DATA AND GRAPHICS: An introduction

By • Humanitarian Mapathon in UCLA

Recording

Show the timestamped transcript
  1. or maybe he's still trying to get his zoom in order.
  2. - Can you hear me?
  3. - I can hear you.
  4. - Yeah.
  5. - This has been a truly dramatic, like, point of view.
  6. (laughing)
  7. I'm zooming you from my phone
  8. because I could not get my computer to zoom in.
  9. I'm so sorry.
  10. I am one of the recalcitrant nerds who insists
  11. on continuing to use Linux,
  12. even though everyone else has abandoned it.
  13. And I use zoom every day with no trouble,
  14. but I was told by your system
  15. that I needed to upgrade my zoom.
  16. And so I've spent the last 20 minutes
  17. trying to upgrade my zoom.
  18. (laughing)
  19. So hard, so hard.
  20. I know, so now I have zoomed you from my phone,
  21. which is probably what I should have done in the first place.
  22. And so I'm really, really, really, really sorry.
  23. Thank you for your patience.
  24. - Oh, no worries, Ben.
  25. We are so thrilled to have you here.
  26. I just wanted to give you a little bit of a preview
  27. of what's been going on prior to you arriving.
  28. You know, this event started on Wednesday
  29. and it's a humanitarian mapathon event.
  30. And the goal was to digitize 20,000 buildings
  31. using hot OSM and open street maps.
  32. I don't know if you can see from my virtual background,
  33. but between our two campuses, UCLA and USC,
  34. we each digitized collectively 10,000 buildings
  35. at each school.
  36. So we reach our miles to 20,000 buildings,
  37. which were digitized mostly.
  38. We were working in Indonesia and Sudan, South Sudan.
  39. And yesterday we had back to back to back workshops
  40. around topics of humanitarian mapping,
  41. data science, Python, Jupyter Notebooks, Tableau, QGIS,
  42. all these kinds of tools that I know you're familiar with
  43. with the work that you do and the team that you manage.
  44. So we're all very excited to have you here.
  45. I just wanted to quickly kind of introduce you
  46. with the work that I know that you've done.
  47. So let me just quickly,
  48. I'm just very briefly share my screen.
  49. So you look quite different from your picture here.
  50. - Yeah, COVID's been hard on us all.
  51. - Yeah, yeah, you have this kind of rock star
  52. vibe about you today.
  53. COVID has made us all into, you know, punk stars.
  54. But Ben, you know, if you're from Los Angeles
  55. and are familiar with the type of work
  56. that the LA Times has done over the years,
  57. we know Ben has this kind of data journalist
  58. who has put together a lot of the data,
  59. you know, journalistic stories around the LA Times
  60. and our local communities.
  61. I happen to, let me see,
  62. know about Ben's work primarily through the pandemic
  63. because when this came about tracking the coronavirus
  64. in California was a site launched by the LA Times
  65. earlier this year.
  66. And Ben led the effort to work with community organizers,
  67. with hospitals to kind of have a one-stop shop
  68. where all this information about the coronavirus was collected.
  69. It's really an amazing resource if you haven't been there.
  70. It has all the almost real-time data statistics
  71. about coronavirus, kind of more focused locally.
  72. So we know we have data from Johns Hopkins,
  73. that's global and kind of more national,
  74. but where do we find neighborhood level data
  75. about what's going on in Los Angeles?
  76. And this is just an incredible resource.
  77. You know, it's full of real-time data graphics
  78. that's updated hourly.
  79. And, you know, unfortunately, it's still relevant today.
  80. You know, back in March, when I first talked to Ben
  81. about this, I was actually thinking, you know,
  82. maybe the last of the summer.
  83. But that's the reality of the moment.
  84. It's an incredible resource with data charts, maps.
  85. And also, what makes this resource so useful
  86. for us in academia is that, you know, Ben's a huge,
  87. as he said, you know, Linux operating systems
  88. about open data.
  89. So, you know, this week we've been thinking so much
  90. about open data, crowdsourced information,
  91. using open street maps.
  92. And the data desk information provided by the LA Times
  93. allows us to really work with data that's transparent,
  94. that's made publicly available.
  95. And this Coronavirus GitHub page is where I believe
  96. Ben's team is contributing to the community
  97. by providing almost real-time data and statistics
  98. about all the information that they're collecting.
  99. So, I'll leave my talk to that
  100. and let Ben take over now.
  101. And once again, thank you, Ben, for joining us today.
  102. - Yeah, thank you.
  103. I had just emailed you a link to my deck.
  104. So, I'm stuck on my phone.
  105. Would you mind calling that up
  106. and just sharing your screen if it's not too much hassle?
  107. - Oh, no problem.
  108. Hold on one second.
  109. - Sure.
  110. I'll spiel for, I don't know, not too long,
  111. but I was hoping to give everybody here an overview
  112. of our team at the LA Times,
  113. the data and graphics department,
  114. kind of what we do, the types of people who are on the team,
  115. the types of things we make,
  116. so you guys can get a sense of, you know,
  117. how these kind of data skills live
  118. within the journalism world.
  119. And then, you know, we can answer any questions
  120. anybody has that really talk about anything
  121. you guys wanna talk about.
  122. And along the way, if you have anything
  123. you wanna know more about,
  124. or if I'm not explaining something good,
  125. just stop me, just talk.
  126. I'm happy to have a conversation
  127. and glad to be here.
  128. So, I also think it's really cool
  129. you guys are doing open street map stuff,
  130. which I myself have dabbled in a little bit,
  131. trying to do one of my hometowns in Eastern Iowa,
  132. where I grew up.
  133. So, well, where to begin?
  134. So, my name is Ben Welsh.
  135. I'm the editor of the data and graphics department
  136. at the LA Times.
  137. And we're a team of about 20 people
  138. who try to use our computer skills to like,
  139. make stuff happen.
  140. And this presentation will give you kind of an overview
  141. of what that ends up being.
  142. So, go ahead and click on.
  143. So, there's me before COVID.
  144. My wife calls this my Bible salesman photo.
  145. And obviously, next is me after COVID.
  146. We already got to this joke though,
  147. so we don't need to stick with it.
  148. Shout out to all the Twin Peaks fans out there.
  149. We can keep going.
  150. And this deck is one that I have presented internally
  151. at our company,
  152. and it initially included some private business figures,
  153. which I've redacted just to not get in trouble.
  154. But I'll give you the gist of what I was trying to get across
  155. when we get to that point, okay?
  156. Keep clicking.
  157. So, the data and graphics department,
  158. its mission, as best as I can articulate it,
  159. is for us to create digital journalism
  160. that's important to the readers of the LA Times,
  161. both in what they want to read
  162. and what they ought to read, right?
  163. With using data development and design.
  164. And by development,
  165. I mean actual computer code development, right?
  166. So the people that are on this team, if you hit the next slide,
  167. are not just reporters and editors and journalists
  168. and people like me who went to journalism school.
  169. They're also at the same time,
  170. computer programmers, data analysts,
  171. information designers, data scientists,
  172. whatever you want to call it, they're nerds, right?
  173. And so, we're kind of one of the more multi-hyphened teams
  174. at the LA Times and that people are called upon
  175. to not just do the traditional skills of journalism,
  176. but also to have these technical skills
  177. that let them find and tell stories in other ways.
  178. And this is something that's increasingly common
  179. across the journalism industry.
  180. Our team is not unique.
  181. There's kind of an evolution happening
  182. in graphics and data departments across the country
  183. where graphics is becoming kind of a higher-skill
  184. data and programming profession.
  185. And our team is caught in the same currents
  186. that are driving that, right?
  187. And so, the people on our team have to be pretty darn flexible.
  188. And so, if you hit the next slide, you can get a look at them.
  189. Like I said, it's about 20 different people.
  190. They come from many of them from Southern California,
  191. many of them from UCLA and USC, or at least a couple,
  192. but also from all across America.
  193. And they're people with a lot of different backgrounds
  194. and skills.
  195. We try to have everybody learn as many
  196. of the different types of things we do as possible,
  197. but the way it works out is some people tend to be
  198. a little better at analysis or maybe a little better
  199. at visualization, but we don't specialize.
  200. We really do kind of try to treat everybody
  201. as being able to take on all of the challenges.
  202. And so, we're always learning and we're always experimenting
  203. and trying to take on new skills.
  204. Just this morning, I was helping someone learn
  205. how to write some spelt, if anybody here is a spelt fan,
  206. to make some of our election graphics for next week.
  207. So, if you hit the next slide, we're gonna get into
  208. what are these things we actually make?
  209. In a lot of ways, to me, a newspaper where I've worked
  210. at the LA Times for 13 years now and seen a lot of change,
  211. but even though we're making different things
  212. and different products, it is kind of an information factory.
  213. So, in a certain sense, a news organization
  214. takes the raw materials of life, of things
  215. that are happening in the world,
  216. and it refines those raw materials into products,
  217. news products that you can sell,
  218. and the people want that help them make sense of the world.
  219. And traditionally, in my painful metaphor,
  220. the assembly lines at this factory were really focused
  221. on making a print newspaper that had blocks of text, right?
  222. With headlines and photos, and that was like,
  223. what the majority of the assembly lines were set up to do.
  224. There were other ones to have sports scores and boxes
  225. and to give you the movie listings
  226. and all these other things.
  227. But in the rewiring of our economy,
  228. that's brought on by the internet and all that,
  229. the factory of the LA Times has to be refitted
  230. with new assembly lines to make different products, right?
  231. That are more in line with what people want
  232. that can compete with other factories out there
  233. in the marketplace, and that can help us shift
  234. to a different type of business,
  235. which is more focused on getting money from subscriptions
  236. and readers who like us enough to pay us
  237. and less so on advertising,
  238. which if you guys are interested in all that business stuff,
  239. we can talk about it later, but I won't be too boring.
  240. And in our little department, it's my view that our assembly,
  241. we basically have three assembly lines
  242. that the people you saw before kind of interchangeably work
  243. on.
  244. The first one makes applications, software websites, right?
  245. The second makes visual stories.
  246. These are stand up, we'll get to those.
  247. And then the third is making digital designs.
  248. And so let's go through those one, two, three,
  249. and I'll show you a bunch of examples of each.
  250. So first up is applications.
  251. This is where we use our software skills
  252. to gather and refine data to serve our digital audience,
  253. writing computer code to go get data,
  254. filing public records requests to get it,
  255. otherwise building databases ourselves from the ground up
  256. that we can then turn into a web application
  257. that is not a traditional story
  258. that people will wanna read and maybe even pay for, right?
  259. And so if you hit the next slide, the first example,
  260. I guess would be the coronavirus tracker.
  261. This is a recent one.
  262. This is where we're gathering data
  263. from a ton of different sources
  264. and turning it into all these pages
  265. that update as frequently as we can get them to,
  266. and let people slice and dice and tailor them.
  267. There's also 58 pages, one for each local county, right?
  268. And we're always working to look and expand on this.
  269. There's a few things I'm hoping we can get out soon.
  270. Yes, that is Netscape.
  271. This is my visual pun for this presentation.
  272. If you'll bear with me, I'm showing my age.
  273. Okay, so then like another example,
  274. if we keep going would be our live wildfires map.
  275. So this is where we're pulling in data
  276. from a lot of public sources and one private one
  277. to try to give people the most comprehensive,
  278. kind of composite view of what's happening
  279. with wildfires this minute in California
  280. or the most recent one possible.
  281. And it's in no way a traditional story, right?
  282. It's a single webpage that updates and stands along.
  283. If you hit the next one,
  284. another example would be our quake bot.
  285. So this is something that isn't like a standalone webpage.
  286. It's something else.
  287. It's a piece of automation.
  288. And so every time there's an earthquake detected
  289. by the USGS, it sends out like a data feed
  290. to the whole world, like kind of little pulses
  291. that say here's the latest earthquake
  292. and how big it was and where it was.
  293. And we have an algorithm that reads every one of those
  294. based on how close they are to Los Angeles
  295. and how high the magnitude is.
  296. We have some sort of editorial barriers.
  297. And then if the data surpasses those barriers,
  298. this blog post is automatically written
  299. and this map is automatically made
  300. and prepared for publication.
  301. And so within a minute of an earthquake happening,
  302. we can have the complete post,
  303. including the map ready to roll.
  304. We still have a human look at it, of course,
  305. before it goes live,
  306. but this is where automation gives us
  307. kind of a quick step out the door immediately.
  308. And then the next one would be, I think,
  309. our most recent police shootings database,
  310. we have a long running project called the homicide report
  311. where we have built from the ground up
  312. our own independent database of every person
  313. killed by another in LA County since the year 2000.
  314. And this year we decided to repurpose a slice of that data
  315. to be a standalone database that just tracks police shootings
  316. and gives people a sense of the latest trends
  317. and that now, like the homicide report,
  318. live updates every time we get new records in.
  319. These databases, they power these standalone applications
  320. and they would probably worth doing just for that reason,
  321. but they also have the side benefit
  322. of allowing us to do enterprise analysis
  323. into these issues that we otherwise couldn't do
  324. if we didn't have the data, right?
  325. So on the next slide, you can see,
  326. since we've been gathering all this COVID data
  327. for months and months,
  328. we've been able to do tons of stories about COVID
  329. that are a little more traditional,
  330. but that have stronger claims and more insightful analysis
  331. because we have this data to draw from.
  332. So one example would be this kind of step back piece
  333. that we did after LA County's sort of botched reopening,
  334. where we were able to use all the data we gathered
  335. and other things we pulled in to kind of tell that story
  336. of what happened here in LA
  337. in a way that wasn't just a data application, right?
  338. And then the next one would come from our homicide report
  339. database.
  340. If you guys follow the news,
  341. you may have heard about the case of Deshawn Kizzie,
  342. who was a young man,
  343. killed following a bicycle stop in South LA
  344. by sheriff's deputies.
  345. And when that happened,
  346. it sort of raised a question of,
  347. well, how common are these bicycle stops
  348. that lead to fatal encounters?
  349. And because we had gathered this database
  350. and had all this information,
  351. we were the only media outlet that was able to give
  352. the public the news that, hey, this isn't a loan incident.
  353. This is something that's happened 15 times
  354. in the last 15 years.
  355. And so it's not especially common,
  356. but it's more common than this being the first time.
  357. And you're able to bring context and insight to it.
  358. You can see if you see that map fly by really quick
  359. that these incidents have been concentrated
  360. in certain areas too.
  361. We have some public records requests out
  362. that I believe will allow us to look into that further
  363. in the future.
  364. So that's thing one,
  365. that's the kind of the applications and the analysis.
  366. We have live election results coming on Tuesday.
  367. So this is all the flowing data feeds
  368. of votes coming in from across America.
  369. We'll go into maps and tables and charts
  370. to help you kind of make sense of what's going on, right?
  371. And so all right, so now we can do thing two.
  372. So assembly line number two is visual stories
  373. is what I call them.
  374. This is kind of an emerging story type
  375. in the world of data and graphics online
  376. that nobody really has a good name for.
  377. So if anybody here has a better name for it,
  378. I'd love to hear it.
  379. You can hit the next slide.
  380. And so visual stories are when graphics reporter reports
  381. and designs visual coverage that is substantial enough
  382. to stand on its own.
  383. Traditionally graphics departments
  384. had sort of made accompanying graphics.
  385. So on the old assembly line,
  386. you might be making a front page story
  387. and you got to have 1600 words of text
  388. and you got to have a headline
  389. and you got to have maybe a couple photos.
  390. And hey, if you're really going to dress up
  391. that front page story, you might get a bar chart
  392. and go with it, right?
  393. And so a lot of graphics departments
  394. had been oriented around being a support desk
  395. that was producing these sort of supplementary graphics
  396. for traditional stories.
  397. And we love those graphics.
  398. They're a good thing, I like them.
  399. But I think we have the potential to do quite a bit more
  400. and I think our audience wants something a little more.
  401. And so one thing I bet everybody here has noticed
  402. reading some of the leading news outlets in America
  403. is that they now produce these visual stories
  404. that are primarily just a combination of graphics
  405. that are big enough that they can just stand on their own, right?
  406. And that's what I call a visual story.
  407. And these are pieces that at most places
  408. are reported, pitched and executed primarily
  409. by the graphics reporter.
  410. It's not a case where a writer of a traditional story
  411. comes to the graphics department and asks for a graphic, right?
  412. So the entire process of how things kind of come together
  413. has changed and the graphics reporter
  414. is more in the driver's seat, okay?
  415. And so go ahead and hit the next slide
  416. and we'll look at some examples.
  417. So early this year, as I'm sure everyone here knows,
  418. Kobe Bryant died in a tragic helicopter accident
  419. near Calabasas and immediately we thought
  420. there's a visual component to the story.
  421. For people to understand what happened,
  422. you need to know the path of his helicopter, right?
  423. And so on day one, it was like a map of where did he crash
  424. and it's just like an X.
  425. And then between day one and day two,
  426. it's like, well, what was the path of the helicopter
  427. and what was really going on and what's the play-by-play, right?
  428. And then what we kind of learned on,
  429. it happened on a Sunday, what we kind of learned on Monday
  430. is like, oh, well, it seems like the final like part
  431. of this helicopter flight is where things really went wrong.
  432. And this was kind of the crucial moment.
  433. And so we decide, well, let's try to do a story
  434. that helps people see or understand as best as we can
  435. what happened.
  436. And so we took the publicly available data about the helicopter
  437. and we turned it into like a three or four step
  438. little graphic here that has, for each chapter in the story,
  439. has like a 3D map that helps you see what's going on.
  440. And by the end, you sort of are able to see
  441. this sort of important thing where the helicopter
  442. began climbing very rapidly in the last few seconds
  443. before a quick turn and then a crash.
  444. And that's the sort of thing that you can describe in words
  445. but is maybe better understood by seeing a 3D model, right?
  446. So the next example on the next slide
  447. would be something from politics.
  448. So we have this very tight DA race, we think.
  449. And after the primary, when we knew we were going
  450. to run off, we get the precinct level results
  451. that show you where people voted in every neighborhood.
  452. And we kind of did an analysis that looked at,
  453. well, where is the incumbent strong?
  454. Where is the challenger strong?
  455. And what would an upset coalition have to look like
  456. based on past election results
  457. and what we kind of know about this election?
  458. So this piece is really about helping people understand
  459. the political calculus of this local race through maps, right?
  460. And then the next one is we did a local mask survey.
  461. So there was this thing over the summer,
  462. the question of how many people are actually really wearing masks
  463. and rather than just quote public opinion polls
  464. or take data from other places,
  465. we decided to conduct our own survey.
  466. So a dozen of our reporters went to three different locations
  467. of multiple times over a few weeks.
  468. We followed a method that Jill Darling at USC
  469. helped us like figure out.
  470. And then we were able to kind of come back
  471. with our own provisional conclusions
  472. about how many people are wearing masks roughly
  473. and the variations between the different locations we looked at.
  474. And then that's presented here
  475. as sort of a designee standalone thing
  476. that's really more about charts and maps
  477. than it is about text, right?
  478. Next slide would be something that's not really data at all.
  479. It's more visual.
  480. And so in the George Floyd protests,
  481. I'm sure everyone saw on social media,
  482. there was a lot of footage of the police here in LA
  483. being pretty violent.
  484. And so we did a piece where we tried to aggregate
  485. as much as that as possible
  486. and then help compare it to the actual standards
  487. for use of force the police are supposed to follow
  488. to show the contrast.
  489. And so this is a case where we're using the design
  490. and the visual graph and the videos to tell the story
  491. and it's not really like charts and maps, you know what I mean?
  492. But it still falls within kind of this
  493. under this visual umbrella that I think is part of what we do
  494. and it's why I call it a visual story
  495. and not a graphic story or something, right?
  496. And we have another example I'm trying to remember.
  497. Oh, and then here's one we did on homeless housing.
  498. This is a type of visual story I bet you've all seen too
  499. that we call scrolly telling, which is a terrible term,
  500. which is where as you scroll things sort of move and happen
  501. and this is another type of visual technique
  502. that we use from time to time, depending on whether it fits
  503. the story we're trying to tell, okay?
  504. I think that's it for those examples.
  505. And then I think we get to things three,
  506. if I remember my deck right, which is digital designs.
  507. Oh, no, it doesn't fit the screen size.
  508. My gosh, you found a bug in my deck,
  509. but this is a case where we're using our digital design skills
  510. to elevate the newsroom's most ambitious journalism
  511. through like what we think
  512. are cool digital custom presentations.
  513. And in this case, we're usually helping to elevate stories
  514. developed by other departments rather than our own work.
  515. And a lot of it has some, you know,
  516. nothing to do with data or graphics.
  517. We're kind of the boutique web designers
  518. for these like flagship projects of the newsroom, okay?
  519. So examples of that wouldn't be on the following slide.
  520. So the first one would be we did an obituaries package
  521. of all the people who died from COVID.
  522. Not all of them of many people who died in COVID,
  523. of COVID here locally.
  524. And then that package has sort of a custom design
  525. to try to make it more dramatic and interesting
  526. and engaging for the reader.
  527. The next example is one from Sunday.
  528. I wonder if anybody here saw it.
  529. We had a piece by Rosanna Shee about DDT
  530. being dumped off the coast back in the day here in LA.
  531. And to try to make that story more impactful and grabby
  532. and to get people we did custom design for it,
  533. which included like a video topper,
  534. which you see here, a little scrolly telling in the middle.
  535. And then the sort of overall design and theme of the page
  536. is customized to fit the story.
  537. It isn't just the generic LA times kind of look, right?
  538. Now the next one is the Chicano moratorium package
  539. over the summer, we had an anniversary
  540. of a famous protest in East LA against the Vietnam War,
  541. which ended up being a series of stories.
  542. And in the past, this might have just been
  543. like a newspaper special suction,
  544. and then a bunch of pages on our website
  545. that aren't really connected.
  546. But what we try to do is to bring a style and theme
  547. to the whole package.
  548. This is kind of the cover page where all the stories
  549. were collected, and then each of the individual stories
  550. was given sort of a variation on the theme
  551. that you see here is the idea.
  552. And I think next up is the one I'm gonna,
  553. oh, then sometimes we do these for other platforms,
  554. not just for the open web.
  555. The Linux nerd in me would love it all just to be
  556. on RWW, but as far as our business goes,
  557. we do need to cross publish these things
  558. on monopolistic corporate platforms, guys, like Apple News.
  559. And so sometimes we will customize stuff
  560. to work on those other places as well,
  561. as you see here with our voter guide,
  562. which ran inside of Apple News.
  563. And then we have, I think the next slide
  564. will just tease one we have coming,
  565. we're gonna try to do a crazier year in review package
  566. that will be more visual and fun
  567. than what we typically done in the past.
  568. So those are the three types of things we do.
  569. The next section kind of gets into what that requires
  570. for a change, and some of this is a little more
  571. in my message for people eternally at the LA Times,
  572. but I think it'll work for you guys too,
  573. which is for this, for our data and graphics department
  574. to make these things and to achieve this mission,
  575. they have to be more independent and ambitious
  576. than maybe a traditional graphics department has been,
  577. and they need to have a higher skill level.
  578. So that requires us to seek out
  579. and develop our own ideas,
  580. to look at the major storylines of the day,
  581. be it COVID or COB or the Dodgers winning
  582. or the election or the George Floyd protests
  583. or homelessness here locally
  584. or whatever those major topics are
  585. that we know both matter to our democracy,
  586. but also to our audience in terms of what they want to pay for.
  587. We need to look at those topics and say,
  588. our job is to develop striking visual and data coverage
  589. that people are going to want,
  590. and that means we have to be more ambitious
  591. than some departments have been in the past.
  592. We need to take more initiative,
  593. and we need to operate more like a traditional news department
  594. in having our own sort of editing system and people in place.
  595. So that's kind of this section.
  596. And you can just jump through the next few slides
  597. and we'll get to where the next section,
  598. I just talk about the results.
  599. So just, you can jump ahead here a couple.
  600. There's Alex Tatuzzi and my deputy editor, my hero.
  601. And then the next slide is our listing.
  602. We have a job listing for a second deputy editor
  603. as we try to increase our editing muscle on our team,
  604. which is one of my major strategic goals
  605. for the current moment.
  606. And the reason we pursue all these goals
  607. is not just 'cause we're nerds and we like to do it,
  608. which is maybe a good enough reason
  609. if you're not expecting to get paid.
  610. But we also are pursuing this different type of journalism.
  611. One, because we like doing it.
  612. Two, because it's better journalism.
  613. The three, because it pays and it's better for our business.
  614. And so the next section is really kind of covers
  615. that these types of things I've been showing you,
  616. they're more popular with audiences.
  617. They're more valuable in that they are seen
  618. to convert more new subscribers to the digital product,
  619. which is our number one business goal than a typical story.
  620. And they're more durable in that they continue
  621. to draw an audience long after they're initially published,
  622. where most new stories kind of peter out within a day or two.
  623. So this next section is all redacted,
  624. so we don't need to look at it, but I can give you the gist.
  625. I mean, you probably aren't gonna be surprised
  626. to hear this if you think about it,
  627. but our coronavirus tracker,
  628. as humble as it may appear to you,
  629. is the most popular page in the history
  630. of the Los Angeles Times website.
  631. No page has ever been visited by more people.
  632. And if you compare it to our top traditional stories,
  633. it is multiple times more popular
  634. than the most popular stories of the year.
  635. And then when you look at conversions
  636. in terms of what's valuable in attracting new paying subscribers,
  637. it is an order of magnitude higher
  638. than the number two story of the entire year.
  639. And so you see that when things like this connect,
  640. they have a sort of home run possibility for the business
  641. that a traditional story very rarely can have.
  642. And then if you look at that over time,
  643. you'll see that it didn't become the top converting story
  644. until three or four months after it had initially launched,
  645. which just shows that it's continuing to draw audience
  646. and new paying customers over time
  647. and that sort of durability I was talking about.
  648. And so for those reasons,
  649. we think that there's a strong business logic
  650. behind investing in what we do.
  651. And I'll kind of close with this, I guess maybe,
  652. but having been a news nerd, as people say,
  653. and having done this kind of thing for nearly 20 years now,
  654. that has been the biggest change
  655. that I have seen in the last five or 10 years.
  656. When I began this as like a computer programmer journalist
  657. in 2000, we were curiosities.
  658. We were either the one person on the investigative team
  659. who had learned to code to try to do a cool data analysis
  660. for a traditional print investigative piece,
  661. or you were some weird person in the corner
  662. who said the internet's gonna be big one day,
  663. and we need to learn how to make new weird things.
  664. And so the justifications for most of the teams
  665. that I've been on in my entire career
  666. were either this is data analysis
  667. to do more impactful and important investigative reporting.
  668. That's actually what got me into it initially
  669. and is still my passion, right?
  670. Or it was this is sort of boutique internet experimentation
  671. to kind of see what comes of it, right?
  672. And we're now far enough into the digital transition
  673. to the information economy
  674. that that's just fundamentally changed.
  675. And what we've seen is that there are certain types
  676. of these products, these data and graphics products
  677. that are so successful
  678. that there's now like a capitalistic economic logic
  679. for producing them.
  680. And there's less and less of a logic
  681. for producing what these departments produced before.
  682. And so that's why I think not to get,
  683. I guess I am at the university
  684. so I can get all Marxist on you guys, you know,
  685. like I think that that is this economic logic
  686. that is driving kind of the evolution of our team
  687. and the justification for growing it,
  688. as well as similar changes you see in other departments
  689. like ours at other news organizations,
  690. some of which are like us in the middle of the transition
  691. and some others are further along down the road.
  692. Like for instance, the New York Times graphics desk
  693. which is already fully converted into this type of thing
  694. that I've been talking about here today.
  695. And I don't work there so I don't know the numbers
  696. but I would estimate based on what we see
  697. and based upon how the number of things they put out
  698. and the quality of what they put out
  699. that like the New York Times graphics department
  700. may be the most read and most profitable department
  701. in the whole newsroom, like that wouldn't surprise me
  702. with maybe the exception of one or two people
  703. who write about Trump, you know.
  704. And that is something that just was not true in the past
  705. when these departments were seen as supplementary, right?
  706. So that's my spiel.
  707. I'm happy to talk about really any of these issues
  708. that were raised, any of the stories in particular,
  709. any, you guys want to talk about Python?
  710. I'll talk D3, I'll talk about that.
  711. Whatever you guys want to talk about,
  712. I'm happy to just jam and hang out.
  713. - All right, well, that was fascinating.
  714. I do certainly want to open it up to the audience
  715. because this is about the students who are here
  716. participating this week.
  717. Maybe I'll have Andy kind of summarize the questions
  718. in the chat room, but in the meantime,
  719. well, you know, go ahead and post questions
  720. that you have for Ben in the chat room,
  721. but maybe I'll kick it off a little bit here, Ben.
  722. You really talk about this kind of transformative age
  723. of data journalism through this kind of visual evolution.
  724. What's just kind of like mind boggling to me
  725. is the fluidity by which you guys go about your business.
  726. Thinking like any one of the products
  727. that you have showcased in your presentation,
  728. if we were to do any one of those in academic institution,
  729. it'll probably take six months to create.
  730. And just kind of, you know, be followed by this,
  731. your ability to, you know, pivot so quickly
  732. with a new story.
  733. And then the next day you have like something
  734. that looks like it took six months to create.
  735. Can you talk a little bit about this kind of this environment
  736. that you work in, the speed that you have to work in?
  737. And you ever sleep?
  738. - Well, I'm glad you think that during COVID, very rarely,
  739. as you can tell by looking at me,
  740. but like after the selection,
  741. the dream is like December, like maybe a day off,
  742. but like,
  743. but more seriously, it's really like any,
  744. like, you know, I like the factory metaphor,
  745. but I'm sure there's other ways to talk about it,
  746. but it's like any process, any like manufacturing process,
  747. you break it down into pieces
  748. and then you like get really good at like, you know,
  749. getting each of those pieces of things
  750. that need to get done kind of like handled, you know,
  751. and you then try to speed up the process as much as you can.
  752. And so like in the case of like a visual story, for instance,
  753. you can think about it in that abstract manufacturing way
  754. in the sense that you need a toolkit
  755. that lets you stand up a single static page
  756. that's outside of like a traditional
  757. rigid content management system, right?
  758. That can be custom coded and designed.
  759. And so what you'll find is that at our department
  760. and every other one like it,
  761. that people have their little quote framework
  762. or their little, you know, publishing rig
  763. that lets them spin up a standalone web page.
  764. And then within that, there's like components
  765. of web development that make it easier for you
  766. to move quickly like a template inheritance.
  767. Well, within that system,
  768. we have a base template that has all the LA time styles
  769. and fonts and the header and footer
  770. and like all the basic pieces.
  771. And so then those are just like there for you
  772. when you like get the page like rolling, you know what I mean?
  773. And you're within that already.
  774. And then by the push of a button,
  775. you can trigger a little deploy
  776. that will send that like blank page live.
  777. And then within the framework, you, you know,
  778. depending on how you set it up,
  779. you have ways to import data files
  780. and then to quickly get them into JavaScript
  781. and other tools so that you can lay out the page
  782. kind of as quickly as possible.
  783. And our system isn't as streamlined as it could be
  784. or as it should be, you know,
  785. but like every, we've developed it ourselves internally
  786. and we're trying to make it better like all the time.
  787. Like this week we were talking about,
  788. well, how could we get it better at producing a headless page
  789. that's for iframing in other places?
  790. You know, how could we make a template
  791. that's just like ready to make an iframe page
  792. that could go on the homepage in 10 minutes, right?
  793. And so you have to have like that toolkit that's there.
  794. You have to have someone on staff
  795. who can like build and maintain such a toolkit
  796. because they don't really exist.
  797. One of my goals is for there to be
  798. a more standard open source sort of like thing for that.
  799. You know what I mean?
  800. You look at tool like when we make an application
  801. like the live election results or the tracker,
  802. we might use Django or some of these
  803. pretty highly developed web frameworks for databases.
  804. But when it comes to static pages,
  805. you really just have kind of like a hodgepodge
  806. of like node framework thingies
  807. that like can kind of halfway make a page
  808. but they don't have the 30 different things
  809. you actually need to make a real page
  810. that you want to publish, right?
  811. And like kind of, I think that there's the possibility
  812. for a stronger open source framework
  813. that goes beyond just the basic tools
  814. and gets into a little more of the opinionated decisions
  815. about how to manage deployment
  816. and static assets and data files and stuff.
  817. Wow, that was way too nerdy.
  818. But like that's part of it.
  819. That's like the technology part.
  820. And then there's the kind of editing and journalism part
  821. which is identifying either planned events
  822. that are ahead on the calendar
  823. that you know are coming like the election
  824. or pivoting when a major storyline emerges
  825. and then focusing your assignment on it.
  826. And so like in the case of like the Kobe thing,
  827. it's like Kobe's helicopter crashes on Sunday.
  828. In a breaking news situation like that,
  829. we tend to think in terms of like rounds of coverage
  830. or stages.
  831. So on day one, we're going to do this.
  832. On day two, we're going to do this.
  833. And then we're going to aim for something
  834. that's like a little more, you know what I mean?
  835. And so like in the case of Kobe,
  836. day one was just like, where is the crash?
  837. What kind of helicopter was it?
  838. You know what I mean?
  839. The basic facts and can you visually do that?
  840. Day two was, well let's get the play-by-play.
  841. What was the like?
  842. When did they take off?
  843. Where did they go?
  844. But you know, so we had a graphic
  845. with the day two story that had that stuff.
  846. Was there a route, a typical route?
  847. Was it an unusual route, you know?
  848. And then what happens typically in that is
  849. after you learn the basic facts,
  850. you kind of get to a point where you're like,
  851. okay, here's what we think the story is.
  852. You know what I mean?
  853. So like within these facts,
  854. here's what we think is worth sort of teasing out
  855. visually for people.
  856. So both has to be something insightful
  857. that goes beyond the basic facts for people,
  858. but it also has to have like a visual component to it, right?
  859. And so in the case of Kobe,
  860. our decision was to focus on the final turns of the helicopter.
  861. All this is of course debatable,
  862. but those were sort of our judgments we made
  863. like in those moments.
  864. Like the conception boat crash,
  865. the fire off Santa Barbara coast,
  866. I guess a year ago now,
  867. was another example of that where day one,
  868. it's just, well, where was the boat? What was the boats route?
  869. Day two, or there are like advanced,
  870. our piece for the weekend then was like,
  871. well, what were the escape routes on this boat?
  872. Where that was what seemed to us to be like the story.
  873. You're kind of betting that that's how it's gonna shake out.
  874. You know what I mean?
  875. As the facts come in.
  876. And then let's make a 3D model of the boats
  877. so we can show people what these escape routes were like
  878. and emphasize that they weren't very good.
  879. You know, it's what it ended up being.
  880. So that's like a breaking new situation.
  881. And then there's kind of in between ones
  882. where it's like there's gonna be this long storyline.
  883. And this is something that we're just trying to really
  884. get better at in our department this year.
  885. And is what I really admire about the New York Times
  886. and some of the more mature departments,
  887. is it seems to me that they have kind of decided on
  888. for the year, here's what their major storylines are gonna be.
  889. And then they come up with a string of visual coverage
  890. to try to attack those major storylines.
  891. And sometimes those can be planned like the election.
  892. And other times like COVID, they just occur
  893. and then you have to be smart pivot, right?
  894. And so for us, we've done that this year
  895. for the Floyd protests and for the COVID coverage
  896. where we put together our own independent visual coverage plan
  897. of like here's ideas.
  898. Everybody in the team pitches things in.
  899. We think about are there applications we could do?
  900. We think about are there visual stories we could do?
  901. And then we try to make them happen.
  902. - Yeah, that's wonderful.
  903. I just wanna jump in.
  904. There's a really, really great question.
  905. I think it can kind of build on what you were,
  906. I mean, what you just talked about.
  907. But the question is about just sort of thought process
  908. and developing stories and data,
  909. but specifically about dealing with biases and storytelling.
  910. And then the follow up to it is,
  911. are there any specific examples that you could share
  912. where you did face pushback on a story
  913. that is sort of data driven,
  914. but maybe there's some sort of part of it
  915. that kind of, you know, there was a stop to it.
  916. - Like internally at the LA Times, that's-
  917. - Yeah.
  918. - Well, you know, or maybe just from you.
  919. Maybe you kind of, yeah.
  920. - I think I have an example, it might not be exactly that.
  921. I will just say I've been here 13 years
  922. and you know, like any institution,
  923. people have disagreements and conflicts and struggles,
  924. you know what I mean?
  925. And I've had my own.
  926. But I will say that in that time,
  927. I have never once felt that if I had a good story
  928. that I knew was true, we couldn't get it published.
  929. That's never happened.
  930. There are people I really disagree with
  931. and to be honest with you, I don't like
  932. and they might not like me,
  933. but I have at different times been able to come to them
  934. with we have the story, check this out, let's do it.
  935. And those people, they rise above all the personal stuff
  936. and they're like, yeah, let's do it.
  937. You know what I mean?
  938. And to me, that's one of the great things about a newsroom
  939. is that people who disagree or who are different
  940. can unite around like a good story.
  941. And like I have nothing but good things to say
  942. about that personally, but like this is an example though
  943. and of something slightly different but related,
  944. which is how and one thing I really love about statistics
  945. is how it can sometimes they can sometimes lead you
  946. to conclusions that you do you would never imagine,
  947. you know what I mean?
  948. Or maybe we're counter to your initial bias
  949. and then you end up telling a different story
  950. than the one you thought you were gonna tell
  951. when it started out, right?
  952. And so you guys remember Hurricane Harvey in Houston.
  953. Oh, I have to send a Slack message really quick.
  954. I have to,
  955. this will just take a second.
  956. I have bought a mechanical keyboard
  957. during this like period at home guys
  958. and it is so loud when I type that this might be deafening
  959. but you guys remember Hurricane Harvey.
  960. This was in Houston a couple years ago.
  961. They had kind of like a big storm come through
  962. and kind of flood the whole place
  963. and it led to these sort of crazy images on television
  964. of people trying to cross the city
  965. in this sort of biblical flood kind of situation.
  966. And it was a Friday night and we were sitting
  967. in the newsroom at the LA Times
  968. and an editor said to me, well, we gotta do something.
  969. Let's come up with something.
  970. It was one of these moments, right?
  971. Like what are we gonna do?
  972. This nuts thing is happening.
  973. We have to come up with a way to cover it.
  974. And like a lot of those situations,
  975. you kind of start with what you're seeing in the news,
  976. what you hear from other people and your own biases.
  977. You know what I mean?
  978. Like what you kind of think about the world.
  979. And like, and then we're trying to come up
  980. with a data story.
  981. And so our initial thought was, hey,
  982. if we get the data about where the FEMA flood zones are,
  983. these officially defined flood zones from the government,
  984. right?
  985. And we compare it to the early damage data
  986. that we see coming in from the government
  987. about where buildings are damaged.
  988. We're gonna find that all the people in these zones
  989. got wiped out and we're gonna be able to add them up
  990. into some big number and try to do a story
  991. about the like extent of the damage, right?
  992. And like, we didn't know exactly what it was gonna be,
  993. but that was kind of our methodological bias
  994. or frame that we like came to it with, right?
  995. Let me just send this message, sorry.
  996. And we're all fired up to do that.
  997. I stayed up to like one in the morning, like it's like,
  998. oh my gosh, the Harris County Property Assessors Office
  999. has all the data online.
  1000. Yeah, grab it.
  1001. And the FEMA is having a conference call
  1002. and they're gonna give out the damage data.
  1003. Wow, there's property values.
  1004. Here's the flood zones.
  1005. I'm gonna put them all together and like find it out.
  1006. And like, and then we pitched that story to an editor
  1007. and we say, this is what story is gonna be.
  1008. It's gonna be about how screwed up Houston got
  1009. 'cause all this.
  1010. And then I did the analysis
  1011. and what I quickly found, or not quickly, not so quickly,
  1012. what I soon found was that most of the damage
  1013. was actually outside of the official FEMA flood zones, right?
  1014. So that was an interesting thing.
  1015. And then we're like, okay, so let's start calling people
  1016. and let's figure out what's going on.
  1017. So we're just getting on the phone calling, texting
  1018. 'cause we have all these addresses
  1019. from the property assessors.
  1020. Just call every person and just see what's going on.
  1021. And we did dozens and dozens of phone calls.
  1022. And what we found is that people inside
  1023. the official FEMA risk zones were actually like fine.
  1024. Like every one of them we could try to was like,
  1025. no, I'm good.
  1026. We were high.
  1027. We had put up like, we had elevated our house
  1028. or we had put in a retaining pond
  1029. and that caught all the water or our roads got redone.
  1030. And it was the people outside who had the problems.
  1031. And what we kind of pieced together then
  1032. from talking to experts and more and more people
  1033. was that our initial assumption was actually reverse.
  1034. And that's because believe it or not,
  1035. public policy works guys or it can.
  1036. It's that if you're inside an official FEMA flood zone,
  1037. there were all these additional building requirements
  1038. and retrofitting requirements for like taking care
  1039. and getting ready.
  1040. And so the people who lived in the quote unquote
  1041. riskiest areas actually ended up having the best outcomes
  1042. because they had to take all these preventive measures.
  1043. And so we ended up writing a story about that,
  1044. about how, hey, these regulations actually work,
  1045. but they're not nearly widespread enough
  1046. to save all these other people, you know what I mean?
  1047. It's kind of what the story ended up being.
  1048. And we had to kind of go back to the editor
  1049. and be like, oh, our initial pitch, it's like not that at all.
  1050. And that was this, it's a little hard to summarize,
  1051. but that was this sort of like weird process
  1052. of gathering numbers, listening to them,
  1053. talking to people on the other end of the data
  1054. and gradually having like a light bulb go on like,
  1055. oh, what everything I think is wrong.
  1056. And then you have, there's, whenever that happens,
  1057. there's always this like fun meeting
  1058. like in the middle of it where like you get together
  1059. and you're like, hey, we're totally wrong, right?
  1060. (laughing)
  1061. And you have to, okay, well, we can't just give up.
  1062. We gotta come up with like some story.
  1063. And so then oftentimes it's a more interesting story,
  1064. you know what I mean?
  1065. Because it's something you didn't expect, you know?
  1066. - Yeah, that's, yeah, that is actually amazing.
  1067. And it's really, I think it's just sort of really helpful
  1068. to kind of get that perspective on how you develop
  1069. and really work through these stories.
  1070. Another question, and I know we wanna be really sensitive
  1071. of having you here for too, too long,
  1072. but we have a couple of really good questions.
  1073. One, I think that would maybe really also be helpful
  1074. for students is just sort of thinking about
  1075. that tipping point between when you're creating a story,
  1076. when is it interactive and when is it static
  1077. and kind of like, how does that decision process get made
  1078. and how does that sort of, how does that happen?
  1079. - That's a tough one, and there's not one answer,
  1080. but I mean, if the question is when is something interactive?
  1081. Look, if you ask it that way,
  1082. the answer is less and less so every month and year.
  1083. And that has a lot to do with the changes we've seen
  1084. in consumer technology over the last five or 10 years.
  1085. As more and more people have moved to mobile phones.
  1086. Believe it or not, you probably will believe it.
  1087. You guys are into tech.
  1088. The majority of people who read the LA Times online
  1089. read it on a phone, like well more than 50%.
  1090. And so in a circumstance where the majority
  1091. of your audience is visiting you on a very small screen,
  1092. which they mostly just like thumb up and down on, right?
  1093. A lot of the interactive graphics that were very intricate
  1094. and made in Adobe Flash or whatever, 10 years ago,
  1095. really no longer makes sense because most people
  1096. aren't gonna interact with them and use them.
  1097. And the reality is, is what we learned from web analytics,
  1098. even before the mobile changes,
  1099. that most people never interacted with them anyway.
  1100. And so that's why you've seen a shift in,
  1101. if you follow kind of our little niche,
  1102. you've seen a shift away from most interactive graphics.
  1103. And so for the most part, we're looking for graphics
  1104. that can work static and small on mobile
  1105. and the way to make them more insightful
  1106. is usually through adding an annotation layer,
  1107. so like a little swoopy arrow or something.
  1108. Or a scrolling change, like you see
  1109. in that scrolling technique or the chart sort of changes
  1110. as the user continues to read the page.
  1111. Where interactivity still makes a lot of sense,
  1112. I think is in cases of personalization or of discovery.
  1113. So we're gonna have a chart that shows you
  1114. the total number of deaths at nursing homes from COVID.
  1115. But then when we have the, we're gonna have a table
  1116. that lets you look up the nursing home and care about, right?
  1117. And then that table should have like a search
  1118. and a filter and a way for you to like dig into it.
  1119. And then there's something like the live wildfires map,
  1120. which is really about one, giving you an overview
  1121. of like how many fires are there right now.
  1122. But then two,
  1123. two, letting you drill down into areas that you care about
  1124. to see kind of the latest data within those areas.
  1125. And so that's the map being there for personalization
  1126. and discovery, you know.
  1127. I think a lot of the interactivity that was just
  1128. like about fiddling, you've probably noticed
  1129. has like mostly disappeared.
  1130. Plus it's a lot, it takes more time to code.
  1131. - Yeah, very cool.
  1132. You know, we definitely wanna ask you this one question
  1133. for sure, because we have so many students here
  1134. who are interested in data and storytelling.
  1135. And sort of what is like, if you had one piece of advice
  1136. for students going out, working on school projects,
  1137. or really just wanting to break into
  1138. Ellie Times newsroom or other places, what is it?
  1139. What should they know?
  1140. - One piece of advice, I mean, I don't know.
  1141. - Or two or three.
  1142. - I can give you a call.
  1143. - Yeah, please.
  1144. - I mean, to me, this is a thing that has become cliche
  1145. and is said by some people who I think are kind of jerks,
  1146. but if you race them from your mind, it is true.
  1147. You gotta learn to code.
  1148. You don't have any, it's just like to really
  1149. do this well, you have to learn to code.
  1150. And you have to learn to code certain types of things,
  1151. not just like coding in general,
  1152. but I do think the fundamentals of programming
  1153. are so important.
  1154. The fundamentals of just like software and package design
  1155. are really underrated, just like how do I write a function
  1156. and then import it into another file and then reuse it
  1157. and like package code.
  1158. Like the basics of coding, I think are super important.
  1159. And then the sort of applied application of coding
  1160. to the types of things we do.
  1161. And so that, I think on the one hand,
  1162. is kind of data gathering and analysis.
  1163. So this is, how do I get data by scraping it
  1164. or reading it out of a database or building my own database?
  1165. And then how do I analyze it to like interview the data
  1166. is how we think about it.
  1167. What questions do I have for the data
  1168. and how can I write code that will help me get the answers
  1169. to those questions?
  1170. And so the tool that I teach a lot of classes on
  1171. that I like, but it's all debatable
  1172. is the Jupyter Notebook with Python data analysis tools.
  1173. They're free is kind of the best thing about it.
  1174. And the Jupyter environment once you figure out
  1175. how to get it running is just a really handy way
  1176. to do like step by step kind of data work.
  1177. And then on the other side,
  1178. there's the sort of how do we tell stories visually
  1179. through web development, right?
  1180. And that really involves learning fundamentals of HTML and CSS.
  1181. You know, again, just the basics are so important.
  1182. They're not even in HTML's case.
  1183. It's not even that hard, right?
  1184. And then how do you use JavaScript to make interactivity
  1185. and data visualizations?
  1186. I'm a big fan of D3.
  1187. And it's also a great way to get better at JavaScript, too,
  1188. because the D3 culture is really built around excellence
  1189. and programming, which I think is important.
  1190. I think there's a lot of free tools that are just about
  1191. cutting corners that don't really help you
  1192. over the long run become to master that type of thing.
  1193. So to me, there's the coding part,
  1194. but then there's the journalism part, which is thing two.
  1195. And this is where, like when I get applicants for internships
  1196. and stuff from data science programs,
  1197. which I'm always glad to get and I love to see,
  1198. you know, sometimes the people just haven't had the time
  1199. or opportunity to really apply their work in a way
  1200. that they're making something for the general public
  1201. or that has kind of a headline take away like point, you know?
  1202. A big part of editing and putting out a piece of journalism
  1203. is you kind of have to have like something to say.
  1204. And just writing a notebook that asks 10 questions
  1205. of the data or just making a graphic
  1206. that lets you fiddle with the numbers isn't enough, right?
  1207. The things that you're making need to kind of take
  1208. that next level of kind of journalistic whatever, you know,
  1209. by saying my finding is this.
  1210. And here's the headline that says it
  1211. and here's the three charts that show it, you know what I mean?
  1212. And you kind of like have a takeaway and a point
  1213. so that you have something to share.
  1214. And so when I see applicants who,
  1215. and I don't expect anyone who's being out
  1216. to really master any of this,
  1217. but when I see people who kind of have clearly put in the effort
  1218. to do both of those two things,
  1219. to like learn how to do the coding stuff to make stuff,
  1220. but also learn how to boil it down
  1221. and to make something out of it
  1222. for regular people, you know what I mean?
  1223. That's when I think you're a strong applicant
  1224. and you're sort of beginning the road to become do what we do.
  1225. But the reality is is learning how to code takes a while
  1226. and the things we're learning,
  1227. we're trying to do are emerging and experimental.
  1228. And so no one including myself really gets kind of good
  1229. at what we're doing until they've spent a couple of years
  1230. really apprenticing at it.
  1231. And that's just, that's just being human, you know?
  1232. - That's great.
  1233. That's really humbling.
  1234. And I think really honest to share that.
  1235. And I hope all the students out there
  1236. really get a lot out of that.
  1237. That's really helpful.
  1238. You know, I don't know if you wanted to kind of
  1239. have a wrap up question or...
  1240. - Yeah, whatever you guys got,
  1241. I'll take, I got a couple of minutes yet.
  1242. - Yeah, okay.
  1243. Well, well, first of all, thank you for kind of validating
  1244. this week for us from what you just said,
  1245. the value of coding, open data.
  1246. This is all that we've been kind of preaching
  1247. in the last couple of days.
  1248. Also should mention that if I look up
  1249. visual storyteller in the dictionary,
  1250. I think I'll find you in there.
  1251. You seem to embody that kind of visual storyteller mentality.
  1252. But I don't know if there's like one or two final questions.
  1253. Maybe somebody can unmute themselves
  1254. if you're out there and introduce yourself to Ben
  1255. and ask a question, hoping somebody can do that.
  1256. - I have a quick question.
  1257. - Please.
  1258. - And that's about, Ben, I see that you're active
  1259. in an organization called Data for Progress.
  1260. And I'm wondering if you'd talk about that a little bit.
  1261. It sort of ties in with what we've been doing
  1262. with humanitarian mapping.
  1263. And sort of how you see that kind of mission
  1264. converging with what you're doing at the LA Times,
  1265. which is obviously more of a commercial than--
  1266. - Yeah.
  1267. I don't know if I'm familiar with Data for Progress,
  1268. but I'm guessing it came up linked
  1269. to some other things that I do, you know what I mean?
  1270. But like for me personally, open source software
  1271. is probably as close as I get to religion.
  1272. Nice, but like, and I think is a really underrated
  1273. as a solution to a lot of problems.
  1274. It just needs more institutional support
  1275. from places like UCLA.
  1276. But what you guys are providing today, which is great,
  1277. you know, but like, and so I've been involved
  1278. in a number of kind of open source data gathering
  1279. and cleaning efforts related to helping journalists
  1280. both improve their skills, but also to kind of
  1281. build software and data infrastructure.
  1282. So we're not all like replicating each other's work
  1283. and we can benefit from kind of group collaboration.
  1284. One of those is called the California Civic Data Coalition.
  1285. And so this was a foundation funded open source
  1286. software project that refines and cleans up
  1287. the campaign finance data put up
  1288. by the California state government.
  1289. So this is the money in our state politics,
  1290. like all these Uber ads you're seeing
  1291. for Prop 22 right now.
  1292. And the data is there, it's just really difficult to access
  1293. because it's just in crappy shape.
  1294. And so there's kind of like this problem of who's going
  1295. to clean it up, right?
  1296. And so we created an open source group
  1297. to write the software that would be the refinery
  1298. for that like raw data, and then all the code
  1299. and all the effort is open source as well as the result.
  1300. Along the way we invented some bulk data loading
  1301. open source software tools, which people now use
  1302. for like totally different purposes, you know what I mean?
  1303. There's like side benefits to doing that.
  1304. And we're replicating that model currently
  1305. when it comes to tracking COVID data,
  1306. where initially the LA Times was doing 100%
  1307. of this COVID tracking ourselves in California,
  1308. but we now have a team of people from eight newsrooms
  1309. across the state who are helping us kind of gather it
  1310. and consolidate this cleaned up database of COVID data.
  1311. And I think that there's one, a lot of potential for that,
  1312. but two, I think as news organizations get smaller
  1313. and struggle, I don't know if there's a better alternative,
  1314. like who, you know what I mean?
  1315. Other than we just hope the New York Times gets big enough,
  1316. they can do it for everybody, you know what I mean?
  1317. Like that's really, the commercial solution
  1318. is going to be news monopoly, you know,
  1319. that's just big enough to do that is what I think.
  1320. But, you know, maybe I'm being too pessimistic.
  1321. So I think we need to band together
  1322. to come up with this alternative.
  1323. - Thank you.
  1324. Great, how about one last question for Ben?
  1325. How about one of the questions?
  1326. - I have a question.
  1327. - There we go, please.
  1328. - So I love your neighborhood map,
  1329. which is, I know, which has been on the page
  1330. for a while now, it's so great.
  1331. And I use it in my classes to talk to instruct students,
  1332. well, instruct students about two things,
  1333. one about the concept of place
  1334. and another about the concept
  1335. of volunteer geographic information.
  1336. And I'm curious if you guys have done other projects
  1337. that you think are within the realm
  1338. of like volunteer geographic information.
  1339. So like people actually map like-
  1340. - Yep, that project is one of my great white whales
  1341. and that we haven't really revived it.
  1342. There was a, the plan was to do it last year
  1343. for this census, but other things have come up.
  1344. And just so people know,
  1345. there is no official source of neighborhoods in LA County,
  1346. even though we all feel that we live in one.
  1347. And so about 10 years ago with some of my colleagues,
  1348. we sort of tried to pull the public
  1349. and then divide up LA County into, you know,
  1350. the LA Times neighborhoods at least.
  1351. And then we connected those kind of like
  1352. with the legos of census tracts.
  1353. So then all the areas have all this metadata
  1354. about demographics that are then merged with them,
  1355. which has allowed us to do dozens of investigative
  1356. and analytical stories to compare East LA to West LA,
  1357. you know what I mean?
  1358. It's not more in social science.
  1359. And one of my personal ambitions is for our team
  1360. to be more of a social science data provider.
  1361. We're trying to do that a little bit here with COVID,
  1362. but I would love to sort of be able to publish
  1363. sort of like the machine readable atlas
  1364. of LA demographics by neighborhood or something
  1365. so that we could see more and more people
  1366. able to do that type of analytical work.
  1367. But at the core of it was this crowdsourcing effort
  1368. 'cause we didn't want to do it entirely alone.
  1369. So we sort of put out our own version of the maps
  1370. and then we had people tell us how wrong we were
  1371. and we modified them and you can't please everybody.
  1372. There's definitely flaws,
  1373. but that was kind of a collaborative process.
  1374. You know, that's the type of thing
  1375. that we haven't turned into like it.
  1376. We're going to do something like that every month,
  1377. but there's a few examples through the years
  1378. that I've had a lot of fun working on.
  1379. Like we finally, we're going to do the like a few years ago,
  1380. the like, what is the East Side story in the LA Times?
  1381. You know what I mean?
  1382. Where everybody brings out their grievances, you know,
  1383. and I will tell you as long as I am the data
  1384. and graphics editor, the East Side begins at the LA River.
  1385. I'm sorry, everybody who lives on Sunset,
  1386. you don't live on the East Side, sorry, you know what I mean?
  1387. But like, but we know people disagree, you know what I mean?
  1388. Like there is no answer.
  1389. And so we did like a one-off that was sort of
  1390. in that same tradition where it was,
  1391. here's a map of LA, draw the East Side, right?
  1392. And we had hundreds of people like draw them all.
  1393. And then we made this composite map
  1394. that was like everybody's East Side like overlaid.
  1395. And it was a little abstract and already, you know what I mean?
  1396. I don't know if it was like the clearest data visualization
  1397. ever drawn, but it sort of had this fun effort of like,
  1398. here's the splatter of like what everybody thinks, you know?
  1399. And though I'm sure our readership is biased
  1400. in a certain direction.
  1401. But so like that was an example of that, I guess.
  1402. - Yeah.
  1403. - You know, to me, the crowdsourcing thing that's out there
  1404. that nobody does much with Viz,
  1405. which we could do with Quakebot,
  1406. is the like, did you feel an earthquake data from like USGS?
  1407. I think it's like all of there for every earthquake
  1408. and like nobody really does much with it, you know?
  1409. I know the scientists at USGS do,
  1410. but like, I don't know if the news media does.
  1411. Yeah.
  1412. - Right.
  1413. Cool.
  1414. Thank you.
  1415. - All right.
  1416. Yeah, no, there's a whole bunch of chatter
  1417. about that project you're talking about mapping Los Angeles.
  1418. - Yeah.
  1419. - It's many of us fondly remember that kind of,
  1420. hey, we can contribute to defining boundaries,
  1421. which is not something we're used to.
  1422. Boundaries are usually defined by somebody up there.
  1423. And we abide by those boundaries.
  1424. And here's the LA Times saying,
  1425. hey, you define your own boundaries.
  1426. So that was a really neat project.
  1427. And still to this day in the urban planning department,
  1428. that's a go to resource for doing neighborhood level analysis.
  1429. - Yeah.
  1430. I was absolutely just going to say that.
  1431. I mean, I think the impact of that project in the classroom,
  1432. I just don't know if that kind of reaches you all,
  1433. but like, I mean, that is in so many different classes,
  1434. so many different students use that as sort of like,
  1435. understanding LA at this point.
  1436. - Yes, we have census tracts.
  1437. Yes, we have these other boundaries,
  1438. but those particular boundaries in that work
  1439. is just super, super influential in the classroom.
  1440. - Yeah.
  1441. I mean, to me, I just see all the times we fail to revive it.
  1442. So I feel a little guilty with that.
  1443. But like, you know, we kind of have, I have some ideas.
  1444. I might be like, this is one of these things
  1445. where like internally it gets caught up in the politics of like,
  1446. do we need a hyper local neighborhood news product?
  1447. You know what I mean?
  1448. And like, it's hard to have a conversation internally
  1449. without it being caught up with like traditional coverage.
  1450. It's my belief that it should be more of a data application
  1451. and like kind of like a social science hub
  1452. that is like easy enough for the average person
  1453. to understand, you know what I mean?
  1454. That's just like what is, you know,
  1455. Chevy bills or whatever, you know?
  1456. And I would like to kind of narrow kind of the mission
  1457. of it to really focus on that kind of popularizing
  1458. the demographics thing.
  1459. And so my hope is, is that we can do that
  1460. when the new census data comes out.
  1461. But as you know, it's the ACS is where the action really is.
  1462. So it doesn't matter.
  1463. I kind of wonder if we could partner that with like,
  1464. can I give you my pet idea?
  1465. And you guys can tell me if it's terrible.
  1466. With like a gentrification analysis.
  1467. - Oh, yeah.
  1468. - If you like that, yeah.
  1469. 'Cause I did some reading
  1470. and you guys may know this better than me,
  1471. but there's these like academic definitions
  1472. of gentrification, which as far as I can tell,
  1473. are kind of limited to a sort of post-war redevelopment
  1474. sort of framework of like, how old is the housing stock?
  1475. And how recently has it been replenished?
  1476. And what is the change in housing value been, right?
  1477. And so like, you could, I look at those methods
  1478. and I could take these longitudinal census databases.
  1479. I can do that for every census tract
  1480. and then I could tell you what's the most gentrified neighborhood
  1481. in LA like that way, right?
  1482. That would be one way.
  1483. But like my hunch is that that's not enough.
  1484. Like in our 21st century life,
  1485. there's sort of this Brooklynized idea of gentrification,
  1486. which has to do with cultural products, right?
  1487. And commerce and like that kind of thing, you know what I mean?
  1488. Like what is the business in Highland Park now
  1489. on that strip on Fig versus 10 years ago, right?
  1490. Used to be a Latino typewriter shop
  1491. and now it's a hipster cocktail bar or whatever, you know?
  1492. And I think that's what people have in mind
  1493. when they send you an interpretation today.
  1494. And so like what I'm kind of reaching for
  1495. is could you come up with a metric
  1496. that would try to capture that
  1497. and then integrate it with the older metrics
  1498. into like a kind of more up-to-date gentrification index
  1499. or something, do anybody have any ideas
  1500. or have read anything that smiles ahead of me on this?
  1501. - Well, you know, there's a Center for Neighborhood Knowledge
  1502. here at UCLA that does a lot of research.
  1503. I don't know if you know Paul Wong,
  1504. he's a professor who--
  1505. - I don't know. - Wong worked with models
  1506. of gentrification and they have a site,
  1507. a project called Urban Displacement,
  1508. collaboration with Berkeley and UCLA
  1509. where they have index S for gentrification.
  1510. - No, really?
  1511. - It just talks about the need for us, you know,
  1512. in academia and also to let you know,
  1513. like UCLA is launching this kind of data acts initiative.
  1514. So it sounds like we are,
  1515. once that behind you guys, but we're right there
  1516. in terms of our efforts to think about,
  1517. like you said, you know, the importance of coding
  1518. in kind of the more social sciences,
  1519. which isn't traditionally been the case.
  1520. So just to think about how us in academia,
  1521. I mean, I feel such a stronger bond with you
  1522. through what we've been discussing today
  1523. and the importance that we work together
  1524. to create some meaningful stories
  1525. that can inform policy decisions across our communities.
  1526. - Yeah, I also love, I really love the idea
  1527. of kind of looking at sort of alternative data sources.
  1528. I was on a call yesterday and a workshop yesterday
  1529. actually looking at the Getty's new Riche project.
  1530. And they're putting all these images out there.
  1531. And this workshop was really trying to understand
  1532. how to look at those images, you know,
  1533. in this, you know, cultural production,
  1534. the streets that were photographed,
  1535. historically looking at, you know, racial change over time
  1536. and really talking about gentrification
  1537. and how a data set like that with those images
  1538. could be used with, you know, all these sources
  1539. that you're talking about, but other data sets
  1540. that are cultural and kind of trying to, you know,
  1541. look at that and also how to sort of code those images
  1542. and doing that coding so that it actually is, you know,
  1543. doing the work that we want it to do
  1544. in terms of understanding, you know, the displacement
  1545. or whatever's happening.
  1546. And not just sort of saying like, oh, you know,
  1547. this was here before and now it's gone,
  1548. but like, well, why is it gone?
  1549. And really kind of, you know, putting the point onto that.
  1550. So yeah, it's really interesting.
  1551. And I think you're right.
  1552. I mean, I think what we've been talking about
  1553. with, you know, really learning how to, you know,
  1554. build coding and, you know, also just stepping back
  1555. and being able to kind of piece, I think,
  1556. like really different data sets.
  1557. And the thing that we were talking about yesterday too
  1558. is just sort of understanding that world of data
  1559. and kind of, especially I think for things
  1560. like gentrification, kind of stepping out
  1561. and kind of looking at or trying to understand it
  1562. in a way that's local and trying to find sources
  1563. that maybe aren't that typical, you know, census,
  1564. you know, way of understanding it.
  1565. So I don't know, I love it.
  1566. Yeah.
  1567. - There's a guy at Harvard I found he did a paper read
  1568. used Yelp data and then he tried to like correlate
  1569. the arrival of certain types of businesses
  1570. with other gentrification indicators, you know,
  1571. like educated population or whatever.
  1572. And so there might be something there, you know,
  1573. but I kind of imagine the index.
  1574. You take like the traditional metric of like buildings
  1575. or whatever, you know, you add in the racial dynamic
  1576. of, you know, white invasion, as I've learned,
  1577. it's called in some papers.
  1578. And then you maybe add like a third cultural dimension
  1579. of these products and then you like mix those into like
  1580. a radar chart or like an index or something
  1581. and that'd be kind of cool.
  1582. My bet is what Adams would be number one.
  1583. That's my bet guys.
  1584. - I would say from personal experience
  1585. with Brooklyn and here in LA, the first major indicator
  1586. should be when the real estate agent decides
  1587. to change the name of the name.
  1588. - There you go.
  1589. That's another one.
  1590. - That's an indicator.
  1591. - One thing I learned looking at the literature too,
  1592. I thought was interesting is traditionally only like
  1593. the lowest quartile could ever be thought of
  1594. as having gentrified.
  1595. So you had to be depressed to ever undergo the process.
  1596. And I don't think that people think about it that way
  1597. today anymore.
  1598. Like, 'cause when I ran the traditional numbers,
  1599. I saw that Venice Beach had a really high score
  1600. on some of the values but wouldn't have been considered
  1601. because it was already a pretty wealthy place to begin with.
  1602. Right?
  1603. And so like there's this way in which I think
  1604. the traditional numeric framework doesn't mesh
  1605. with how people really think about it
  1606. or talk about it today.
  1607. And like, that's the thing I just feel like
  1608. there must be some way to wrestle that.
  1609. But.
  1610. - That could be a,
  1611. and this is a very geographically nerdy
  1612. but a modifiable aerial unit problem issue
  1613. where because you're looking at Venice as a whole.
  1614. - Right, we've predefined it.
  1615. - Yeah.
  1616. - Very wealthy parts get, you know,
  1617. averaged with the less wealthy parts.
  1618. So those.
  1619. - That's a good point.
  1620. - Yeah.
  1621. - Mm.
  1622. - But I'm thinking also that with the sort of combining
  1623. what Andy was talking about with the coding
  1624. and like maybe taking for each neighborhood,
  1625. choosing kind of like the commercial strip
  1626. and looking at Google Street View
  1627. and sort of like coding that over time
  1628. for the main, like the main drags of different neighborhoods.
  1629. - Yeah.
  1630. Or even if we did the number for the whole county,
  1631. you could pick the place that had the highest
  1632. and up new merit score and then go in
  1633. and do that closer study on that place.
  1634. - Yeah.
  1635. - You know, and then you wouldn't have to do everywhere.
  1636. - Right.
  1637. - I also have to think about how to do less.
  1638. Sorry.
  1639. (laughing)
  1640. - Now, I mean, I think that's one of the things
  1641. that we all struggle to is sort of kind of, you know,
  1642. looking at LA, but I think a lot of us push our students
  1643. and just people that we work with to focus in on areas.
  1644. And I think that's sort of like a nice bridge to,
  1645. you know, I think what you at the times do
  1646. is kind of putting in the person into our data stories.
  1647. And like, you know, not always sort of like having that
  1648. bird's eye view of everybody, but like, let's zoom in.
  1649. Let's like really kind of understand what this data set
  1650. is doing on this street, on this block.
  1651. - Yep.
  1652. We talk about them and like, there's usually,
  1653. there's like two different couples we're looking for.
  1654. There's like, there's either Mr. and Mrs. Outlier
  1655. or there's Mr. and Mrs. Central Tendency.
  1656. And like kind of depending on like which story
  1657. you're telling, you're sort of like aiming
  1658. for those people, you know.
  1659. - That's great.
  1660. I love that.
  1661. - I feel like we're in a newsroom right now
  1662. making decisions for the next story.
  1663. But I wanna wrap us up here,
  1664. taking too much of your time.
  1665. Thank you so much.
  1666. The conversations in the chat room has been just nonstop.
  1667. You've touched a lot of topics that really mean
  1668. so much to many of us.
  1669. So just wanna say a big thank you to Ben.
  1670. I'm sure it's shared by all of us here.
  1671. There you go.
  1672. And Ben, let's definitely get together again.
  1673. When there's a little bit more time to relax.
  1674. I don't know when that would be.
  1675. But I'd love to kind of have a conversation
  1676. about how we in academia can work closer together
  1677. with you guys at the LA Times.
  1678. - Great.
  1679. - Yeah.
  1680. - For having me.
  1681. - Thank you.
  1682. Thank you for being so generous with your time.
  1683. It's really amazing.
  1684. - Mm-hmm.
  1685. Thank you so much.
  1686. - Sorry, I was late.
  1687. - I love the new luck.
  1688. I love the-
  1689. (laughing)
  1690. - I'm gonna report that to my wife.
  1691. - All right, thank you so much, Ben.
  1692. - Thank you, Ben.
  1693. - Mm-hmm.
  1694. - Bye.

Downloads

Recording video · Timestamped transcript