1、State TechnologyModernizationA Discussion on Managing,Protecting,and Sharing Data Across the USEXPERT DIALOGUEThe American Rescue Plan Act of 2021 along with the Coronavirus Aid,Relief,and Economic Security(CARES)Act of 2020 have provided trillions of dollars in economic stimulus funding.Included in

2、 this legislation are billions of dollars through which state and local governments can modernize technology and provide improved services to citizens.In partnership with Snowflake,the Government Business Council(GBC)interviewed Chief Data Officers(CDOs)from across the country in order to better und

3、erstand how states are using this federal funding to modernize network systems,improve data sharing,and address unemployment fraud.INTRODUCTIONTHE EXPERTSDR.RAVI KRISHNANChief Data&Analytics OfficerState of North DakotaAs CDO for that state of North Dakota,Dr.Krishnan is in charge of all things data

4、 for the state,including analytics,data management,business intelligence,data warehousing,and data strategy.Dr.Krishnan and his team also work with robotics processes,machine learning,and develop custom applications.PATRICK MCLOUGHLINChief Data OfficerState of MarylandAs CDO for the state of Marylan

5、d,Patrick McLoughlin is responsible for coordinating and directing data use and management across Marylands state agencies.Mr.McLoughlin works with state agency officers to set guidance and policies for data that enable better-informed decision-making through data analytics.ADAM CARPENTERChief Data

6、OfficerState of MontanaAs CDO for the state of Montana,Adam Carpenter works with state agencies to understand their data challenges and then help them develop strategies to solve them.Mr.Carpenter leverages his background in consulting and machine learning engineering to guide the states decision-ma

7、king around data.THE INTERVIEWHOW ARE STATES PLANNING TO LEVERAGE STIMULUS FUNDING FOR TECHNOLOGY MODERNIZATION?DR.RAVI KRISHNANAs we are looking at leveraging funds,there have been a lot of investments related to broadband.I think North Dakota is actually one of the premier states in the country wh

8、en it comes to our broadband capabilities.We have a lot of investments in that front,and a lot of focus on automation too.We actually have funds available that are going to help us automate some of our processes as related specifically to healthcare.A lot of funds are also in that realm of cybersecu

9、rity to ensure that we are protecting our data,and of course theres also several projects related to digital transformation.So,all in all we have many projects related to various agencies and we are trying to optimize those CARES dollars to spend on the areas that are most necessary.PATRICK MCLOUGHL

10、IN Marylands Department of Information Technology handles a lot of the overall technology modernization.From a data management standpoint,however,part of my responsibility is developing a statewide strategic data plan.The strategic plan will drive what were looking to do as far as goals and objectiv

11、es over the course of the year and subsequently what that may involve from a technology standpoint.We have our framework in place right now and are going to continue over the next 60 days to put that in place.I think the intention from a technology modernization standpoint is that were seeking to be

12、tter centralize information about the data we have.Weve recently completed the first statewide data inventory and well be looking to implement a statewide data catalog.We will need to determine the best approach for implementing those as effectively as possible and provide a centralized mechanism in

13、 order to 1)understand what data assets we have,2)provide information about our data assets,and 3)facilitate better access to and sharing of that data.Were also looking for an opportunity to leverage innovative technologies,specifically those related to AI and machine learning on the analytics side,

14、and robotic process automation for data ingestion and exploration.Those are areas weve identified as opportunities to enhance or adopt as part of the strategic plan.ADAM CARPENTERA large part of the American Rescue Plan Act money was earmarked to improve broadband access,especially to rural or under

15、served areas.As the state with the least broadband coverage in the country,thats a big lift for us.This also meant that we needed to produce a map and some reporting on demographics and infrastructure so that we could identify what internet exists in the state and who is being served at what speeds.

16、We needed to gather data from a massive number of people in order to really get a good view on it.The American Rescue Plan Act allowed us to take that step.We have a map now that allows us to see broadband coverage in Montana and make better decisions on how to distribute grants.Additionally,when co

17、mbined with census data,we can see which areas report lower than average access to computers.Theres a budget in the digital equity part of the Infrastructure Investment and Jobs Act that could potentially allow us to give these people access to the tools they would need to access the broadband thats

18、 now available.HOW ARE STATES MODERNIZING THEIR SYSTEMS IN ORDER TO APPROPRIATELY MANAGE THE FLOW OF DATA COMING IN?DR.RAVI KRISHNANThe investment that we are looking at in the state of North Dakota is primarily around leveraging cloud.We have a lot of legacy systems that have been around for 10 to

19、30 years or maybe even longer in some instances.We need to prioritize which applications can be moved to the cloud to help us with cost savings and save time.We have several options to accomplish this,including re-hosting,refactoring,replacing,or retiring older networks.We want to help the state bec

20、ome a cloud-centric enterprise,but there will always be a presence of on-premises data centers.We want to have highly resilient systems that are backed up with all the data thats required and have an option to also execute some of the more critical ones as needed.We are in the early stages of our fo

21、rmal data strategy for the state,which will go hand in hand with the technology strategy.We are looking at both of those aspects as we move forward.PATRICK MCLOUGHLINUltimately what were trying to do is get the right information to the right people at the right time so that they can make more inform

22、ed decisions by leveraging data.In order to do that,we need to implement security requirements and provide easy access to the appropriate data once the security requirements are met.Users should be able to quickly act upon the data thats available to them.Really what were trying to do is drive that

23、from the perspective of modernizing our access points to that data.Ultimately what were trying to do is get the right information to the right people at the right time so that they can make more informed decisions by leveraging data.”PATRICK MCLOUGHLINChief Data OfficerMarylandADAM CARPENTERA state

24、is like having 15 large organizations in one.Different departments within a normal organization create silos and problems with data movement.Its even worse when you go up to the agency level.There are whole industries dedicated to data connectivity and movement but some are fairly pricey and require

25、 specific skill sets that are more difficult to get in a state with only one million people.So,what youre starting to see states do is move into a Software as a service(SaaS)-first posture.There are models like Snowflakes and a number of others where its pay as you use it,so you dont have to spend m

26、assive taxpayer dollars on speculation that the tool is going to be useful to you.You can acquire it at a low cost,and then if its useful,youll spend more.Solutions like that actually eliminate a huge part of this data flow problem.In Snowflakes case,for example,we run into multiple vendors that we

27、already do business with who have moved their backend operations to Snowflake,and because Snowflake is essentially coded to be one large tenancy,those organizations can now share their data to us with a couple of clicks at far lower costs.This sharing capability eliminates entire industries that the

28、 state has to engage with and entire skill sets they have to hire.HOW ARE STATES MANAGING DATA SHARING?HAVE DATA SILOS BEEN AN ISSUE?DR.RAVI KRISHNANYes.Data silos have been an issue.In some instances we try to break down those silos by pulling them manually to one spot and just making them talk.We

29、can make bridges and connections,but ultimately its really up to the enterprise-wide data strategy that will help to remove these silos and future silos.Theres this whole concept of open data frameworks.Each state is making the data available for the public to process and use for their own data scie

30、nce and algorithmic platforms.However,one of the bigger challenges there is anonymization.The state of North Dakota works with the hospital system to make anonymized health data available to the public,but we have to be really careful about data privacy.We can make bridges and connections,but ultima

31、tely its really up to the enterprise-wide data strategy that will help to remove these silos and future silos.”DR.RAVI KRISHNANChief Data&Analytics OfficerNorth DakotaOther larger things within the state like licenses or contract tracing data does not lend itself well to data sharing due to privacy

32、implications.As a small state with less than a million people,it becomes difficult to anonymize data.Especially if you have geo data,because,in some areas of the state,there may not be people for a couple miles around.So,that gives us a little pause on making all the data completely open.The other b

33、igger aspect,which is actually more important for the near term,is data sharing between our agencies.Regardless of the state agency,we want each citizen to be recognized the same everywhere.Each agency has their own infrastructure and systems.Were trying to bring it all together into one enterprise-

34、type solution where we can actually take advantage of cross-pollinating ideas.Some other states are also sharing data externally.We want to solve internal data sharing first and then eventually improve the external.PATRICK MCLOUGHLINI think data silos are naturally going to be an issue everywhere.Th

35、is is certainly something were looking at in Maryland.We do have a lot of data sharing agreements across the state,between various agencies and also to external partners where appropriate.Essentially were looking to bring down some of these silos by developing a standardization in language and appro

36、ach that can be adopted by all or most agencies rather than having several one-to-one type relationships between agencies.Establishing that common language and standardizing data security and data privacy will allow us to streamline how we provide access across the state.This would also reduce the o

37、verall time needed when data is required to make decisions.ADAM CARPENTERI find that data silos are an even bigger issue in the public sector.Snowflakes data sharing component works within the state as well.Thats really what we were looking for when we got it.It eliminates the need for that whole in

38、dustry of data sharing tools to transfer data between the agencies.Of course there are still concerns when youre pulling in data from an external source.You have to secure it.Thats something that everybody is fairly good at,but when you move data,thats where the security piece starts to fall apart.A

39、s soon as the data is handed over,its out of your control in multiple ways.Its out of your control from a security perspective,but you also have the problem that data is immediately stale.If there was an error in that data that could badly misrepresent conclusions in the data,you have no opportunity

40、 to correct that.Everything we do is in trust for the citizens that we serve,and that includes holding their data.Our duty to protect that data is sacrosanct above all else,and that has to be more important than enablement,at least to some extent.So,you have to find ways to share data securely.Platf

41、orms,like Snowflakes have the ability where you can set masking policy.For example,you can set a policy that says,“Anytime you see a social security number,mask the whole thing or all but the last four.”That allows you to redact sensitive data sets by default,meaning even if the data gets out,youre

42、not leaking the most sensitive version of that data.Through Snowflake,we can also share data account-to-account within the state so that it never leaves its secure,encrypted cloud storage.DO YOU THINK THERE IS A NEED FOR A NATIONAL STATE ASSOCIATION FOR STATE CDOS,SIMILAR TO THE NASCIO?IF SO,HOW COU

43、LD THIS BE ACHIEVED?PATRICK MCLOUGHLINWe actually do have a state CDO network thats facilitated from the Beeck Center out of Georgetown University.We are in a position right now where all the state CDOs meet at least on a monthly basis and are able to coordinate.We also have breakout sessions that a

44、llow us to have more focused discussions around particular initiatives or processes that are going on.We can collaborate and understand how other states have either successfully implemented programs or share some lessons learned from processes that have been done before.We exchange guidance on whats

45、 worked well and what hasnt worked well.I think so far thats been a really important program because it not only allows us to share some of that information,but we also build those relationships where we can speak freely with each other and do so on a regular basis.I think thats worked out pretty we

46、ll.DR.RAVI KRISHNANThe state CDO network has been working so far,but I agree that if something like NASCIO is formalized,I think it would give us a lot more support both internally and externally,and we could have a more powerful forum.ADAM CARPENTERI think that there needs to be a national state as

47、sociation for state CDOs,even if its just an additional responsibility that NASCIO takes on.As we start to solve the interagency data sharing problem over the next few years,the next objectives on everybodys plate are going to be more effective interstate data sharing and data sharing down to the lo

48、cal level and up to the federal level.NASCIO has money for research scientists and writers that publish work.Our ability to do that is directly dependent on the time the members have to contribute to those documents.Id like to see the State CDO Network enriched,at the very least,with more Theres no

49、question about it.The State CDO Network has been invaluable to all of us.”ADAM CARPENTERChief Data OfficerMontanafunding and capability.But what theyve accomplished with what they have is remarkable and extremely helpful.Theres no question about it.The State CDO Network has been invaluable to all of

50、 us.I cant thank Tyler Kleykamp and Georgetown enough.HOW WOULD YOU DESCRIBE THE LEVEL OF FRAUDULENT CLAIMS ACTIVITY AND WHAT ARE STATES DOING IN RESPONSE?DR.RAVI KRISHNANLooking at it from a technology point of view,we still use mainframes.In this day and age,the mainframes are really powerful mach

51、ines,but they also have limitations.We are looking at using some funding to modernize some of the mainframe platforms that we have.I will also say that there has been a slight uptick in the number of unemployment claims and we are encouraging people across the state to report all of those.I think it

52、s part of the nationwide trend,so its hard to crack down on every one of those and we are asking for support to help track all of that.We are working with the agencies and providing them the support from an IT perspective so they can do the best job possible with all the available data.PATRICK MCLOU

53、GHLINWithin the Maryland Department of Labor,we have our specific Division of Unemployment Insurance.Theyre the ones who are directly responding to this and acting as effectively as they can.They actually partnered directly with LexisNexis to provide an overall assessment of what claims are out ther

54、e,while addressing any claims that may be providing incorrect identity and responding accordingly when necessary.ADAM CARPENTEROne of our most elite data teams in Montana is the team that works on unemployment insurance and fraud.I cant say that the fraud problem is totally buttoned-up,of course,but

55、 our small population size makes the issue easier to manage.Its like an arms race where fraudsters are inventing new ways to steal funding and youre having to stop them.Our fraud prevention team is taking advantage of funding to prevent fraudulent claims.Im actually really proud of how well the stat

56、es doing.IS THERE ANYTHING ELSE YOU WOULD LIKE TO TELL OUR READERS ABOUT THAT WE HAVENT YET DISCUSSED?DR.RAVI KRISHNAN Were investing in robotic processes,automation,AI,machine learning,proof of concept,and so on,but as CDOs,we also need to temper our expectations.Its very important to be realistic

57、about where things are so were able to achieve our objectives.We have one of the most efficient government services in America and we want the citizen experience to be world class.We are making progress towards this goal,but its going to be a slow and steady climb in the beginning until we get our f

58、oundations in place.Typical transformations take anywhere between two to five years,and even longer in government.You need to invest in the initial infrastructure,bring together tools and technology,and train teams on how to work in the cloud.Once we have that,we can really shoot for the stars.PATRI

59、CK MCLOUGHLINTheres value in having a Chief Data Officer in place.Having that coordinated lead on how data can be maximized and most effective helps support the business of government.Through the executive order,the state of Maryland not only created the role of Chief Data Officer,but also the posit

60、ion of Agency Data Officer.I think thats really important for the success of the Chief Data Officer.Having that buy-in and required participation from all the agencies is going to allow for continued success by really driving continued adoption and the use of data as the valuable asset that it is.AD

61、AM CARPENTERDont try to boil the ocean.I think the private sector has shifted over the past 10 years to these shorter return-on-investment timeframes.That seems like a bad thing at first,but its really intended to incentivize breaking projects down into smaller chunks.What Ive learned in my career i

62、s exactly what Desmond Tutu said:“How do you eat an elephant?One bite at a time.”You cant always wait for the perfect solution that solves the entire problem.You really have to break things down into bite-size pieces.Theres a certain size past which projects become no longer intuitively manageable.T

63、hats when you start missing details and letting things slip.You end up with a solution at the end that was more expensive than it should have been,took longer than it should have taken,and still doesnt solve the problem.Instead we really need to start breaking down problems into smaller pieces and m

64、ake a vertical slice that will still deliver value at the end.If you can solve a really annoying or costly problem for$50,000 that saves you$80,000,you may have just funded your next project.You need to invest in the initial infrastructure,bring together tools and technology,and train teams on how t

65、o work in the cloud.Once we have that,we can really shoot for the stars.”DR.RAVI KRISHNANChief Data&Analytics OfficerNorth DakotaINDUSTRY PERSPECTIVEWHY IS TECHNOLOGY MODERNIZATION AND DATA SHARING SO IMPORTANT FOR STATE AGENCIES?Over the last two years,weve seen the positive impacts that technology

66、 modernization has had on the entire public sector.Our response to the COVID pandemic was driven by the sharing of data sets,the integration of modern technologies,and the replacement of legacy systems that kept data away from decision makers and the public.Each of my fellow panelists have communica

67、ted their clear intent to modernize so that better policy decisions can be made across the board and,perhaps more to the point,the data that drove those decisions can be shared in the public view for increased transparency with citizens.HOW CAN PRIVATE INDUSTRY PARTNER WITH THE PUBLIC SECTOR TO HELP

68、 MANAGE DATA FLOWS?Tools.In the engineering community we often refer back to the golden 80/20 rule.For the majority of our public sector partners,custom development of tooling is a significant cost center.Having to develop solutions based on fragmented portfolios of products slows the time to value

69、and really chews up resources.Eighty percent of most mission systems can be built with commercial-off-the-shelf products that simplify the integration and upkeep of the complex systems that drive our public sector technology stack.The cost savings from adopting a ready-made platform allows our partn

70、ers to focus on that last 20 percent,making a platform,not a portfolio,that quickly and efficiently handles the humongous data flows that we have today and expect to grow in the future.WHAT CAN STATE GOVERNMENTS DO TO MINIMIZE FRAUDULENT UNEMPLOYMENT ACTIVITY AND HOW CAN INDUSTRY HELP?The answer to

71、this question,like most questions,is MORE DATA.The simple fact is that patterns are very difficult to detect in small data sets.Larger data sets allow complex systems to establish relationships,common indicators,and statistical outliers more easily.This leads to better insights.You dont know what yo

72、u dont know is probably the best phrase in the world when it comes to data analysis.Weve had hundreds of our current customers adding data like weather,COVID cases,and census data to their own internal data sets to better understand external factors and how they affect an organizations mission.Getti

73、ng access to accurate data sets in an architecture that can handle the size and complexity of those data sets is key.Especially since our public sector organizations are typically charged with a singular mission,but a shared purpose.Unemployment fraud can be detected through the Department of Labor,

74、Internal Revenue Service,state tax data,and a variety of other information sources.Being able to combine those sources from each stakeholder allows the mission to move forward faster and more accurately.NICHOLAS SPEECEChief Federal TechnologistSnowflakeThe discussion with these experts shows that te

75、chnology modernization and improved data sharing is essential for the future of state governance.Transitioning state networks to more cloud-centric platforms and training appropriate staff on how to manage these systems could take years,but the investments will be worthwhile.Data sharing between sta

76、te agencies and between states appears to still be in its infancy;however,formalizing the state CDO network could go a long way toward improving data sharing and exchanging lessons learned from prior efforts.FINAL CONSIDERATIONABOUT GBCAs Government Executive Media Groups research division,Governmen

77、t Business Council(GBC)is dedicated to advancing the business of government through analysis,insight,and analytical independence.An extension of Government Executives 50 years of exemplary editorial standards and commitment to the highest ethical values,GBC studies influential decision makers from a

78、cross government to produce intelligence-based research and analysis.For more information,email us at .ABOUT SNOWFLAKESnowflake is the Data Cloud.Snowflakes Data Cloud is a global network where thousands of organizations mobilize data with near-unlimited scale,concurrency,and performance.Inside the

79、Data Cloud,organizations have a single unified view of data so they can easily discover and securely share governed data and execute diverse analytics workloads.Snowflake enables every organization to mobilize their data.Customers use the Data Cloud to unite siloed data,discover and securely share data,and execute diverse analytic workloads.Wherever data or users live,Snowflake delivers a single data experience that spans multiple clouds and geographies.Find out more at https:/