WEBVTT - Smart Talks with IBM: Tracking COVID-19 0:00:00.120 --> 0:00:03.279 In smart Talks, we chat with people who are making 0:00:03.400 --> 0:00:06.880 innovative use of advanced technologies designed by IBM and an 0:00:06.880 --> 0:00:11.479 effort to make real world change. There's really no way 0:00:11.760 --> 0:00:15.880 to introduce this episode without stating the obvious. The coronavirus 0:00:16.000 --> 0:00:20.439 and COVID nineteen have caused massive disruptions and pretty much 0:00:20.560 --> 0:00:24.840 every aspect of our lives. We're faced with tough decisions, 0:00:24.920 --> 0:00:28.479 which become even more difficult when we find ourselves lacking 0:00:28.560 --> 0:00:32.440 critical information. And that's going to bring us into today's topic. 0:00:32.600 --> 0:00:35.320 But that's just one way that IBM is making its 0:00:35.360 --> 0:00:39.280 technology available in the fight against COVID nineteen. The company 0:00:39.320 --> 0:00:43.800 is working closely with scientists, doctors, leaders, and experts to 0:00:43.840 --> 0:00:47.239 fight COVID nineteen in many ways, all while leveraging some 0:00:47.680 --> 0:00:52.400 really impressive technology. Whether it's using supercomputers to help researchers 0:00:52.440 --> 0:00:55.080 find a vaccine and that's a little bit of a 0:00:55.080 --> 0:00:59.160 teaser for an upcoming episode, or aggregating enormous amounts of 0:00:59.200 --> 0:01:02.200 informations so that the average American can get a local 0:01:02.200 --> 0:01:06.120 ice snapshot of what's happening in their communities. IBM technology 0:01:06.200 --> 0:01:09.080 is playing a big part. So what are we talking 0:01:09.120 --> 0:01:12.959 about today? Well, the Internet is a phenomenal way to 0:01:12.959 --> 0:01:16.840 share information, but that's a double edged sword. Over the 0:01:16.880 --> 0:01:21.880 past several years, we've seen misinformation and even disinformation spread 0:01:21.920 --> 0:01:25.920 across online communities, clouding our understanding of matters from the 0:01:25.920 --> 0:01:29.800 trivial to the critical. If we're lucky, it's the trivial, 0:01:30.160 --> 0:01:32.959 like the fact that people kept photoshopping the date on 0:01:33.000 --> 0:01:35.160 the time circuit in the Back to the Future films 0:01:35.160 --> 0:01:38.800 to make it today's date when Marty McFly goes to 0:01:38.880 --> 0:01:44.319 the future. But we're increasingly seeing more bad information about 0:01:44.440 --> 0:01:49.000 important things, from politics, to climate change to yeah, the 0:01:49.080 --> 0:01:52.600 spread of the coronavirus, and in that kind of environment, 0:01:53.120 --> 0:01:57.640 making the right decisions becomes increasingly more difficult to do, 0:01:58.120 --> 0:02:02.400 just as it's becoming more urgent. That brings us to 0:02:02.400 --> 0:02:05.840 today's topic. I sat down with Cameron Clayton, the general 0:02:05.880 --> 0:02:09.200 manager of IBM Cloud Ecosystem and of the Weather Company 0:02:09.240 --> 0:02:13.600 and IBM Business Well we were both sitting down, but 0:02:13.760 --> 0:02:16.800 we also happened to be in our respective homes speaking 0:02:16.840 --> 0:02:19.520 over the Internet in an effort to keep ourselves and 0:02:19.639 --> 0:02:23.560 others safe and really, in many ways, that's what this 0:02:23.639 --> 0:02:27.000 boils down to. But I'm getting ahead of myself. One 0:02:27.040 --> 0:02:29.520 thing I do want to mention is that the unusual 0:02:29.600 --> 0:02:33.359 circumstances mean this episode sounds a bit different from other 0:02:33.440 --> 0:02:37.040 tech stuff episodes, because real life goes on while we podcast. 0:02:37.760 --> 0:02:42.120 Cameron's team at the Weather Company have done something extraordinary. 0:02:42.240 --> 0:02:44.800 If you've ever visited weather dot com or use the 0:02:44.840 --> 0:02:47.959 Weather Channel app, you know you can get an incredibly 0:02:48.040 --> 0:02:51.480 localized report down to the zip code in the United States. 0:02:52.000 --> 0:02:55.400 And just so you know, there are forty one thousands 0:02:55.400 --> 0:03:00.120 seven to zip codes in the US. I counted them, 0:03:00.160 --> 0:03:02.680 which was tough because somewhere in the fourteen thousand range 0:03:02.680 --> 0:03:06.120 of lost count had to start all over again. Aggregating 0:03:06.160 --> 0:03:09.440 that much information and presenting it in a meaningful way 0:03:09.840 --> 0:03:13.079 is no small feat. But what Cameron's team did next 0:03:13.480 --> 0:03:16.320 was in some ways even more astounding. They took that 0:03:16.440 --> 0:03:19.400 general approach and they applied it to the spread of 0:03:19.520 --> 0:03:24.080 COVID nineteen. Here's my conversation with Cameron Clayton, with only 0:03:24.120 --> 0:03:28.000 an interruption here and there to clarify some things. Cameron, 0:03:28.040 --> 0:03:31.160 thank you so much for taking time to join us 0:03:31.160 --> 0:03:34.640 on the show today. I really appreciate it, absolutely my pleasure. 0:03:34.639 --> 0:03:37.840 It's great to be here now. I think it's pretty 0:03:37.840 --> 0:03:41.440 safe to say this is not an exaggeration that everyone 0:03:41.920 --> 0:03:46.320 has been affected by the COVID nineteen crisis to some extent. 0:03:46.440 --> 0:03:48.840 But I think it can be a little tricky for 0:03:48.880 --> 0:03:53.000 people to get a big picture grasp on the global 0:03:53.120 --> 0:03:56.200 impact of this crisis. Can you kind of speak a 0:03:56.240 --> 0:03:59.480 little bit as to your perspective on the world impact 0:03:59.480 --> 0:04:04.920 of COVID eighteen. Certainly so. As the Weather Channel, our 0:04:05.000 --> 0:04:09.880 mission is to map the atmosphere. So as a company 0:04:11.200 --> 0:04:14.680 and a collective of people, our whole job is to 0:04:14.720 --> 0:04:18.400 try and predict what's going to happen in the atmosphere tomorrow. 0:04:19.320 --> 0:04:26.120 It's almost this impossible science math mother nature problem that 0:04:26.120 --> 0:04:28.159 that we're that we've been working on for a long 0:04:28.240 --> 0:04:32.760 time and make tremendous progress on. Said differently, though, and 0:04:32.800 --> 0:04:37.480 I think this is really where it comes to COVID nineteen. 0:04:38.320 --> 0:04:41.680 At the end of the day, we're making the invisible 0:04:41.880 --> 0:04:46.760 visible and and so doing that, it's easier to make 0:04:46.800 --> 0:04:54.360 decisions When you take something that's intangible, you can't see tangible, 0:04:54.640 --> 0:04:56.960 and then therefore you can make a bit of decision 0:04:57.000 --> 0:04:59.280 as a result of that. And so that's what we 0:04:59.400 --> 0:05:02.159 tried to do with with COVID nineteen. And we've got 0:05:02.200 --> 0:05:05.640 so much inbound interests from our our fans all around 0:05:05.680 --> 0:05:09.679 the world saying, hey, you you make something we can't 0:05:09.760 --> 0:05:14.839 see easier to digest, easier to understand, and easier to 0:05:14.880 --> 0:05:17.360 make decisions on which you please do the same thing 0:05:17.440 --> 0:05:20.440 for COVID nineteen. And so that's what that's what we've 0:05:20.480 --> 0:05:24.760 been doing, is is trying to make the invisible visible. 0:05:25.680 --> 0:05:29.120 And like like you said, I think every single person 0:05:29.160 --> 0:05:33.560 on the planet is impacted by COVID nineteen, just like 0:05:33.640 --> 0:05:38.479 they're impacted by the weather, right. And so as a 0:05:38.520 --> 0:05:43.679 result of our reach in our scale with literally hundreds 0:05:43.680 --> 0:05:47.240 of millions of users around the world, uh, we're able 0:05:47.279 --> 0:05:50.599 to also communicate with them in a way that they're 0:05:50.640 --> 0:05:54.240 familiar with as part of their daily habit already. H 0:05:54.480 --> 0:05:56.719 And so you know, we'll be able to try and 0:05:56.720 --> 0:06:01.640 provide trusted data to our fans around the world. You know, 0:06:01.680 --> 0:06:07.520 as you point out, Cameron, contextualizing data is really critically 0:06:07.520 --> 0:06:10.240 important for people to be able to make use of it, right, 0:06:10.320 --> 0:06:15.359 to be able to take something that is conceptually this 0:06:15.560 --> 0:06:18.200 huge thing, but it's very hard for us to boil 0:06:18.279 --> 0:06:22.279 that down into actionable things that we can do as people. 0:06:23.120 --> 0:06:26.120 I think one thing that kind of helps again, you know, 0:06:26.279 --> 0:06:28.400 as humans were not really good at dealing with big 0:06:28.480 --> 0:06:32.680 numbers just on our own. One thing that really helps 0:06:32.880 --> 0:06:36.440 is is kind of anchoring things to personal experience as well. 0:06:36.520 --> 0:06:42.240 So before we jump into all the background on IBM 0:06:42.279 --> 0:06:47.080 and the weather companies work with UH, the technologies to 0:06:47.200 --> 0:06:50.240 track COVID nineteen, I was wondering, can you talk a 0:06:50.279 --> 0:06:53.840 little bit about how this crisis has affected you personally 0:06:53.960 --> 0:06:57.479 so far? Oh? Yeah, sure, I think. Uh So, I 0:06:57.480 --> 0:07:01.400 have four children, I'm married with with all kids, UH 0:07:01.440 --> 0:07:05.840 and a dog, and about two and a half weeks ago, 0:07:06.360 --> 0:07:10.640 all our kids, you know, came home and have been 0:07:10.880 --> 0:07:16.560 on sort of Zoom and Google classroom, you know, with 0:07:16.960 --> 0:07:20.520 x is all all day every day, along with with 0:07:20.560 --> 0:07:25.080 me doing the same thing on UH working here from 0:07:25.080 --> 0:07:31.040 the house. And you know, I think we're thankfully safe 0:07:31.080 --> 0:07:35.040 and healthy and doing well. But Kevin fever is a 0:07:35.080 --> 0:07:39.680 real issue, especially with two boys. It's been a big 0:07:39.760 --> 0:07:43.080 change in the sense of we often ask, you know, 0:07:43.440 --> 0:07:45.120 how's your day going. Are you having a good day? 0:07:46.560 --> 0:07:49.200 And I think many of us are still asking that question, 0:07:50.560 --> 0:07:53.560 but we're not used to the answer being actually no, 0:07:53.720 --> 0:07:58.480 I'm not. I'm really not okay. Uh, And and we're 0:07:58.520 --> 0:08:00.480 getting a few of those answers now. I had a 0:08:00.480 --> 0:08:04.760 couple of those onswers this morning. And so how we 0:08:04.840 --> 0:08:08.760 rallied together as a community, how we relied together ash 0:08:09.240 --> 0:08:15.000 as companies, how we relied together as humans really really matters. 0:08:15.280 --> 0:08:18.560 And I'm certainly super proud of the way our teams 0:08:18.640 --> 0:08:22.800 rallied together and and IBMS relied together. Were also super 0:08:22.840 --> 0:08:27.120 proud about our clients and partners and neighbors and uh 0:08:27.160 --> 0:08:34.040 and others like Uh. We had a block street party 0:08:34.040 --> 0:08:38.160 where everybody was in their cause. They decorated their cause 0:08:38.240 --> 0:08:41.000 and drove by the five old people that live in 0:08:41.000 --> 0:08:43.800 the street and who are in the window to try 0:08:43.840 --> 0:08:46.960 and share them up. So this just amazing touches of 0:08:47.040 --> 0:08:49.960 humanity happening. That's that's really cool to see. And that's 0:08:49.960 --> 0:08:54.480 a great way to segue into talking about what the 0:08:54.520 --> 0:08:57.800 Weather Company, what IBM are doing and in a way 0:08:57.840 --> 0:09:02.360 to give people more tool so that they can make 0:09:02.480 --> 0:09:06.520 decisions that are critically important for themselves and for the 0:09:06.559 --> 0:09:10.760 people around them, whether it's relatives, coworkers, loved ones, just 0:09:10.840 --> 0:09:14.760 strangers on the street. We all have this responsibility So 0:09:14.960 --> 0:09:18.840 let's talk kind of in general terms what exactly you 0:09:18.880 --> 0:09:23.319 guys are doing in an effort to give people these tools. 0:09:23.320 --> 0:09:26.120 So kind of from a very high level, as I 0:09:26.200 --> 0:09:32.199 understand it, you're pulling data that's localized to specific regions, 0:09:32.920 --> 0:09:36.240 contextualizing that and presenting it in a way that's easily 0:09:36.280 --> 0:09:40.679 digestible so that people have and up to date understanding 0:09:40.800 --> 0:09:43.720 what's going on in their communities. Is that more or 0:09:43.760 --> 0:09:49.520 less correct? Yeah, that's that's right. We're we're trying to 0:09:49.559 --> 0:09:58.880 make this invisible virus contextual and localized. The important thing 0:09:58.880 --> 0:10:00.959 when you do that is it has to be from 0:10:00.960 --> 0:10:04.200 a trusted source. So the sources of the data have 0:10:04.440 --> 0:10:12.280 to be really high integrity, and integrity beats all other 0:10:14.520 --> 0:10:18.360 aspects of data. Right. It's more important than timeliness. It's 0:10:18.400 --> 0:10:22.200 more important than you know, how large the field is. 0:10:22.679 --> 0:10:25.559 The most important thing is it's a trusted data source. 0:10:25.600 --> 0:10:30.200 And so all the data that we're collecting and displaying 0:10:30.360 --> 0:10:36.520 in our solution is from local government, state government, or 0:10:36.520 --> 0:10:41.320 federal government sources. So we're not we're not doing crowdsourcing. 0:10:41.400 --> 0:10:45.840 We're not pulling in social media opinions or things like that, 0:10:45.880 --> 0:10:48.200 although those things have their place and are helpful in 0:10:48.240 --> 0:10:52.360 their own way. This is really about about aggregating and 0:10:52.360 --> 0:10:56.520 collecting all the local sources so that you can see 0:10:56.520 --> 0:10:59.880 what's going on in your community, right And I think 0:10:59.880 --> 0:11:03.800 that is really important. You know, I have a have 0:11:03.960 --> 0:11:09.120 a uncle that was in Louisiana. He's a great guy, 0:11:09.520 --> 0:11:11.800 but he's a free spurt. He does his own thing. 0:11:12.000 --> 0:11:15.960 He lives his own way, and he doesn't really listen 0:11:16.480 --> 0:11:19.640 that well too, friends or family. He's going to do 0:11:19.760 --> 0:11:23.439 his own thing. And so, you know, as we started 0:11:23.440 --> 0:11:26.000 building building this tool and I started seeing the data 0:11:26.080 --> 0:11:27.679 that was going on there, I was able to use 0:11:27.720 --> 0:11:30.880 it to show him that, you know what what and 0:11:30.880 --> 0:11:34.880 this was you know, last weekend and Monday. That would 0:11:34.920 --> 0:11:39.240 show him, Look, there's really some really serious outbreaks going 0:11:39.280 --> 0:11:43.199 on in Louisiana right now. This is really accelerating. Uh. 0:11:43.280 --> 0:11:46.040 You might not take this seriously, but your people in 0:11:46.080 --> 0:11:50.560 your community really need you to to listen to the 0:11:50.600 --> 0:11:55.600 local authorities and stay put right, stay inside and and 0:11:55.679 --> 0:11:59.360 behave um. And so I think, you know, you take 0:11:59.480 --> 0:12:04.760 my example and extrapolate that out across Every American has 0:12:04.800 --> 0:12:09.200 a story somewhat similar to had I think, Uh, and 0:12:09.240 --> 0:12:12.240 so we're all looking and seeing what's happening to friends 0:12:12.240 --> 0:12:16.680 and family around the world and across the United States, 0:12:16.720 --> 0:12:21.600 and uh and you know, staying home, stopping the spread 0:12:22.200 --> 0:12:25.840 ultimately saves lives, right it is that that's simple right now, 0:12:26.920 --> 0:12:31.000 and be able to show people what's happening in their community. 0:12:31.040 --> 0:12:34.560 We're doing in three ways. So we're showing them the data, 0:12:34.679 --> 0:12:39.760 how many people have been tested and shown as positive 0:12:39.920 --> 0:12:43.520 in their county, how many deaths have been recorded. We're 0:12:43.559 --> 0:12:47.200 showing the trend line for that county day over day. 0:12:47.280 --> 0:12:50.319 So is it getting steeper, is the curve getting steeper 0:12:50.400 --> 0:12:53.000 or is it plateau ng or hopefully at some point 0:12:53.000 --> 0:12:56.000 here we'll see it declining in places, but right now 0:12:56.040 --> 0:13:00.160 we're not. It's still on the upswing. Then we we 0:13:00.240 --> 0:13:02.560 let them choose between their state view or their county 0:13:02.640 --> 0:13:06.920 view of that trained analysis. Then we're making it tangible 0:13:06.960 --> 0:13:09.160 on a map. Right What we found with with weather 0:13:09.240 --> 0:13:12.360 and we're doing now with the COVID virus is plotting 0:13:12.360 --> 0:13:14.800 it on a map so you can actually see it 0:13:14.840 --> 0:13:19.200 in a geographical context. So your county versus the county 0:13:19.240 --> 0:13:22.600 next to you, and across the entire United States, every 0:13:22.640 --> 0:13:27.520 county that that's producing data we're ingesting. There are some 0:13:27.640 --> 0:13:32.160 places that aren't producing data, particularly in some rural counties, 0:13:32.160 --> 0:13:36.079 but for the most part, the major population centers are 0:13:36.080 --> 0:13:39.640 all producing this data, and you can see what's going on, right, 0:13:39.720 --> 0:13:44.319 and I think we've seeing people make better decisions as 0:13:44.320 --> 0:13:47.920 a result of it, and that's the whole purpose. That's 0:13:47.920 --> 0:13:50.959 why we did this. One of the things you mentioned 0:13:51.120 --> 0:13:53.840 was about the trusted sources. I'm glad you went into 0:13:53.880 --> 0:13:57.240 that and explained where you're pulling information from, because obviously 0:13:57.320 --> 0:13:59.880 we're right now in an era where there's missing for 0:14:00.080 --> 0:14:05.439 nation and even disinformation running rampant online. So it's good 0:14:05.440 --> 0:14:08.920 to be able to point to a tool and actually 0:14:08.960 --> 0:14:14.040 know where where's this tool pulling information from. And it's 0:14:14.040 --> 0:14:16.200 also really interesting to me that it's taking a very 0:14:16.240 --> 0:14:18.959 similar approach to what the Weather Company has done. I 0:14:19.040 --> 0:14:23.240 think everyone has had the experience of using either the 0:14:23.320 --> 0:14:27.200 app or the website and looking at, you know, weather 0:14:27.280 --> 0:14:31.320 forecasts for specific zip codes and so kind of taking 0:14:31.360 --> 0:14:35.320 that same thinking and applying that to the COVID nineteen 0:14:36.040 --> 0:14:42.040 outbreak is really interesting to me too. I'm wondering, um, 0:14:42.080 --> 0:14:47.320 with that in mind, uh, are the sources you're pulling from? 0:14:47.360 --> 0:14:50.000 You know there are local, state and federal government. Are 0:14:50.000 --> 0:14:53.240 they in different formats? Because to me, that would present 0:14:53.240 --> 0:14:57.800 itself an enormous challenge because machines. Humans are really good 0:14:57.840 --> 0:15:01.640 at pulling information from whatever format we encounter, machines typically 0:15:01.840 --> 0:15:06.600 are not. So this is actually the heart of the 0:15:06.640 --> 0:15:13.080 technology challenge we've had and and we've basically used IBM 0:15:13.160 --> 0:15:18.000 S Watson AI tools to be able to ingest this 0:15:18.280 --> 0:15:23.200 data from literally thousands and thousands of data sources, all 0:15:23.280 --> 0:15:27.000 in different formats, right, everything from a PDF that's actually 0:15:27.040 --> 0:15:33.560 an image like a photograph essentially two HTML files to 0:15:34.440 --> 0:15:38.600 or everything else in between. And every single counties website 0:15:38.640 --> 0:15:43.720 is built differently. There's no technology standard that's been applied. UH. 0:15:43.760 --> 0:15:46.440 And so you know, I think that one of the 0:15:46.600 --> 0:15:50.160 one of the amazing things from a technology only perspective 0:15:51.120 --> 0:15:54.960 in this is that the Watson's AI was able to 0:15:55.800 --> 0:16:00.280 INGI and learn all of these different formats based really 0:16:00.320 --> 0:16:04.720 on its own and ing all this data from all 0:16:04.760 --> 0:16:08.440 these different technology types, uh and put it into a 0:16:08.480 --> 0:16:11.760 standardized format that we could then produce and and and 0:16:11.880 --> 0:16:15.640 show in our websites and our acts and and that 0:16:16.000 --> 0:16:20.880 whole training from when we started to when we had 0:16:20.920 --> 0:16:24.680 the first standardized file was thirty six hours, so we're 0:16:24.720 --> 0:16:28.840 not talking about two weeks or or anything. This is like, 0:16:29.120 --> 0:16:33.720 you know, a data hoff to learn and train and 0:16:33.800 --> 0:16:36.960 collect the data and put it into a standardized format. 0:16:37.600 --> 0:16:41.080 Cameron mentioned training the system on data, and some of 0:16:41.120 --> 0:16:44.760 you might wonder what that actually means. It's a term 0:16:45.040 --> 0:16:48.640 used in machine learning in which engineers feed information into 0:16:48.680 --> 0:16:52.720 computer models in an effort to produce particular results. And 0:16:52.760 --> 0:16:55.920 typically you start out doing this by knowing what results 0:16:56.000 --> 0:16:58.960 you want ahead of time when you're first training the system. 0:16:59.320 --> 0:17:01.920 That way, you can see what comes out of the model, 0:17:02.160 --> 0:17:04.560 you compare it to what you expected to see, and 0:17:04.600 --> 0:17:06.720 if the two don't match up, you can go back 0:17:06.760 --> 0:17:10.200 and tweak the model. So, for example, let's say you're 0:17:10.240 --> 0:17:14.159 training an image recognition computer model to recognize pictures of 0:17:14.400 --> 0:17:17.960 fire hydrants. You might feed a ton of images of 0:17:18.000 --> 0:17:20.720 fire hydrants to the model. Then you might introduce a 0:17:20.760 --> 0:17:24.360 mixture of images that include fire hydrants and other stuff, 0:17:24.680 --> 0:17:27.520 seeing if the model can tell the difference sorting the 0:17:27.560 --> 0:17:31.119 images properly. So you analyze those results and if they're good, 0:17:31.160 --> 0:17:34.399 you keep going. You might use millions of points of 0:17:34.520 --> 0:17:37.800 data to train your model. Over time, and often this 0:17:37.880 --> 0:17:42.760 is a painstakingly slow process, particularly when engineers need to 0:17:42.760 --> 0:17:45.720 step in and change something about the model that isn't 0:17:45.800 --> 0:17:49.560 quite working. Once you are able to get reliable results 0:17:49.600 --> 0:17:52.600 from the model, you can put it to more practical use, 0:17:52.880 --> 0:17:56.440 though it may require future tweaks when the model encounters 0:17:56.600 --> 0:18:00.399 something well outside the norm. Okay, let's get back to 0:18:00.440 --> 0:18:03.880 my conversation with Cameron Clayton, the general manager of IBM 0:18:03.920 --> 0:18:07.560 Cloud Ecosystem and of the Weather Company and IBM business. 0:18:08.200 --> 0:18:12.240 You're talking about things like natural language processing, being able 0:18:12.280 --> 0:18:15.600 to access all these different styles of data, whether it's 0:18:15.640 --> 0:18:18.400 an image file or it's something that can be searched 0:18:19.080 --> 0:18:23.160 with an algorithm, and then taking the meaning of that 0:18:23.280 --> 0:18:25.320 not just the fact that the data is there. You 0:18:25.359 --> 0:18:28.040 have to the MA gene has to understand what the 0:18:28.080 --> 0:18:31.479 meaning is in order to put it into its model 0:18:31.560 --> 0:18:36.240 and then present it. These are really tough engineering challenges 0:18:36.440 --> 0:18:40.719 in computer AI in general, and so to see that 0:18:40.800 --> 0:18:46.639 application being put so quickly in place, exactly can you 0:18:46.680 --> 0:18:48.880 give us an idea of how long it took too, 0:18:49.760 --> 0:18:52.800 from the point of ideation to the point of implementation 0:18:53.480 --> 0:18:56.480 that you guys went through in order to produce this. 0:18:57.040 --> 0:19:01.680 So the timeline from when we started UH this getting 0:19:01.720 --> 0:19:06.520 it getting it live was probably about ten days. We 0:19:06.640 --> 0:19:09.960 got inbound interest from our fans around the world. We 0:19:10.080 --> 0:19:13.760 then mocked up how we wanted to present the data 0:19:13.880 --> 0:19:17.480 from a visual point of view. We then, in parallel, 0:19:17.520 --> 0:19:20.800 we're rapidly trying to source all these local data sources. 0:19:20.800 --> 0:19:22.760 We wanted it from the very beginning of as local 0:19:22.800 --> 0:19:26.960 as possible UH, and we rapidly realized this was not 0:19:27.040 --> 0:19:30.280 something that you could do manually. You had to do 0:19:30.400 --> 0:19:34.080 this UH in an automated way at scale, and so 0:19:34.080 --> 0:19:37.760 that's when we brought in Watson and AI to help. 0:19:39.000 --> 0:19:42.760 Thirty six hours later we had data. We had, you know, 0:19:43.320 --> 0:19:45.800 a format, but then we really spent a lot of 0:19:45.840 --> 0:19:50.080 time testing right to make sure that the data was correct, 0:19:50.160 --> 0:19:53.560 to make sure that you know, when when you know, 0:19:53.760 --> 0:19:58.080 Governor Cumos speaks in New York at eleven thirty every morning, 0:19:58.359 --> 0:20:00.240 within a few minutes, we were able to up dat 0:20:00.320 --> 0:20:03.359 the data based on the numbers he's sharing with the media, 0:20:03.440 --> 0:20:07.919 but on the New York State website, for example. And 0:20:07.960 --> 0:20:12.440 so there's all these different factors both technology and format 0:20:12.800 --> 0:20:14.840 that we had to take into account, but also just 0:20:15.480 --> 0:20:20.000 testing this. The second is we always, due to our 0:20:20.040 --> 0:20:22.520 scale and the number of users we have, we have 0:20:22.600 --> 0:20:26.040 to load test and make sure that technologically we can 0:20:26.080 --> 0:20:29.720 deliver to the millions of people that use our properties 0:20:29.720 --> 0:20:34.480 and platforms. And so we've got this uh live a 0:20:34.520 --> 0:20:39.720 few days ago and beginning to end is probably seven days. 0:20:40.600 --> 0:20:43.800 There's a remarkable achievement. I mean, you're you're talking about 0:20:43.920 --> 0:20:47.360 everything from coming up with the idea to the design 0:20:47.440 --> 0:20:50.240 saying what do we actually do to make this possible? 0:20:50.320 --> 0:20:56.080 To even even something that seems to someone like me simple, 0:20:56.200 --> 0:21:00.560 like how do you present that information to the consumers 0:21:00.640 --> 0:21:03.639 so in a way that makes sense, Like we we 0:21:03.720 --> 0:21:05.600 only see it at the end, right, we see it 0:21:05.640 --> 0:21:07.959 after you've made all those decisions, and we look at 0:21:08.000 --> 0:21:10.040 it and we say, oh, yeah, of course that makes sense. 0:21:11.080 --> 0:21:14.359 But you have to get there first on the design side. 0:21:14.480 --> 0:21:17.000 And I think a lot of people don't understand or 0:21:17.040 --> 0:21:19.760 appreciate how challenging that part of technology is to not 0:21:19.920 --> 0:21:21.760 just the making it work, but making it work in 0:21:21.800 --> 0:21:26.680 a way that is ultimately useful to the end consumer too, 0:21:26.720 --> 0:21:31.520 so that yeah, it doesn't just work, it works for you. Um, 0:21:31.640 --> 0:21:33.720 and we were you know, one of the other things 0:21:33.720 --> 0:21:36.160 I wanted to talk about was that this design process 0:21:36.240 --> 0:21:39.639 not only was it rapid, but obviously it was unusual 0:21:39.800 --> 0:21:43.719 in that you weren't all working in the same workspace 0:21:43.760 --> 0:21:47.560 at the same time because of the concerns, the health concerns. 0:21:47.600 --> 0:21:50.160 So what was that Like, how did your team respond 0:21:50.320 --> 0:21:56.600 to working in a decentralized approach. So I gotta say 0:21:56.640 --> 0:22:02.200 the team really has leaned in here hod like countless 0:22:02.680 --> 0:22:07.400 almost before. We had sixty people touched this project. Over 0:22:07.440 --> 0:22:12.600 the course of the week. I would say fifty five 0:22:12.680 --> 0:22:16.679 of the sixty worked multiple twenty four hour days in 0:22:16.760 --> 0:22:21.399 that time period. Uh And so sleep deprivation was a 0:22:21.440 --> 0:22:25.800 real issue as well, right, But but we have we 0:22:25.840 --> 0:22:29.679 have great tools, you know, whether it's video conferencing and 0:22:29.720 --> 0:22:36.160 collaboration tools where we can actually iterate and design products remotely, 0:22:36.480 --> 0:22:38.560 but all on the same screen at the same time. 0:22:39.359 --> 0:22:42.160 So our design is ractually able to you know, one 0:22:42.160 --> 0:22:44.400 of them draws a line and the other one can 0:22:44.520 --> 0:22:48.800 raise it as as they're drawing it. Uh. And so 0:22:49.640 --> 0:22:53.359 that was both fun and challenging at the same time. 0:22:53.680 --> 0:22:56.760 I would say the goal of the product was to 0:22:56.840 --> 0:23:02.960 make it clean and simple, so it is digestible, and 0:23:04.160 --> 0:23:06.560 I think you know, we can always add more data 0:23:06.600 --> 0:23:10.919 fields over time and and add more information, but the 0:23:11.000 --> 0:23:15.360 real purpose of this was to help people understand what's 0:23:15.359 --> 0:23:17.560 happening in the community so that they would stay home, 0:23:17.840 --> 0:23:21.000 stop the spread, and ultimately save lives. Right. Making that 0:23:21.600 --> 0:23:24.879 invisible visible was was the goal of this UM and 0:23:24.920 --> 0:23:28.640 so I think we've we've achieved that, but it's only 0:23:28.680 --> 0:23:32.600 through hard work of a small group of folks that 0:23:33.640 --> 0:23:41.080 that you know, really worked hard for seven days without sleep. Wow, 0:23:41.160 --> 0:23:44.440 I mean, that's that's incredible. So, so they're working hard 0:23:44.640 --> 0:23:47.960 putting this all together. Meanwhile, you've got the AI and 0:23:48.040 --> 0:23:51.400 tools working hard in the background to synthesize all that data. 0:23:51.440 --> 0:23:53.639 Can you talk a little bit more about the specific 0:23:54.080 --> 0:23:57.520 technologies running in the background. We've mentioned Watson, but is 0:23:57.560 --> 0:24:00.800 there anything else along with that. This is a whole 0:24:01.040 --> 0:24:05.199 whole variety of things that that live in the background 0:24:05.240 --> 0:24:08.920 that make this possible and so and almost all of 0:24:08.960 --> 0:24:10.960 those things are things that we don't think about as 0:24:11.040 --> 0:24:15.680 consumers of whether dot Com or our mobile apps. So 0:24:15.920 --> 0:24:20.800 one is just the cloud infrastructure itself. The fact that 0:24:22.040 --> 0:24:27.639 you know on Monday night when we launch, you know, 0:24:27.920 --> 0:24:30.520 we were alive for like three hours. We had about 0:24:30.520 --> 0:24:35.679 a million users in three hours used the property. On Tuesday, 0:24:35.840 --> 0:24:39.600 we had three and a half million users start using 0:24:39.600 --> 0:24:44.560 the property. On Wednesday it was up to like five 0:24:44.640 --> 0:24:47.959 and a half or six million users. You can't scale 0:24:48.080 --> 0:24:51.879 like that, and those are each individual, unique visitors, some 0:24:51.920 --> 0:24:55.840 of them visited multiple times and checks, you know, tens 0:24:55.840 --> 0:24:59.440 of locations around the country for friends and family. Uh. 0:24:59.480 --> 0:25:01.800 And you can't have that kind of scale without having 0:25:01.800 --> 0:25:05.760 a really robust cloud infrastructure behind it. And so uh, 0:25:06.000 --> 0:25:10.640 IBM Cloud was has been and continues to be instrumental 0:25:10.960 --> 0:25:15.640 and sort of invisibly in the background, keeping the infrastructure alive. 0:25:16.240 --> 0:25:17.520 And one of the one of the things about that, 0:25:17.560 --> 0:25:19.639 I'll take a quick story on it, that was super 0:25:19.680 --> 0:25:23.000 impressive to me. And I see this these kinds of 0:25:23.280 --> 0:25:27.720 launches fairly often with our products. But as we launched 0:25:27.720 --> 0:25:30.879 on Monday night, what I was not used to seeing 0:25:31.040 --> 0:25:37.439 was security h automatically being implemented. And so what I 0:25:37.440 --> 0:25:39.640 mean by that is we were actually having a denial 0:25:39.680 --> 0:25:43.280 of service attack, so hackers, we're trying to hack into 0:25:43.359 --> 0:25:47.639 weather dot com as we were launching the side, and 0:25:47.680 --> 0:25:52.480 because of the security elements of IBM Cloud, it didn't 0:25:52.520 --> 0:25:55.160 stop out. I say to our team, should we stop, 0:25:55.359 --> 0:25:58.520 like this is seems like a really big deal, and 0:25:58.560 --> 0:26:01.800 they were like, no, this is totally fine. We deal 0:26:01.840 --> 0:26:04.919 with this all the time, all the securities you know 0:26:05.240 --> 0:26:07.880 in place. I don't say that to try and bring 0:26:07.920 --> 0:26:13.640 on anymore uh challenges. We don't definitely don't want that. 0:26:14.359 --> 0:26:16.320 But it was really impressive to me to see how 0:26:16.400 --> 0:26:20.360