good eveningor afternoon everyone depending on onwhat time zone you’re inmy name is kath madison i’m actingexecutive director of thetech talent project and i’m so glad thatyou’re with us tonight to celebratethe work of the inaugural class of thetech executiveleadership initiative so the teleprogramcame together this year as acollaboration between the aspen techpolicy hub project redesign and the techtalent projectthe goal was an eight-week skillbuilding initiativeto prepare experienced technologyleaders to engage effectively withpublic sector challenges andessentially tele program participantsmet once a week for eight weeksbuilding skills through real-worldengagement with government policy andhuman-centered designand through this experiential learningmodelthey tackled real world service deliverychallenges to hone their skillsand abilities to work effectively withdedicated public servants to mitigaterisks ontechnology projects and tonight we’ll beintroducing the inaugural cohort to youand showcasing some of their work sounlike most of the team who pulledthe tele program together my backgroundis statenot federal technology implementationand while the scale may be different thechallenges facing government and itsabilityto leverage technology effectively areare the sameso i still vividly remember the nightbefore vermont’s little health insuranceexchangelaunch launched and you know those of uswho were in the weedsknew enough to know that it was going togo poorly despite the publicassurances that were being made and iremember getting justa few hours of sleep on a couch in theoffice as we awaited the final go no-godecisionin the middle of that night i alsovividly remember the disaster thatunfolded in the months after that launchand it was like nothing i had everexperienced before inin my professional career and at firstour workto right the ship was singularly focusedwe wanted to bring the site back up andhelp keep applications from timing outbut really our problems just quicklyescalated from thereballooning into a backlog of tens ofthousands of vermonterswho were just trying to get theirapplications through or to makecritical updates to their applicationinformationand during that time i would say thepublic pressure was pretty relentlessand rightfully so you know we had failedto be good stewardsof public need and taxpayer dollarsand all the while you know customerswere angry and staff were in tearsand as a small rural state in vermontwith relatively few resources andleverage it really took us years to digout of that situationand you know while i wish thatexperience was uniquesadly it’s not um you know many of youin the audience todayexperience healthcare.gov challengesfirsthandand we see it still almost 10 yearslater with state websites crashingas people tried to access unemploymentbenefits during the pandemicwith the small business loan program andthe effort to get out financialassistance to individualswhen government fails to delivertechnology effectivelyit’s costly and inefficiently at bestbut at its worst it can do real harmbut this isn’t the only story that wecan tellwe can tell the story of thousands ofcareer public servants whoare true heroes who work tirelessly tohelp those in need using outdatedsystems and broken business processeswe can tell the story of what a smallteam of technologistswere brought in to rescue healthcare.govwere able to doonce they were able to take onleadership roles to foster opencommunicationand encourage really a badgelessenvironment where collaborationand communication between teams wasvaluedand we can tell the story of hundreds oftechnologists who’veleft their successful careers in publici mean in private sectorto join organizations like the unitedstates digital servicethe presidential innovation fellowsprogram and 18f to applytheir unique skills to serve the countryand sothere are really two big lessons herefor menumber one is that in the years sincehealthcare.govthere are just as many bright spots outthere as failures and i think part offiguringout how to move forward successfullydepends on what we choose to focus onandchoose to amplify and number two youknow the team that you havemakes all the difference and when ireflect on the tele program that’sthat’s why it’s so focused on peopleworking with some of thebest and brightest tech and techadjacent leaders in the country toreally inspire them to serve and toprovide them with the knowledge skillsand ability to succeedin the public sector this is a loftygoaland given that this was our first classwe have improvements to makebut i would say that the the work thatwas done and the relationships that werebuilt over this eight week period ornothing short ofof inspirational and i’m so excited thatyou’ll get the chanceto hear about some of this work tonightbefore i turn it overto my sister who will talk a little bitmore about the projects and introducethe team i just want tothank a few people so first thank you tothe incredible stewards of ourcurriculumdana chisnell and betsy cooper ourexecutive teamincluding jenny hunt and jenniferanistof the hubs deputy director maisislaour program of court program coordinatormeja lualiaand our fellow haley ponte to the fordfoundation schmidt futuresand the beck center for social impactand innovation thank you for yourgenerous supportthank you to the aspen tech policyhubfor hosting the program and to all theguest speakers who joined us to shareyour wisdom and experienceand most of all thank you to the teleinaugural classboth for being a part of this experimentand for your time energy and commitmentto public servicewith that i will turn it over to mythanks so much cass for for thoseopening remarks describing the impetusbehind the telly programand good evening or late afternoon toeveryone else my name is mai sisla andi’m the deputy director of the aspentech policy hubfor those of you who are new to us thehub is a west coast policy incubatortraining tech policy entrepreneurs weregularly run in-person fellowshipprograms in the bay area to traintech experts on how to engage in thepolicy process and we’re eager topartner with the tech talent project andproject redesignto also build a program that targetedsenior level executives who couldparticipate in a program like this whilecontinuing their full-time jobthus the tech executive leadershipinitiative waswas born um this evening we’ll beintroducing our inaugural cohort ofleadersthese leaders will also be presentingresults from the culminating exercise ofthe tele programa 48-hour challenge sprint exercisefocused on improving the us censusthroughout the tele program we askedleaders to participate in a variety ofexercisesfrom writing policy memos to creatingstakeholder mapsto working on an eight-week challengeproject focused on improvingunemployment and immigration systems inthe ustheir culminating project asked them totake all the lessonslearned from their eight weeks anddevise a solution to a tricky policyproblem in just48 hours in order to give them a reallife experience regarding the quickpace pace with which policy is oftendevelopedwe partnered in developing this exercisewith real leaders involved in the realsenses so that the work the leaders docan have the largest possible impactthis specific exercise was focused onimproving census data collection in thelight of challenges this yearas many of you may know response ratesfor the dieneal census seat this yearsuffered because of the coven-19pandemicand natural disasters like wildfires incaliforniaas part of this 48 exercise hourexercise promptwe asked our telly leaders to one thinkof creative solutions for compensatingfor missing or biased data from the 2020censusand two propose an idea of how toaccurately assess housing needsgiving these missing data we set thelocation of this prompt in the fictionalcountyof nacremento california and asked theleadersto present their proposals to fictionalcounty and state leadersgiven the similarity of our fictionalsetting to a real location in the us thereal location may appear on some of ourparticipant slides and for that weapologizeum the leaders were split into fiveteams for this project and as such we’llhave five teams presentingtheir own unique solutions to thisprompt and we just want to reiteratethat these teams only had 48 hoursover a weekend to come up with theirsolution as you’ll see they’ve come upwith outstanding work in this limitedtimewhich just goes to show the power thatthis type of creative thinking can haveto improve policy and governmentservicessome brief logistical announcements onhow this evening is going to be runafter each presentation we’ll leave acouple of minutes for qa um so please submit your questions inthe q a box below and we’llhave team members either answer thequestions liveor via the the chat bot um weunfortunately do not havewe aren’t able to have all the leaderspresent here tonight so we haverepresentatives from each team butthe the remaining leaders will be ableto answer your questions in the q a boxum and in addition um we’ll introduceour keynote speaker for the event amandaamanda renteria of code for americamidway through the team presentationsso please stay tuned for that umso with that i’m thrilled tointroduce the first five telly leadersum on the first team um also called teamcardinalum you’ll soon see a pattern in the teamnames soon enoughthis first team includes linda lanen acivil and environmental engineering techexec who is currently managing directorof riverside technology advisorsmimi liu a career health technologistwith deepexperience scaling primary careimproving the lives of seniors andengineering electronic health recordsleah moser the director the head ofdriver legal at uber and former productcouncil at amazon supporting globalcompanies on regulatorycommercial privacy and employmentmatters tim rileyfounder of riley consulting and formerhead of analytics and data science atconstant contactdelivering critical solutions at theintersection of tech finance andoperationsand last rogers weed a clean tech execand angel investor with experience inthe private public and non-profitsectors umfor this first presentation we’ll havelinda lainen and rogers weed presentingum so over to you linda and rogersgreat my thanks a lot for thatintroduction and let me see if i canbring my slides up here to get us allstartedokay hopefully everybody can see thoseand so let’slet’s jump right into it uh so umteam team cardinal uh team cardinalfocused the prompt into a problemstatement with two key points ofemphasiswhich are underlined here on the slidethe first pointis that the missing or biased datareferenced in the prompt will almostcertainly result in anundercount of the true population of thecountyand the second key point is that forhousing impacts and ideas to mitigatethosewe chose to focus on studentsexperiencing homelessness in the countyi’m going to address our thoughts onrectifying the undercountand my partner lindo will speak tomitigating homeless student housingimpactswe did map the state’s housingallocation process that’s runevery year every eight years and calledthe regional housing needsallocation methodology or arenamethodologyand we noted a couple of interestingthings about thatfirst of all on the far left we saw thatit’s theamerican community survey or acsthat provides the key data input thatstarts the processfor california rather than the censusdata which is probably a good thingand the second thing that we notice thatthere’s still time as noted by thecircle on the right hand side of theslide to impact this cycle’s processbefore local governments complete theirhousing elements planswhich will inform the next six or soyears of housing policyin the area this slide just confirmswhat was asserted in the promptthat for many reasons the 2020 censuswill result in an undercount of thepopulationwhich left unaddressed will haveknock-on effects in the allocation ofall kinds of government resources overthe next decadeas we researched ways to mitigate thiswe were fortunate to findthat california has computed a hard tocountindex for every census block in thestateits original purpose was to guide thestate census organization’s use ofoutreach resources to conduct the countthis index is a function of 14 differentvariablesthat all correlate with an area beinghard to countincluding the six that are shown here onthe slidethe index ranges from 1 to 136and is an aggregation of each censusblock’s assessment on the 14 variableswe see an opportunity to create a toolthat we’re calling the census adjustmenttool orcat the cat is a spreadsheet thatoperates at the census tract levelits key inputs are the 2010 censuscounts by trackthe hard to count index for each trackand a new total 2020 population estimatefor the countythe tool then uses the htc index toallocate the difference between the 2010total population countand the estimated total 2020 populationof thecounty across all of the county’s censustractsthe allocated surplus can then be spreadacross demographic categories withineach census tractusing the 2010 census as a guidethe key to using the cat is arriving ata revisedtotal estimate of the population ofnecromancer countywe believe that by reaching out to cellphone providers as well as key utilitiesit’s possible to get a much moreaccurate total count of the populationthan the census is likely to produce anda number of the teams that follow ustonightare going to get into details of howthat might be donethe cat then allows you to distributethe difference between thatrevised estimated count and the 2010censusacross every census tract in the countyas you can see on the slide we’reshowing you a screenshot of the toolwhich we actually built during theexerciseand which we’ve uploaded to the aspeninstitute folks to make available to anyentities that would be interested inexploring it furtherand now i’m going to hand it over to myteammate linda to describe our thoughtsabout mitigating the impacts of censusunder countson students experiencing homelessness inmacrame countythanks rogers when we undertook thestudy we reallywanted to we we looked at the studentsexperiencing homelessness as apopulation that could beimpacted positively for a very long timein the futureby experiencing homelessness or beingtransient umstudents who don’t have a steady homeenvironmentare impacted usually for the rest oftheir lives and we felt thatin order to change the long-termoutcomes for these peopleuh the ability to provide stability withthe home was one of the key things wecould do and make a differenceas you see here there are six specialneeds groups defined by state housinglaw and students experiencinghomelessness is not a discrete grouprather they primarily exist in thosefivegroups that you see circled there peoplewith disabilitieslarge households female-headedhouseholds people experiencinghomelessnessand farm workers next slide pleasewe found that the mckinney-vento actrequires the collection ofrobust data on students experiencinghomelessness this data is managed by theus department of education is requiredto be tracked by federal lawthe data is used for creating educationpolicybudget decisions and determining theplacement of grant money awardedpublic school districts the outcome ofthis data collection is to facilitatethe enrollment attendance and successof in school of these youthnext slide please here you can see anexample that we found froma non-profit in washington state calledschoolhouse washington which hasleveraged this data from mckinney ventuinto an actionable dashboard you can seein this casethey show that in king county 3.2percent of all students experiencedhomelessness which totaled almost tenthousand studentsyou can also see the breakdown by gradegroups on the lower leftas well as the type of homelessness onthe lower right which includesdoubled up motels sheltered andunsheltered you can also clearly seethat 83 percent of students in thissituationare students of color next slidebecause the students live in a varietyof situations we recommend partneringwith a variety of ngos that mayfocus on those other housing groups forexampledomestic violence non-profits unitedfarm workersthe department of health and humanservices office of family assistanceand youth outreach reach groups such aswind youth servicesin nacremento as well as any local lgbtcommunity centersas we all know lgbt youth are oftendisplaced from their families of originwhen they come out and are particularlyhigh risk group for homelessnessalso in in the march 2019 californiacommunity colleges reportwe found that sixty percent ofrespondents intheir survey were housing insecure inthe previous yearand 19 of community college studentswere homeless in the previous yearnext slide please i wanted togive special thanks to my friend denisewise for her help with the entire groupuh denise’s ceo of the venturacalifornia housing authorityand answered a text on a saturdaymorningand jumped on some phone calls with usto help all of us navigate the publichousing equation which was hugelyhelpfulback to you my for the q awonderful thanks so much linda rogersand the rest of your teamfor your incredible work on this projectumas a reminder for those of you in theaudience to please submit your questionsin the q a box belowuh and some of our tele leaders may beanswering questions specifically in thatbox or we’ll have themanswered live um so for linda and rogersand the rest ofuh team cardinal um one question that wehad is what made you all decide to focuson narrowing the prompt to studentsexperiencing homelessness when there’splenty of otherum hard to count groups this yeari think that the the biggest thing was aan outcome from ourour conversation with denise actuallywhich where she pointed out the issueswith homeless youth andum and in just the long-term impact thathas we felt likethat was a population we could reallyimpact positivelyyeah something i would add in that toois we we found a very robustalternative data set in the form of thatmckinney-vento data that reallyallows you to have a pretty accuratecount of what these student populationslook like it’s i think it’si think it’s surveyed every year throughthe public school system sothat was another factor great wellthanks rogers linda and the rest of yourteamum next i’m thrilled to introduce thenext five leadersum from team dove um[Music]let me just the this second teamincludes osama alami an engineer wholeads developer programs engineering forgoogle cloud platform and has previouslyled engineering teams for google mapschromecast and google glasschuck forges military director ofanalytics at the navalair warfare center aircraft divisionwill be transitioning from naval servicein 2021 and is focused on datamanagement and digital strategyleonard spicer a startup founder ofmultiple organizations including cloverthe largest point of sale platformshalina taki director of consumerinsights of the global advertising andtech firm pmgwho has a history of leading agenciesand brands to transform data intoactionable insightsand sharon zema general counsel andchief data ethics officer at themarketing cloud company acousticwas worked at the intersection of lawand policy for tech companies like goproand pax labs for this presentationspecifically we’ll havechuck borges presenting so over to youchuckcan everyone see me yesgreat and can everyone see thepresentationwe do not see the slides currently nowtry againone more timehow aboutnowfantastic all right as um i said i’mchuck borges team doveand i’m here to talk to you today aboutour data aggregation toolwith the focus on census tractinformation we all knowthat census tract census data theamerican housing survey data is criticalin helping states and municipalitiesdirect funds where they are neededthroughout the united statesand we all want government funding toget to its intended recipients tominimize fraud waste and abuseto maximize the effectiveness of thatfundingbut the challenge is that it is veryhardnot only to keep track of where peoplelive especially nowbut that lower income and heart attractpopulations are especially transientand one of the examples we found uh inone area of nacrementouh incomes of less than fifteen thousandyearly had up to a fifty percentnon-response rate to the most recentcensusduring our research in our 48 hourproject we found that there werehundreds of public and private data setsthat are resident throughout the statethat can helpclose the gaps in that census data theyall bring different pieces of the puzzlebut they’re all relatively stovepiped soour concept is toaggregate them into a data hub and usethem togetherwe’re taking our inspiration for thatfrom the github movement just as theopen source movement enabled multiplecode bases to be combined togetherwe think we can do that for data anddevelop a data github for public serviceour proposal is essentially to combinemultiple data sets into one aggregateddata huband as we create that data hub tomaximize its efficiency initially inaugmenting the census we would want toaggregate that datawith the census tract mapping as one ofthe initial keys for that data hubon the back end of that data hub youcould build in visualizations anddata analysis to find new associationsand correlations betweendifferent data sets that previouslycouldn’t talk to each otherwe can bring in census data other publicdataand if you automate this process thedata hub will continuously updateas new information is fed into thesource databaseshow is this different from what’salready available wella lot of what has to happen now at theespecially the federal level and thestate and local levelis that when you want to do a data callyou need todo an individual pull of disparate datasets you need to normalize itfor that particular data call and youcan’t repeat the processby building a data hub where data isautomatically fed into itand the the databases talk to each otherwithin the aggregation toolyou’re combining that data intelligentlyand allowing yourself to reachmeaningful conclusions over and overagainyou’re also continuously updating thatdata to keep it currentand with using the census tract data asa common aggregation point for each datasetyou you’re building almost like a key ora primerthat allows all of the differentdatabases that you’re bringing into your data hub to kind of[Music]communicate with each other this willfacilitate both project objectives itwill help you with a census gapevaluationbut because you’re bringing in differentdata sets from housingfrom income from tax records you’ll alsobe able to do a more robust housingneeds analysisand it also supports the 2020 californiadata strategywhich asks that data’s data aggregationaligns longitudinal data sets which thisdoes helps develop an open data mindsetand maximizes interagency data accessand exchangeso how do you aggregate disparate datasources well the first thing we wouldwant to doand since we’re starting with censusdata but this is scalable and i’ll getto that in a few slidesis uh do a data survey figure out whatthe criticaldatabases are that contain the mostinformationthat you need to aggregate census dataamerican housing surveysome housing data education data perhapssome data from theforestry department because of therecent wildfires out westbring them into a data scheme thatextractsthe databases or ingests them into somesort of a central database like asql-based databasecleanse it to clean up any errors uhperhapstwo databases have different ways to tolog a social security number and youneed to correct for thatconform those pieces of data transformthem and load them into the centralaggregate data solutionthis does require agency cooperationacross the different agencies whereyou’re pulling data sources frombut again that’s part of california’sdata strategyso we think that we could generatebuy-in for this type of solutionon the back end you could create almostany type ofvisual or visualize output that you’dlike usinga number of business intelligence toolsout there and as you’re bringingdisparate data sets together into thisone tool you’ll be able to createnew visuals or answer questions thathave always been nagging at youthat you just couldn’t get theinformation for before you could startto map populationdo financial analysis and really finda high base of database justificationsfor disputingfederal apportionments another key uhbenefit to this output is it eliminatesnon-repeatable data calls because nowall the data is coming inbeing normalized with itself and iscontinually being updatedevery time you need to do a data call inthe future you can compare it to a datacall that you did in the pasthow would this work for nacrementocounty well there are many hard to countgroupsbut and i’m not going to go through allof them andthey pull information to for those hardto count groups fromdozens and dozens of public and privatedatabases and recordsright now it is very difficult tocompare data from the national housingtrust fund tosnap data to library records thissolution would solve that problemwhat we would do is use one or morenon-census databases to bridge the gapsin that census datawe’ve mapped out all of those databasesthat i just kind ofglossed over and tied them tosome of the hard to track groups and youcould see if you really wanted to trackfrequently moving renters you would pulldata from those field four databasesextreme poverty on the upper rightuh really relies on snap data sectioneight housing datayou could start to identify which arethe hard to track groups that you needtosolve the problem for first and ingestdata from those databases into thecentral datadata aggregation tool essentially we’vebuilt an ascent data schemeon which databases need to flow to trackwhichhard to track groups here so how do yougo about combining these data setsand fund them well there are a coupleways to fund ainitial data hub we’ve identified somesponsors and some potential fundingsources herebut and even though we didn’t have 48hours to do thiswe had 48 hours to do this ourselves wedid not create the toolone possible approach is a 48 hourhackathona hackathon is a contest a prizechallengewhere you bring in industry teams givethema problem statement give them 48 or 72hoursand have them come up with creativesolutions it’s a great way to create aprototype of a tool with minimum fundingcalifornia has some california iapologizehas some experience with hackathonsthere was a 2015 uh green gov challengefor their pilot portal and they also dida safe drinking water challenge in 2018.in the california safe drinking waterchallenge itthey did a four month hackathon butagain you can do these in 48 or 72 hourswith a simple problem statement such asbuild a central data hub they took 25machine readable data sets had 15 eventsand it cost them about 12 000 in prizesand an associated amount of logisticaland setup coststhe way the hackathon works is you pullin somesome data sets and uh sanitize them foranonymo anonymize the data sets for piiprovided them to the proposing teamswith a problem statement and a challengeand you get a number of outcomes backsome outcomes may actually answer thequestion that you wantedsome outcomes may not but what you’regoing to get is a lot of innovativesolutionsyou’re going to get a lot of externalinsight into how your data can bemanaged and manipulatedyou’re going to get the opportunity froma data hackathon toput a team that really answers yourquestion on contractquickly and because this is a one-offevent you’re minimizing thecost and risk that comes with along-termcontract to develop an unknown quantityin practice the stakeholder is going toreceive some census resultsthey’re going to log them into thedataca.govpull additional data from that data hubabout the htc group in questioncompared to the census data figure outwhere the gaps areand the under counts are and then theycan use the data hubas a tool to either again disputefederal apportionmentsvalidate census results or find new waysto use thedata to identify how we get resources tothe needy populationsthe great thing about the data hubconcept where you’re bringing indatabases to answer a question it is itis infinitely scalablethere’s no reason why we could notpartner withprivate data entities such as facebookand googleto bring in the information that theycollect and they collect a magnitude ofdata we all know it we all see theadvertisementsum in times of crisis you could use yourdata hub to do some predictive modelingnow that you can see how a housingmovement compares toforest fire um problems you can start topredictively model how populations willmigratedepending on where the fires um come inyou can’t do that now on a repeatablebasisit also um helps you de-duplicate if youdoi’m sorry if you do duplicating you’llbe able to make theirdata sets more effective and useful inthe long run also bycreating a central data hub anddeveloping rules against abuse you wouldbeable to more holistically protectthe pii and thedata from outside actors from cybersecurity problemsbelieve it i’m sure you all know thisbut the morecentralized your data solution theeasier it is to actually protect againstcybersecuritya decentralized data posture can be verydifficult to protect fromoutside actors so this actually improvesyour cyber security posture and alsoprevents um a lot of identity and piiissuesin summary you asked us to find acreative solution to compensate formissing our bias data in the 2020 censuscensusand an idea to accurately assess housingneeds we believe that an aggregated datahub fulfills both of these asksit allows county planners to help fillin data that is missing and plan toserve their constituentsit serves the residents of nacrementocounty by allowing a more accuraterepresentationin a holistic manner of where people arewhat they need what they useand it supports the strategic goals ofthe 2020 california data strategyby bringing currently stove-piped datasets togetherinto a more central more open and moreaccessible data solutionand that is my last slide so i will turnit over toquestionsgreat thanks so much uh chuck and andteam doug for this great workon creating a github-like repository soone question that we’ve gottenfrom a couple members of the audiencewhich you touched on a bit at the endbut i’lli’ll allow you or any of your other teammembers to elaborate isare there any other significant privacyconcerns that we should be aware of withattaching thesethese data sets together especially whenwe’re bringing in data from privatecompaniesso there could be and this could beanswered by how you structure theinitial data hubyou might want to start your data hubwithdata sets that don’t have a lot ofprivacy concerns maybethere’s a few data sets out there thatdo not have social security informationthat rely on more publicly accessibleinformation like housing data hasmaybe name address phone number but doesnot contain a lot of piiso you could start with a smaller datahubthat is focused on you know lessprivacy-based data and then develop youryour data governance rules around thescalability aspect of that as you startto bring inthe more concerning pii issues as far aspartnering with private industryi i could see some serious privacyconcerns there you could address thatby making the pipe flow only one way thedata can only intakefrom industry you could also probablyset some parameters up in yourpartnership thereto address any privacy concerns with thedata that you’re either getting fromthem or again you don’t have to sharedata with them in any instance in thisscenario it’s in industry’s bestinterest depending on how it’s marketedto partner with this type of solution uhespecially with the press that socialmedia gets nowadaysuh it would it would show a cooperationwith government thatthat they’d probably welcome great wellthanks so muchuh chuck and team duff for for thisgreat proposalum and we have some questions here inthe chat that the rest of your teammemberscan answer as the rest of thepresentations go on tooso next i’m thrilled to announce thekeynote speaker for our eventamanda renteria amanda is currently thechief executive officer of code foramericaa non-profit organization founded toaddress the widening gap between thepublicand private sectors and their effectiveuse of tech and designamanda has had a unique and diversecareer starting out in the financialindustrymoving to education as a teacher in hersmall hometown communityand finally spending the majority of hercareer in public serviceshe has served as the chief ofoperations at the california departmentof justiceoverseeing over a thousand publicservants and an 850 million dollarbudgetnational political director forsecretary clinton managing the politicaland outreach strategyfor the 2016 presidential campaign andas a chief of staff in the u.s senateduring one of the most productiveperiods in our country’s historyin addition to her policy work she hasalso run for congressional officeand governor believing that empoweringothers is at the heart of public serviceamanda will be delivering our keynoteremarks on the importance of bringingtech and design expertise intogovernmentum so over to you amandahello i got that right all rightum well first of all just hi everyoneum it is wonderful to be with you i wantto thankum cass and dana and the fantasticpanelists and all the tele teamsand also really thank the aspeninstitute for this programand the important discussion i am lovingthese presentationsi think a lot about my time in publicservice as i’m as i’m watching andreading and looking at all the greatwork that you’re doingum so i want to say it’s an honor toshare this virtual space with youbecause it’s awesome to see leaders intech ready to roll up their sleeves andtackle some of our toughest publicsectorchallenges at such a critical time iknow you’ve heard my name is amandarenteriaor para todos gabriel espanol amandarenteria i use the pronounsshe and her and as i just stated i’m theceo of code for americaand i started about now seven months orso agoand so this is a incredibly interestingtime to beuh in this space now i know most of youare familiar with code for america butjust in case we have anyone thatthat code for america is new to themwe’re one of the founding organizationsthat helpseed today’s civic tech ecosystem fromthe first national day of civic hackingwith the white house now almost a decadeagoto our annual code for america summitswhere we bringcivic tech folks together um every yearand so keep an eye out may of 2021 weexpect to see you all therein a virtual likely in a virtualenvironmentwhat we do now at code for america hasevolved over timeour mission is to partner withgovernments at all levelsto help them use technology to create aresponsivepeople-centered government and we have aparticular focuson people who are often left out orforgottenand speaking of that um i thought ishould mention i know there was a littlebit of aboutabout my background mentioned earlierbut it’s actually central to why i’mdrawn to this worki grew up in one of those communitiesone of those forgotten communities inthe central valley of californiaa place called wood lake in tularecounty for those of you who are familiarwith itand i know the importance of governmentworkgovernment programs my parents wereformer farm workers andtaught us to explore opportunities umbut opportunitieswere frankly hard to come by and so i’vealways been inspiredto make sure any kid there or placesthat or places like thathave the opportunity to reach theirpotential so for me i’ve been blessedas you’ve heard with a 20-plus career inpublic serviceworking at all levels of government umand nowi’m at when someone looks at my careeror someone says wellwhy code for america now and i think asyou look througheverything that i’ve done in all thedifferent places whether it’s a publicschool teacheror working in the u.s senate there’salways been this through threat for mycareer which has been to make governmentwork for truly everyoneand code for america was one of the veryfew organizations that has focused onprograms thatimpact the hardest to reach communitiesandknow the potential and power of gov techciv techto enable a government that truly caresfor everyone in all corners of thecountry and i believeright now as we think about the nextcouple of years we have an opportunityto create a transformative moment forgovernmentand right now is actually the criticaltime to act here’s whyfor the first time in my career inpublic service we are seeing an opennessto making significant changes in bothgovernment programsand the way government operates becauseof the effects of the pandemic andsocial distancing protocolswe’re also having critical conversationsabout the need to bring new perspectivesinto governmentprioritizing racial equity and livedexperience and limited experiencesand finally we’re seeing a real sense ofurgency with unprecedented economicchallengesa new administration on the way and thelargest electorate in historyready for change so it’s it’s a time oftremendous opportunityif we get it right and i want to doubleclick on thatbecause if we get it right matters howwe make a difference for what kind ofgovernment and country wehave is really about what we do now forgenerations to comeour focus at code for america onpartnering with governments to rethinkand reimaginethe way government could and should beserving the public is rightfor a new way forward for our country sohow do we do thiswell i want to use three differentstories in three areas of our workto capture what we do as a way ofarticulating the potential of civic techin this moment we’re in right nowso what i’d like to do is now share my ishare my screen with youand put that in so i want to start offwithjust the very simple notion we know atcode for america that the two biggestlevers for changing people’s lives atscale or technology and government and iknow everyone on this callrecognizes that as well but how we putthem togethermatters why code for america we havefive principleswe believe in government we believe inbuilding equitable systemswe believe in a government that putspeople first centering their needsoften called human-centered design webelieve that both policy and servicesneed to be informed by real on thegroundevidence we also know that forgovernment to succeedpeople within government need to beempowered for actionto frankly put together a lot implementa lot of thefindings you are you are having in theseprojects and finallythat no project or tech solution orpolicyis ever done but that we cancontinuously continuously learn andimprovejust as the american experienceexperiment isokay so that’s our list of principlesbut how does this all show up in thepractice of our workso one of our key um program areas iscreating a stronger safety net and weare a part ofall of these different programs theseare kind of the three buckets of workthat we are doingbut i want to start with how we approachthis workon a long-term basis so one we really dothink about before we do anythingwe understand that not all systems werebuilt equitablytherefore as we think and start anyproject and program we start with thispremise that we have to figure out thesystem itselfhow is it actually serving people tobegin with and making sure that we’renot just makinginequitable systems better and fasterbut we’re actually getting to theunderlying process to make it better andmore equitableso i want you to imagine the last thingyou spent abouttwo dollars on maybe a coffee or soda orsome timeat a parking meter to go groceryshopping now imagine you’re a singleparentof three kids who got who just got laidoff from your job and is trying to makeit through the pandemicthere’s a federal program calledtemporary assistance for needy familiesor tanf that came out of the welfarereform program in that in the 1990stanf would give you cash assistance butthe amount you getvaries from state to state in californiayou get about ten dollarsin illinois it’s about six inmississippi it’s 1.89the other thing about tanf is that it’sflexible step bendingand states have quite a lot of leeway todivertfederal funds to non-cash paymentgovernment programssome states trust communities and givemore than eighty percent of that moneydirectly to familiesand in others it’s less than fivepercentcan you guess the single most importantvariable in explaining the difference inthe administration of this benefit fromstate to stateit’s the percentage of the state’spopulationit’s the percentage of the state’spopulation that is blackthink about that for a minute in factthere are studies that have foundracial inequities in tanf implementationcontributedto the impoverishment of more than aquarter million black children peryear from 20 from 20 2012 to 2014.eliminating this inequity in tanf alonewould reduce the black white childpoverty gap by 15 percentnow i know sorry now i know that’s amuch bigger problem than we at code foramerica can solve alonebut what we can do and what everyone incivic tech can do is we can work withstates to help them rethinkhow they implement their safety netservices not just in tech solutionsbut in how to create and implementpolicies that lead to equitable outcomesso the second the second piece of worki’d like to use to illustrateour next point or our next principle isour cleanor clear my record where we leveragetechnology to design to advanceautomatic record clearance and clear alleligible records in the usand this is our second principle putpeople first sometimesfolks on the hill call itpeople-centered approach folks in techcall it human-centered designbut i want you to imagine you’re someonewho lost your job when the pandemic hitand you’re looking for work ever sinceyou find a new job that you’re reallyexcited aboutyou go through the application it’slooking really good like you’re going toget it and then at the last minuteyou get sort of an abrupt rejection whybecause the background check flagged amarijuana related offense on your recordfrom a decade agothis could happen to any of the one inthree people in the u.s who have acriminal recordand it just doesn’t affect you it has aripple effect nearly half of thechildren in this country have a parentwith a recordand just to give you a sense of of theimpact of thatthere are about 47 000 legal barriersfor folks with convictionsthese barriers can keep people fromgoing back to school from getstable housing obtaining a good jobgetting a professional licensehere’s the good news many states haverecord clearance laws on the books thatcan remove these barriersin fact it’s about 23 different statesover the course of these of2018 and q1 of 2019.the bad news is no one thought aboutthose laws or the implementation ofthose lawsfrom the perspective of the people itintended to reachhow do we know this well because of themillions of people with criminal recordsthat should be expungedonly about seven percent of the morethan 70 millionplus people actually go through theprocessnot because they don’t want to notbecause they like to be barred from 47000 different empowering activities it’sbecause you essentially need a highpaid lawyer to make it through and fromthe lens of code for americawe decided this was something we shoulddip our toe into and so a few years backwe started a website to help connectpeople looking to clear their recordswith attorneys who could help themnavigate the processwe immediately saw the uptake and theneed for morebut after a while with the partnershipof some foreign thinking das incaliforniawe started asking what if you didn’thave to applyto clear your record at all what ifthose record clearance policies meantthatall eligible records were automaticallycleared by the governmentas they should be so we set onon on track a first of its kindautomatic record clearance pilotbuilding a no-cost open source tool thatcould be used by allby da’s across across the state here’swhat we learnedwe learned clear my record can analyzeconviction eligibility for approximatelytenthousand individuals in under one minutethat’s approximately five hundredthousand lines of conviction data inunder one minutethat technology basically showed whatwas possible to da’s across the statereally da’s across the countryand the vision for automatic clearancehas now helped reduce or dismissmillions of convictions across thecountry and we along with a growingcoalitionof justice reform advocates are just atthe beginning of this journeywe see a future that really puts peoplefirstwhere government routinely clearseligible recordsthat to us is this concept of reallystarting with how doesgovernment put people first as theydesign and develop policies the lastbucket of work i’d like to share isreally aroundtax benefits and this is particularlyimportant and hascome to light during this period of timeof economic inequality of economicchallenges people are facingwe also believe and one key piece to itis that we believe government servicescan and should be as good as what youcan find in the private sector and iknow many of youfeel the same way we know and this iswhere informed with evidence and powerfor actionand improve continuously is part of theprinciples we live byso my final thought exercise i want youto imagine you’re working in a low-wagejobtrying to make ends meet in the springof 2020.you start hearing about stimulus checkson the news and know that money thatthat 1200 dollars could give your familya lifelinebut then you find out that in order toget a check they need to confirm yourincome via your tax returnand you haven’t filed taxes for the pastthree yearslegally you haven’t had to but now theentire process seemscomplicated daunting and downright scaryyou don’t have the resources to turn toturbo tax or h r blockto walk through this complicatedsituation here’shere’s the good news about whatgovernment is trying to do as it turnsout thereis a government service where volunteershelp people file for taxes for freevita but what good is it if people whoneed that helparen’t aware or that in-person sitesrequire to take a day offstand in line and hope that maybe you’regoing to get seen that daynever mind the fact that you’re puttinga job that you have at riskso good for america started out injanuary of 2020planning to run a very small pilotcalled get your refund that would helppeople file their taxes toaccess earned income tax credit we areoriginally planning to work with just afew partnersin that free in this free irs programwhich we would work with vita sitesthen covet happen and in-person vitasites had to abruptly close in thespring andwhen congress authorized stimuluspayments that would go through the irswe realized get your refund could alsobe an entry pointto help people get that critical cashassistance as wellso we built the first free first freemobile appto help people prepare their taxes webrought on dozens of new partners andempowered them to connect with theirclients virtuallyand we quickly learned from the datathat an easy form was only the firststepwe kept seeing this data come in butthen people would bounce offand what we recognized pretty early onour engineers and our data teamwas that a lot of folks needed not justneeded the quick form but they neededhelp through the formsome with quick frequently askedquestions other resothers really needed a one-on-one helpso we fired up our 25 000 volunteernetwork and developed a 900 person taxtrained volunteer brigadeto help three people through an addedchat function on our siteso that people could go through thatone-on-one help and we were able to doit in spanish tooby the end of the tax season i have tosay the end of the tax seasonsort of kept being extended but everysingle day for our team we were havingthat data come in and trying to changein real time much like the privatesector doesby the end of it all we welcome 2.2million people to our new sitedistributed over 65 million in taxbenefitsand help 500 000 people with their taxesin some formand in code for america fashion and muchof what i’ve seenon in the presentations today we areanalyzing the data this yearto gear up for next year to find newways we continue toimprove the service and empower ourvital partnersso the way we think about it at code foramericais as we think about what does ourcountry need in this moment we dothree three pieces of work building astrong stove safety net ridingthe wrongs of the justice system andreally having this active civicengagementand we believe that by partnering withgovernment to reallychange those services that’s how youbuild a resilient country a resilientgovernment but alsoa resilient country so for my finalthoughtslet me stop my share for a secfor my final thoughts i just want toleave with this so here we arewe’re 15 days since election day 63 daysuntil the next administrationand i have to say i am more confidentthan ever that this workthis way of believing in a bettergovernment is ripe for the moment we areinin fact two of the four top prioritiesof the biden biden-harris transitiontied directly to the work i just talkedabout racial equity and economicrecoverythat means we are two months away from aonce in a generation opportunityto rethink and remake the relationshipbetween government and the people itrepresentsand serves as i said at the outset i’venever seen a moment like this in myentire career in public servicebut i also know from my experience thatthe kind of systemic change we’re alllooking forwon’t come from a top-down approach itwill happen because all of uspublic servants technologists policyadvocatesacademics community members all of usroll up our sleeves and push for agovernment that is ready to meet ourtoughest challengesdeliver better outcomes and commitourselves to serving all people withdignity and respectso i do welcome you all into this worldof believers to create a government wecan all be proud offor generations to come thank you againfor having meand being part of creating a betterfuturewonderful thank you so much amanda foryour inspiring words discussing theimportant workthat code for america does and moregenerally the importance of using techand design thinkinguh in improving government services um iespeciallylove your last bit when you’re talkingabout not umworking towards a top-down approach thatchange really comes from each of usthat is a lot of what the spirit of ourwork is trying to do so we reallyappreciate itum and thank you again umso next i’m thrilled to announceto introduce the members of our thirdteamum of the tele program their membersinclude mark day who is founder andprincipal of tri-sectorand an experienced exec in business andgovernment in the nonprofit sectorsum marcella escobar lava who’s chiefinformation officer of monarch landscapecompaniesand is a transformation change agentacross a broad range of industries andcompanies bob malen a product advisorconsultant and executive withconsiderable experienceforging early stage product andengineering teamsmeche kengla who is chief datastrategist at data products llcand who’s focused on educating folks ondata and modern technologybuilding up analytic skills datacompetencies and optimizing operationsjessica rossman who’s vp general counseland chief compliance officer at trulinkfinancialand who has almost 20 years of legalexperienceum in organizations like ebay vmwarewalmart incredibleand lastly julie yarosh an executive andentrepreneur redefining business andtech partnershipsand an expert in tech strategyoperations it business alignmentand organizational management um markdaywill be presenting for team finch todayso mark over to youthank you maigood evening everyone my name is markday and it is my pleasure to presenton behalf of team finch we decided tocall our recommendationthe virtual safety net or vision 2020.let’s start with the top line ourrecommendation is two partone to collect a targeted set of datato address the immediate undercount ofthe housing insecureand two implement a framework forcollecting andanalyzing existing and new data tomaintain an accurate countgoing forward so what is ourunderstanding of the problemtraditionally the housing insecure arehard to countand this year kobit 19 and wildfireshave madecounting that much harder without thosedatanachmano county simply doesn’t have thecold hard numbersneeded to inform decision making orjustifyfunding so it should be saidthat the gap between the supply anddemand of affordable housingis significant and well known in thecountyand it’s likely to get worse if interestrates remain lowand the stock market continues to gohigherso the key here is to actually put ameaningful and defensible numberto that population and quantifytheir specific needs so the 2020 censuswillcapture some of that population with thevision 2020 estimatewe believe that the county can capturemuch moreand do so in a minimally invasive andcost-effective wayso to get there we’ve actually dividedthehousing insecure population intosubgroupsbecause it’s certainly not a monolithand it’s tough to countuh without appreciating some of thedifferences betweenthe subgroups with better figures foreach of these subgroupswe believe that the county cancross-reference to get a solid revisedcount and a much better betterunderstandingof needsthe first of our recommendations issomething that we think could be tackledover the next six to eight weeksso you could jump in and be prettyaggressive about thisand it should address the majority ofthe census under countlet’s go ahead and dive in to thosesubgroupsso overcrowding happens when you’ve gotgreater than one person per roomuh there are a couple significantcontributing factors that we identifiedso international immigration highhousing costsand our recommendation is that you’dactually look at covet incidentsso one of the risks of overcrowding isthe spread of diseasebut we happen to be in the middle of apandemic consequentlyyou should be able to compare theincidentsby census tract to see where there’sactually more covidthan you would otherwise expect givenprevious estimates or countsfrom the american community surveyso what is residential mobility ortransience that occurs when you’ve gotmore than one move within a county orstate in a given yearthere are several contributing factorsunemployment underemployment mentalillness substance abuse addiction andforeclosureall of these things we would imaginewould go up with covetand the wildfires and our recommendationis that you would lookspecifically credit bureau change ofaddress dataso this is something that you can grabthese data from the credit bureauand be able to calculate the rate oftransients or thenumber of address changes month of themonth by censustracksubstandard housing happens when you’vegot one or more unlivable conditionsincluding bad wiring or lack of workingplumbing uh there are a number ofcontributing factors hereaging housing stock negligent landlordsunintended uses and directmodifications our recommendation wouldbe to actually look at tenant codecomplaintsso this is something that you could doand calculate thepercentage of substandard housing scottusing violationsby census tract and depending on thequality of data that you’re able to getyou may be able to finer tune thegrading of qualitysubfamilies occur when you’ve got two ormore families living togetherto share costs of necessity a number ofcontributing factors including highhousing costsfarm workers aging population or covetedswirluh specifically related to collegestudents right nowand we would suggest that you look atschool enrollment data andcross-reference the school enrollmentwith irs filings as well as data fromcovered in californiathis could be augmented with census andacs datato see when you’ve got more than onefamily registered at a single addressand you can also look at how this wouldchange over timeand then homelessness of course is whenan individual or familydoes not have a fixed regular andadequate nighttime residencethe contributing factor is very similarto the ones that we’ve seen beforeexacerbated by kovit up here uh we wouldrecommend that you look at nacrementosteps forwardso this is an organization that appearsto maintain consistent counts of thehomelesswe’d also recommend monitoring evictionand foreclosureseven if they have not happened withgreat frequency becauseof policies put in place because ofcovid there’s probably a bit of a buildupso you may end up seeing this populationgrow fairly significantlywhen those moratoriums no longer are inplacethere are other counties in californiathat have usedthermal imaging of drones to be able toidentify and help counters go to areasthatmay not be traditional uh or may not beareas that folks are aware of wherethe homeless liveso quickly getting a solid handleon numbers and need is incrediblyimportantagain that’s why we wanted that to bethe first part of our recommendationbut perhaps a close second is being ableto have a reliable handleon the numbers and need at any giventimein the future which leads us to part twohereso our longer term solutionuh should take no more than 12 months tobe able to put in placeuh should be able to provide you withmore timely counts and put youin control and not let you have to bepurely reliant on the acs or other datasources outside the countyand with the right additional datasources provide the ability to actuallydo some forecastingso let me orient you to this this slidehereuh the green boxes and outlines arepart one uh what we would suggest youcan do over the next six to eight weeksthe rest is part two so first at thebottomuh of the slide identifying additionaldata sourcesas you’re moving up connecting thosedata through a virtualization layerthis allows you to keep them in place uhand not have to continue to make changesto those different data sources uh tonormalize themthen aggregating analyzing and gettingsmarter over timewith an analytics engine and thenultimately pushingthese analytics and data out to theright audiences for consumptioneither through new products or improvingexisting oneslike the california tcac opportunitymapsso we believe that you can be aggressiveuh and finish part oneby the end of the year or shortlythereafter this is getting at quick hitdatathese data already exist uh or you cangather them really quite quickly and beable to get very very close we thinkuh to a good count the majorityof the way uh to doing uh or accountingforuh the under count of the census uh therest of of our proposal so part twouh we believe that you could get throughmost of this in 2021uh first identifying and collectingother data sources uhsecond being able to develop and executea procurement planuh to complement uh to make sure thatyou get a full complement of datasources needed for that ongoingintelligence i’ve been building out theinfrastructure in order to be able toaggregate and analyze the dataand then finally and we would actuallysuggest that you do this kind of inparallelbut then really roll it out at the endis the outreach with partnerorganizations in order to be able tocollect those data and be able to putthem out into the communityin meaningful ways so we would definesuccess in threeprimary ways one does the initial vision2020 estimateuh is it complete within the six toeight weeks uhis the 2020 estimate uh or can it beused to challengethe 2020 census count if needed and thenfinallydoes this 2020 estimate align withfutureacs figures soto nacramino county we thank you forsharing your challengeand giving us the opportunity to helpwith that i’ll stopquestions back to you mikewonderful thanks so much uh mark andteam finchum here’s one question that we have fromthe audience um so for the housinginsecure data sourceshow do you eliminate double counting umor overcrowding counts due to covetedincidents and transient individualsor of transients and homelessnessyeah um that’s a good question so weactually think thatthe risk of overcounting or doublecounting is probably pretty lowuh we think that a lot of the uhthe risk is actually severely undercounting uh and sowe actually think that um you’ll be abletojust get a decent representation maybeyou end up doingsome back the envelope math to accountfor hey we’re probably going to have alittle bit ofoverlap between these different groupsand ultimately this is going to be anestimate it’s not going to be perfectso again the the risk is for under countuh we canfactor a bit into the math of someoverlap uh but ultimately we think thatum you know you’re the the risk is stillunder count not over countgreat wonderful well thanks again markand team finchand for those of you in the audienceagain if you have questions feel free toput them in the qa box and the team members will eitheranswer them liveor in the box um at a later timeum next i am thrilledto introduce members of our fourth teamum team robin their members includeravi bala who is a tech executive intuitand former cio at prudential financialand has considerable experience leadingdigital transformationsand launching consumer products in thefinancial services industrybarbara cohen who is chief data officerat the colorado department oftransportationand who has held multiple leadershiproles in state and local governmentmark conlon who’s vp of softwareengineering at bcd traveland has specialized in modernizinglegacy systemssoftware infrastructure and processesalisa harvey dawson who is chief legalofficer of gusto and former generalcounsel at sidewalk labsshe’s an expert in building teams thatenable small businesseswhere work empowers a better life and dvig an engineering exec who is seniordirector of engineering at broadcomand has specialized in transformingsystems and teamsto improve efficiency and quality umalisa and dee will be representing teamrobin so over to you alisa and deethank you mai so we will beoffering two proposals to address bothasks from necromancer countyso our first proposal is to use noveldata sources to compensate for missingor biased senseswe zoomed in on three potential novelsourcesthe first is a mobile phone basicallyusing that as a proxy forpopulation i will explain the rationalebehind thatthe second one was using aerialphotographyto figure out the density of populationsin different areasand the third one is to do sewageanalysisuh there is technology to have tofigure out unique biomarkers that can beused to pretty accuratelycalculate population from from thischemical analysisfor the for number three uheven though it’s it’s promising it’s notquite ready foroperational uh use for forat least not a scale for the remotesensingaerial photography thereit is not granular enough to be accurateso it’s good as a secondary source forcorrelationso we’re really going to focus on themobile phone as a source foruh for figuring this the missingpopulation because that’s what the focuswill beso we believe that thedata from phones or actually just thepresence of the phones themselvescan be used to augment and validate thepopulation countsso the reason we picked this phone asthe novel source wasit is the mobile phone is in widespreaduse around the worldand actually the us according to pew’sresearchtoday said 96 of us adults have a phoneso it is it is automatically detected soit’s like a heartbeatso it’s it’s just every phonealmost represents an individual in in inthe areauh the data can be easily de-identifiedfor privacyso that’s that’s a big plus and the datais real timethere is no operation required to gocollect itso you can get population changes overtime seasonal changes what’s happeninguh that can be acquired in real time andthe last point being that theaccuracy of the data is only going tocontinue improving and that is right nowthe phone industry is in a 4gor fourth generation they’re moving to5g what thatmeans for this for this exercise isthere’ll be a lot more towers there’smuch smaller but there will be a lotmoreso the so the counts that will be comingin would beuh far more accurate and alsouh the the location identity thelocation of these phones will be farmore accurate thanwe can get today so on the flip sidewith mobile phones is the uh how do weaccount for people who don’t have mobilephonesso today at least most children don’thave it 53of kids under 11 do not have mobilephones so how do we account for thatuh and some people have multiple phonesuh often for work and for personal useso how do we how do we figure that partoutthere is you know through some cleveranalysis of this data we can actuallyget time-lapseduh databases to figure out which phonesare trackingtogether between towers and we canfilter that kind of information out thethe other one is it how do you knowsomebody is resident ina geography versus just happening to bevisiting transitioning through how do wehow do we separate that part out andthere againtime lapse databases can be used tofigure that outby by figuring out who’s thereover time versus who just happens to bepassing through especially in the middleof the nightthese these phones are going to be tiedto a specific towerso if you take time to your time lapseanalysis you will find thatthe the ones that are moving throughwill could be filtered outand the last one is the geo boundaryproblem so if you have a countyyou have towers all over the all overthe over the county and these towersare the ones that are close to the edgewill pick up phones fromoutside the county so how do you how doyou address that part and theresome level of approximation has to beused touh to to take a portion of those butalsoas 5g comescomes into play which is very it’salready startedthat accuracy will only improve becausethe tower range gets much smaller so thedata you get is much more accurate sothat’s the reasonuh we are we are proposing thistechnology it has been uh useduh around the world we have studies fromfrance portugal and japan where they’veactuallyproven that is us usableso the main assumption we’re making isthat the county can actuallyacquire this data from either mobilecompanies or third-party companies thatcan that can provide this and ideallynot just theuh you know here are the the the actualphones we don’t need the numbers we justneed to knowyou know 100 million how many phone orthe phones and ideallya location as far as a uh you know in interms of uhas a proxy use a uh cell tower forfor that so uh the process would bestart with baseline population data u.scensus acscombine that with the data from mobileproviders or third party that has themobile dataand then we validate that againstsecondary data sourceswhich there are plenty of and we get anupdated population countso for validation purposes there are uhfor many data different data sources welisted uplisted a few there are probably more uhso the idea being that we have wantto pass you know the smell test i havesome confidencein this updated population count andonce we have this countwhich is uh now can be useduh for for the next part of thechallenge which isthe housing need and there uhwe want to use this for the housethe the number of uh houses we need aswell as the type of housing we needand also with some clever analysis wecan also figure outwhere exactly do we need this kind ofhousing so for thatmy partner lisa will go through that forthe second proposalover to you lisai’m just going to jump to um the nextslide um because youyou address some of this so i want to goquickly through how we calculate itbecause essentially we can use thismobile data to help to calculate ahousing gapand how you do that is you start withthepopulation data that you got from thecensus for county1.5 million and once you do thatyou then look at how many expectedphones that would be and you take thatpercentageof people who are likely to use phonesand you come up with a number that’s1.4 million umi’m sorry 1.37 million when you use themobile datathe mobile data however tells you that1.4million phones were detected so thatessentially means that when you’recalculating the uncounted populationthere’s a 30 000 gap so you have 30 000more um in population that does nothowever give you the final numberof what could be the potential housingneeds instead we use this model from umand we’ve used one from sacramento areacouncil of governmentsum to assess what of those 30 000actually indicate housing needs and thatmodel indicates that two points forevery 2.7 peopleyou have um uh one unit of housingthat’s neededso essentially doing the math you end upat the end of the day with anuh gap of eleven thousand which you canthenuse to potentially get additionalfunding sourcescertainly to calculate um the needs ofvery low and low income populationalso using the psychology model and soon and so forthso at the end of the day you can takethat datathat you you have from mobile tovalidateand show if you have an under over countand then use that under countinginformationto um make your housing plans next slidedand last slide um so these novel datasources like mobile phones can be usedto augment and validate the census countum it can be used to assess the housingneeds by identifying the housing gapsand then ultimately since it’s onceagain based on populationany population based service could alsotake advantage of that information soyou can use it for manyother multiple services that you want todeliver at the end of the dayand i’m now over to you my fourquestionswonderful thanks so much umdee and alyssa for your team’s creativeproposal umto utilize mobile and remote sensing andsewage data to complementcensus data um i’m going to pick aquestion that i thinkencompasses a couple of the questionshere um for either you or your teammembers to respond toum so is there any concern especiallywith the aerial surveillance data or anyof the other data that this data couldbe used for something other thanpopulation estimationfor instance can another governmentagency access the data such as lawenforcement using facial recognitionum for their purposesuh so good question as far as privacygoes we reallyare for at least for mobile we aretalking about getting completelyidentified data so if there are amillion phones we don’t needmillion phone numbers we just need onethrough million so thesethose are just numbers and locations interms of towers there’snothing in there that identifies who theperson is thatthat there is the possibility that weknow there’s10 000 people in this section of thecounty butthat information can be easily availablethrough google maps or many other areasany other technology so there’s nothinguniquethat that would be potentially violatingprivacy that would beintroduced with the solutiongreat well wonderful thanks again uhalisa and dee and to the rest of yourteam for thisum creative proposal umso we are now on to our last and finalteam butbefore i introduce the last team i didwant to introduce one leader who was notable to participatein the 48 hour exercise specifically soshe’s not included in the team slidesbut whom i did want to honor hercontributions during the telly programher name is katie push and she’s a datascience product director at low movetechnologiesum and has a cross-functional softwarebackground so just wanted to honorkatie’s contributions toto this program um and now last but notleastum i’m excited to introduce the last setof leadersum in team sparrowso team sparrow includes anirma guptageneral counsel and business leaderexperienced with startups and largecompaniesand issues at the intersection of techbusiness and lawsteve liebman a software engineeringleader with experienced growing startupsfor socially beneficial applicationsfrom zero to 100 million in scaleand lewis who is chief technologyofficer at moveon.organd has spent her career usingtechnology to power collective actionashley lawrence who is currently chiefof the intelligence systems center atthe johns hopkins applied physicslaboratory and who works at advancingtech to address national and globalchallenges and lastly david weekleyfounder and ceo of multiple startupsincluding med quarter pb wiki and hackerdojoand who formerly worked in product atfacebook and googleum david and ann will be presenting forteam sparrowso over to you david and ann thanks mymapping the uncounted by team sparrowso the problem that we’re tackling hereis covered by a bunch of the other folksas wellis trying to figure out who was missedin the 2020 censuswhere those people may be and to getmore time granular updatesabout what’s going on with thepopulation obviously this uhfeeds into planting the local state andfederal levels and the people whoare missed are likely to need thesegovernment services the most badlyso we’re gonna try and find the folkswho are missingso first we’re going to get into how thecensus works todayand while many states carefully preparedfor the 2020 census and years prior theenumeration surveying and non-responsefollow-up processes in place weresignificantly disrupted first by thepandemic and then by natural disasterswe need additional ways of countingpeople that put a minimal burden onthose whose lives have been disruptedwhen thinking about how to find peoplewe thought about what people doas other teams mentioned as well cellphone penetration is well north of 98even in a bunch of the vulnerablepopulations that we want to make sure tocountpeople also eat and so we can have theopportunity to go and trackinflow of produce and grocery shipmentspeople sleepfinding places where that people couldsleep at whether encampments orbuildingsuh can help in estimations wastewaterhas been mentioned by other teams we didwe dug deep onto this one we actuallyfound academic research is promising inthis areaand also people give off heat andtherefore infrared signaturescould be used to enumerate humanpopulations we picked four of these todive deep intoand the first one of these is mobiledata andas other teams have noted 96 ofamericans have cell phones so this is apowerful data source to dig intouh when cell phones are on they connectto cell towers and these cell towerstrackaggregate volume statistics aggregatedata measures like this can accuratelyestimate population without violatingthe privacy of individual cell phoneownerswe’ve also seen other imagery includingsatellite imageryand aerial imagery contribute topopulation counting this is being usedin a numberof different locations uh in includingthe robinson at all in 2017 to go and dopopulation estimateslandscan also already uses this imageryto go and refinecensus results and we think that thiscould be very helpfulfor refining a census counttech companies like google have beendriving gps camera vans around the usfor more than a decade to create thehigh quality maps that we use every dayand as a result tech companies have ahuge corporate reliable location-basedimage data that can be used to fill inkey hard to count population and datagaps like homeless populationsgoogle in particular has made this dataavailable for use by the census and byother projectsone of the areas that was mostsurprising to us as a team and turns outto havemore promise than we first realized ismeasuring certain chemicals that areproduced by the humanas the as waste water is ingested into awastewater treatment plant this iscalledinfluent we were able to talk to anumber of different researchers who’veexamined thisand who shared with us state of the artpapers including linked herechoi 2018 thai 2019 and pandopoulos 2020to show that this is actually a usabletechnique for populationestimation and we could work withcalifornia association of sanitationagenciesto go and gather this data for doingpopulation estimateswe also were able to identify a numberof otherpopulation estimate sources includingtaking a look at aggregate radiofrequency uhcaptures to measure energy emitted bycell phones and the like using ct clampsfor electrical residential distributionchangeswhich is actually a proven techniquethat’s been used in new york and manyother techniques including doing manualspot checks to goand cross validate the dataso how do we put this all together inall this promising data together into areal working solution for the censusfirst we have to take all the new datasources together and use best practicesfrom the field of heterogeneous dataintegration to combine and augment datain ways that add the right details inthe right placesand control for errors then we can addthis new population data into censusdataenriching data per address and addingnew addresses as neededto make this work we propose creating anew role in the census data gatheringprocessthe trusted integrator the trustedintegrator can be a company selected bygovernment procurement processes thathas expertise in data architectureintegration and pipeliningthe trusted integrator will create dataprocessing and update pipelines thatlive alongsidelocal state and federal census processesworking with agency data teams to createprocesses that work for agenciesthe trusted integrator will beaccountable to contractually boundstandards of datadata quality latency privacy as well ascommunity feedback as neededand as a part of our investigationprocesswe’ve identified a competent vendor thatcan do this work and here’s an exampleof theirassigned letter of interestso when the trusted data integrator isin place and has created data pipelinesto enrich census dataprocesses this data pipelining processcan help update census data morefrequentlymaking it possible to do quarterlyversus annual population estimateupdatesmaking census-based services even moreefficientonce we have estimates we need to figureout howaccurate they are how do we do that ifwe’re trying to correct census dataitselfthe way that we’ll do this was alludedto by other teams as wellwhich is to go and take a look at highconfidence areas from the census countsuse those to cross calibrate thecorrectness ofour data set captures and then use thatcross-calibrated data set capture to goand backfill the censuslow confidence areas we can also checkthis against acs resultsand the existing census populationestimate programin doing our overall research we madesure to connect with apretty wide variety of folks acrossacademia commercial researchand folks who had spent a fair amount oftime thinking aboutmapping and populationthanks from team sparrow back to you mywonderful thanks so much david and annefor and the rest of team sparrowum for your great data in a greater ideaum one question i have isfrom an audience member so tracking theflow of food waste and heat to measurepopulations is a really creative ideaout of curiosity how did you come upwith this so we’re thinking just fromfirst principles we’re trying to findpeoplewhat things are fundamentally true ofpeople um i also happen to have a littlebit ofunfair background in this area becausewhile at facebook under internet.org ibuilt the world’s first map of thedisconnectedtrying to find where are all the peopleon planet earth who are not currentlyonlineand then what what can we do to helpthem butthese sort of signal markers are thingsthat a fair number of other partiesincluding other governments ngos etc aretrying to finduh we also use them at google for crisisaccess responseteam to try and understand uh how thingswere changing with refugee camps interms of wherenew cell towers and other internetinfrastructure would need to be stood upto ensure that refugees had high qualityinternetwhich turns out to be very veryimportant for their successwonderful well thanks so much to davidann and the rest of team sparrowand frankly to all the teams andspeakers for their outstanding work onthis exerciseum as i mentioned before the fact thatthese teams were able to produce theseideas in just 48 hours is incredibleand goes to show the power of thisthinking and improving public policyum before i turn over to our lastspeaker of the evening who will giveour closing remarks i did want to thankeveryone who made this specific eventpossiblethis includes our wonderful aspen staffmayhemsevilla pitt and our director betsycooper along with cass madison of techtalenthaley ponte of the beck center and ofcourse our keynote speakeramanda and the telly leaders themselvesum i now want to introduce one of ourinstructors for the telly pilot danachisnow who will be providing closingremarks for the eveningdana is a senior fellow at the nationalconference on citizenship and partnerfounder at project redesignshe’s also a co-founder of the centerfor civic design and founding member ofthe u.s digital serviceover to you danathank you so much mai it’s been my honorand pleasure tobe one of the lead instructors for tellyon human-centered government servicedeliveryin addition to delivering some of thecourse content i worked really closelywith the teams on their eight weekchallenge projects to make governmentservicesbetter is really hard work it’schallenging it’s frustrating the stakesare incredibly highand most of us go into it not knowingmuch about what thehell we’re in for so telling leaders youshowed that it’s possible to begin toprepare people for public service fromoutside government you went from knowingvery little about government and policymaking to the awesome proposals thatyou’ve made tonightthis is seriously impressive even in aremotepart-time program during a pandemicyou’ve taken the challenges veryseriously exploring massively complexproblem spaces to wrap your arms aroundsome of thesolvable parts of each of the problemsand then designedreally thoughtful implementablehuman-centered approachesto addressing the problems you definedwelldone what you’ve done is incredibly rarein government and i look forwardto seeing you in public service so thankyou for your excellent workfor your curiosity and your humility andyour perseveranceit has been an utter delight to workwith youand thank you to the entire teachingteam all of the partners and thestakeholders in the audience and all ofthe friends and family who are presenttoocongratulations telly leaders with thiswe close out our public portion of theprogram for the evening and bang thegavel on the inaugural cohortof the tech executive leadershipinitiativeagain thank you all for coming and goodnighttelly leaders we’ll see you in thefeedback session up nextcheck on the slack or in your email forinstructions for joining the last partbring your questions for our censuspartners anda celebratory libation for yourselfyou
On November 18, 2020, the Aspen Tech Policy Hub, the National Conference on Citizenship’s Project Redesign, and the Tech Talent Project hosted a public event celebrating the launch of the pilot Tech Executive Leadership Initiative (TELI). TELI is a pilot 8-week skills-building initiative to prepare experienced technology leaders to engage effectively with public sector challenges.
The pilot TELI program began on August 22nd and ran part-time through mid-October. Participants include senior leaders from well-known tech advocacy organizations, non-profit organizations, and private companies; entrepreneurial executives who have built and grown successful startups; and innovators who have led digital transformation efforts at large organizations on their own path to modernization.
During the event, TELI participants presented solutions they devised during a challenge exercise sprint on improving Census data collection. The event also included a keynote address by Amanda Renteria, CEO of Code for America, and a celebration to mark the end of the pilot program.