okay, fine, we'll get your corporate character. We called him Ringo. He had a French beard. He was slightly better dressed. But then they said, no, now we need colors because it has to fit with our team. So we said, okay, here's Aryan. But for some reason, Aryan didn't really stick. But the real hit character was Zoe. For some reason, people just loved Zoe either because of the choice of colors or just because of the gender, I don't know. But this was a character that really became very popular. And these characters were created, some of these characters were created by the community, not by us. This is an open source library. Anyone can use it, anyone can contribute to it. We are in fact working with news organizations who say, oh, can you do that for us? We said, yeah, sure, just give us all your comics and we'll put it together in a structure where you can start composing characters. The trick here is to recognize, and this is the pattern I was talking about, that any given comic character is composable based on a series of elements. It could be head and body. It could be head, body, feet, which is how there's an existing library called Humans that does that. Or you can have a face in which you can have different emotions and expressions. That's something that we'd been playing with for some time. Some of you may have heard of this experiment called Sharnoff Faces, where you can have the expressions of a face controllable based on certain parameters. So that's a thin face, that's a fat face. More importantly, that's a rich person, that's a poor person. That's a happy person, that's a sad person. That's a Caucasian, that's an African American. You get the factors that can affect it. And all of these are controllable by data, which means that I can have individual elements of a face being driven by attributes that are outside. So we took these principles, but at the moment we are growing the community. And if there are any of you who are artists, who are, of course, sketchers, who just want to play around, even if it's stick characters, do reach out to Shami Contact Details or just search on Comic-Gen. There are detailed instructions on how you go about creating and it's fairly simple, you just have to draw it and send it to us and we'll take care of the rest and grow this library. But the question now comes. Fine, we've got this, what do we do with it? What are people doing with it? Comics are cool, nice, I don't mind using it, but is this useful, right? The where pragmatism starts hitting upon what we like. Well, turns out that the community definitely likes it. They've been using it for a number of things. There's this whole bunch of Japanese and Chinese sites that are using it, but I haven't been able to figure out what it is that they are saying because Google Translate does not work with images. So we're just stuck. But the ones that I'll show you a few that we know. So this is guy called Anil, for instance, who created something, which is a simple link. You always tease me about being on the phone all the time, but I've been off the phone for eight straight hours. What, you've been sleeping? Simple illustration about the Facebook database expose. So if there are 419 million phone numbers out there, why do I bother needing to share my phone number? Just look it up on the Facebook leak, right? Or another one somebody put together on the mosquito, saying I'm feeling invincible because mosquitoes are, in fact, the deadliest animals in the world in terms of the number of deaths that they cause. Now, these are created by people who can't draw, who can't, who don't have access to a design team, but just said, I'm gonna take this and put it together. And this gave us an idea of how it's gonna be used in the wild. Even within Grammar, the way it started getting used was first by the marketing team who said, we want to send out an invite, they said. So there was this event that we were conducting called Data Science in a Box. And the invite for this event was literally a comic-gen strip, a comic strip, based on comic-gen characters. They just went there, copy pasted the relevant portions and put it in. In fact, there was a comic-gen tutorial that we created. And the invite for that was, again, using comic-gen. So here's a comic-gen workshop, and somebody says, wait a second, hold on. How would I go to a comic workshop when I can't draw? So D explains, well, it's not really about teaching you how to draw. You don't need to draw, and that's the whole point. So, okay, in that case, why do I need a workshop? Well, there is something called comic-gen. You don't need to draw. You can actually directly use it. The workshop is about how to use comic-gen. Ah, okay, perfect. So then why don't you call it a how-to-use comic-gen workshop? Okay, fine, change the title. This incidentally reflects the way in which Ramya actually went through this sequence and put her experience out as a tutorial. It started feeding into our projects as well. We were doing a piece of work with the Gates Foundation, working with the UP Ministry of Health on tuberculosis awareness. And one of the comic strips that was created explaining tuberculosis was using comic-gen. Now, the trouble with some of these is that as you start thinking about comics, the brain starts working in weird ways. So we had to struggle very hard, not to say TB or not TB. Catch the drift. Process communication as well. For instance, choice of colors. How do we pick choice of colors? So this was a comic strip that we used internally to say, look, pick any image that you want. Upload it on cooler.adobe.com and whatever is the image. It will generate a palette based on that. And you can update your dashboard using that color scheme. And that's how the teams are picking their color palettes these days for the dashboard. Another one was with a bank where we realized that there was one particular role that they were recruiting for. That took as many as 45 days. Normally, managerial positions just take two to three days to get closed. But this particular role took a solid 45 days, which means that the candidate is going to say, what, are you kidding, one and a half months just to get back to me on what the next steps are? A number of other things we found at the bank. So it turns out that as the women stay longer in the organization, the attrition rate tends to drop. This is not true for men, only for women. And for older women, the attrition rate again is lower. Now these are things that the organization kind of broadly knew. But when put in as a comic, the level of attention started increasing. Of course, that could be a good thing, that could be a bad thing because there was someone said, ha, okay, then we don't have to worry too much about them during the pay rise, right? So it hits both ways because the attrition rate is lower than they say, okay, fine, then why do I have to bother with such a high pay rise for some of these people? Annotating our visuals was another thing. We do a lot of data visualization. So this was sort of a bank where this was the original visual that was put together in Excel showing for each branch what's the performance, which in itself is pretty unreadable. But then when you add a comic to it saying, okay, branch one tops the chart with the maximum business and that's branch one, it's got the maximum business. But branch one also has a higher default payment, default payments as compared to branches four and six. So their net, their NPA bad loans are pretty high too. That's not a great thing. So in that case, which one has the best performance on NPS? Well, it turns out that that happens to be a zero point. Okay, branch four. So branch four tops the chart with overall business and minimum default payments. That's pretty much the best branch you're looking at. Okay, how about return on investment? Branch six has the best return on investment with 21.2. Now I'm showing you a screenshot which is progressively animated but the actual visualization would draw these lines and tell people exactly what was happening as a narrative story. Or the cricket match India versus Afghanistan. One of my colleagues Richie got totally excited by the match and started creating an alternate view of the scorecard. So every single ball is represented by the number of runs that they scored over by over. Each column is one over and here they scored a four, here they got a wicket, here they scored one run. And this is the total number of runs scored across the match but annotated based on what happened at different points in the match with at the end, India managing to get only 225. Well, actually this turns out that this is not very unusual. Most countries score pretty low against Afghanistan, especially when they're batting second. It turns out that when you play against Afghanistan, you don't really need to score much in the second innings. Your match is done by the time you score a reasonable amount but in our case we were playing first so that wasn't necessarily a good sign. As well as annotating live on dashboards. You see the thing is this is programmatic. So when the numbers are increasing, we can have a smiley face and when the numbers are dipping, we can have a sad face and put it across multiple languages. So this was a dashboard where we were showing the performance of health across different regions and Jhansi for instance is one of the few districts whose performance is starting to come down after we started monitoring it. And the wording was to the health officer in Jhansi letting them know that your rank has come down. Now we had to be very careful because people said, whatever you do, whatever criticism you make, make sure that there's also something that you say that is good about them. So the trade off between what to say and what not to say is a pretty tricky one. But let me come back to the point about this being programmatic. See the thing is that because it's a library, it means that as the data changes, you can have the code react to this. So this is a simple piece of code which as time changes, the color updates and as the color updates, the character's emotion changes, which is a pretty simple way of applying it. Now, what are some practical uses of this? Let me show you one. Actually, no, let me show you more than one. This is a strip that we built where people can search for any word. So let me search for let's say epiphany. It behind, wait for it to load, but behind the scenes what it does is looks up the Miriam Webster dictionary and automatically creates a comic strip. It says, do you know the meaning of epiphany? Yeah, sure, I know what it means. And this is the pronunciation of epiphany, which Dee also explains. And Day says, okay, fine, epiphany means twas, blablabla, which if you haven't, if you don't recognize it, that's Jabberwocky. Dee says what, what rubbish? Okay, so what does it really mean? It means a Christian festival held on January 6th in honor of the coming of the Three Kings, Jennifer and Jesus Christ, oh, I see, okay. You've got to learn to use the dictionary day. Why would I need a dictionary with you around? Comic strip created based on any arbitrary word with a few random variations tossed in that creates a whole series of strips. So similarly, for instance, if I searched for, let's see, disaster. Now, one of the other things that this does is looks at the meaning of the word. And based on the meaning of the word, the characters have a different expression. So eventually when Dee comes on this side, yeah. Dee actually has a shocked expression because disaster is not such a pleasant thing and certainly not something to be happy about. Whereas if it were something pleasant, then she would have had a happy expression. And this is automated. An actual dashboard that we put this in was a World Bank report which published how men and women have different laws in different countries. The way in which men and women access institutions, get a job, approach courts, et cetera. These are different. So for example, if you take accessing institutions, there are 19 countries where married women must obey their husbands by law. That includes countries like Saudi Arabia, Brunei, Malaysia, Iran, Jordan. If you look at these 37 countries, a woman cannot apply for a passport in the same way as a man. And in as many as 11 countries, you can't apply for a national ID card the same way as a man. And there are three countries where you can't open a bank account the same way as a man. And just in one country, Equatorial Guinea, you can't even sign a contract the same way as a man. In all of these cases, the expression that Dee shows is entirely based on the data. And you simply say how many countries are in, neutral and how many countries are exceptions. So in this case, he's pretty happy. There's only one country that has this problem. Whereas in other cases, there's a huge amount of a difference. But net net, what has happened as a result of this is that we can create full-fledged dashboards from data. One way is to create plugins. This is a plugin that we created using Power BI where as the sales increases, let's take some other region. So I'm filtering for Americas where the sales has gently increased last quarter. So Dee is pretty happy. But if I go to EMEA, nope, there's a decline. So Dee is, Dee is a bit unhappy. And in APAC, there's an increase. So he's pretty happy. The emotions can get more sophisticated, of course. But what we have now is a system where it's possible to transfer emotion from data to a human through comics without human intervention, at least to a certain level. And we are getting to that stage where you can actually craft some pretty powerful engaging stories with comics without needing to know any kind of drawing. And even more importantly, without even needing to draw them in the first place, they can actually get used automatically. We're still at the nascent stage of things and really love feedback on how this works out for you. If you do try it and please do try it. If nothing else, it's going to be an interesting experiment and absolutely want to hear from you on your success with comics. Thank you very much.