Skip to main content
Search form
  • 00:04

    [Speaker Series #3 With Keith Porcaro][How Technology Fails Us and What We Can Do About It]

  • 00:09

    KEITH PORCARO: But modern life is built on promises--our work life, our romantic life, our family life,our social life.Whether we're doing business with a strangeron the other side of the world or the friend sitting nextto us, we rely on the promises that others makeabout what they'll do next.And our digital lives too are built on promises--promises about how this technology will change

  • 00:30

    KEITH PORCARO [continued]: your life or who will maintain that codeor what will happen to our data.These promises are hard to keep.Organizations making them can have a change of heartor a change of fortune.WhatsApp gets bought and decides it's in the advertisingbusiness now.iRobot decides that now isn't the time

  • 00:50

    KEITH PORCARO [continued]: to sell the data that your Roomba makesas it zips around your apartment,but it might be later.A cancer research facility in Boston goes under,and 10 years worth of research is incinerated.So part of the problem here is that we're tying promisesabout technology, about code, about things,to the financial health of organizations--to business models that could fail or change.

  • 01:11

    KEITH PORCARO [continued]: But even governments can change their promisesand disappear scientific data from the internet.Most new organizations fail.Most open source projects fail at a rate of about 80%.So 80% of new organizations will fail and take their technologywith them.80% of new open source projects fail.

  • 01:32

    KEITH PORCARO [continued]: And that's OK-- things fail.Technology goes obsolete.Businesses go under.People change their mind.Things fall apart.Stuff happens.I'm not here to say that we're going to stop organizationsfrom failing.I'm here to talk about how we can deal with these failuresand how we can be better.So my name is Keith Porcaro.I'm a fellow at the Berkman Klein Center.

  • 01:53

    KEITH PORCARO [continued]: And I'm working on a project called Digital Public.And today, I'm going to talk about how we handle failurebecause, right now, we're not great at it.Our way of handling failure is to sort of ignore it.Failure is a thing that happens to other people,or failure is just another stepping stoneon the road to success.And here when I'm talking about technology failing,

  • 02:14

    KEITH PORCARO [continued]: I'm not really talking about all technologyor all software or all code.Move fast and break things is fineif you're a bike sharing startup or if you'remaking quizzes about which Hogwarts house you belong in.I'm a Ravenclaw myself.I'm talking about public interest technology--technology that solves a social problem or a civic problemor supports public research.

  • 02:35

    KEITH PORCARO [continued]: In other words, something where we're using technology as ameans, as a tool to keep a promise about a social missionor a social movement.Here move fast and break things is less OK.We often talk about how technologycan solve some problem, and there'sa nice appealing story to this.We have a problem.Just add technology and then the problem is fixed.

  • 02:55

    KEITH PORCARO [continued]: The story doesn't always end there.Public interest technology creates dependencies.Communities depend on these projects,on code, on data in order to keep these solutions alive,whether we're trying to mitigate the effects of climate changeor cure cancer or use software to deliver government servicesor use data to help local fishermen maintain industryor building software to help translators subtitle videos.

  • 03:16

    KEITH PORCARO [continued]: We're trying really hard to keep these promisesand keep these projects alive, but all the evidencesays that these promises will be broken.We promise we won't sell your data,unless we change our mind.This research will be publicly accessible,until the administration changes.This technology will keep working as longas the government keeps paying the contractor, and thensomething changes.

  • 03:37

    KEITH PORCARO [continued]: And these broken promises have real harmif we don't prepare for them.So before coming to Berkman, I worked at a nonprofit,and I worked with technology.And I saw technology for social good projects come and go.And when they did, what happened to their datawas often a mystery, leaving their missionsto sort of twist in the wind.So on the one hand, we aren't going to prevent failures

  • 03:57

    KEITH PORCARO [continued]: from happening.But on the other hand, these projects, this data,this code is valuable to the public, to communities.So what can we do?We can improve I think our stewardship of the dataand code that really matters for massive joint research projectsto the data and code that a single organization generates.

  • 04:18

    KEITH PORCARO [continued]: And that's what I'm going to talk about today.How can we be better stewards in an uncertain digital world?So I'm going to talk a little bit about some of the workthat we are doing around data trusts.It's a model for digital stewardshipfrom helping find public interest technologyprojects' new homes to facilitatingdata and technology sharing across siloed organizations.I'll explain what all that means in a minute.

  • 04:40

    KEITH PORCARO [continued]: But first, I want to talk about two waysthat we deal with technology failure nowand why it just isn't good enough.The first is open sourcing.When organizations decide to start--stop supporting a project, they tend to open sourceit, which usually means they put the code upon GitHub with a license that says anybody can use or modifyit.If you're lucky, you might get some documentation too.

  • 05:01

    KEITH PORCARO [continued]: This feels a little bit like abandoning a projecton a desert island in the middle of the ocean.Technically, anybody can come and access it,but they probably won't.Don't get me wrong--open source software is great.I use it every day.My computer runs Linux right now.But that doesn't mean that open sourcingis great for every project.Technology is hard, and even well-supported projects

  • 05:22

    KEITH PORCARO [continued]: are difficult to maintain.Linux may be the best supported open sourcesoftware in the world--takes an average of five years to fix bugs in its kernel.And even well-maintained projectsmay not be well supported.Open SSL is software for secure web communicationsthat's used in the majority of the world's web servers.You may have heard of it because of the Heartbleed bug--a security vulnerability that affected every web server

  • 05:44

    KEITH PORCARO [continued]: that was running it-- again, a majority of web serversin the world.Before this bug, Open SSL, the project,had one full-time employee and averaged about $2,000 a yearin donations, and these are the popular projects.A study of Sourceforge through 2009said of the 150,000 or so open source projects on it,just one in six got to a first release.

  • 06:05

    KEITH PORCARO [continued]: GitHub, in the meantime, has tens of millionsof repositories--25 million that were updated in 2017 and another 20 millionthat haven't been updated in over a year.So just saying that we're going to open source it may notbe the best thing for the project for maintainer ato find your project, it might be like, well,trying to find a desert island in the middle of the ocean.

  • 06:26

    KEITH PORCARO [continued]: And in the case of public interest technology,abandoning a project in the public domainmay not be in the best interest of the project's mission.The people who depend on a project and the people whomaintain a project are different.They have different skills and they aren't lookingfrom the same places for help.And so without somebody supporting it, an organization,a group, a person, it may be hard for an organization who

  • 06:48

    KEITH PORCARO [continued]: actually needs the project to use it.It feels to me a little bit like abandoning a seeing eyedog on a desert island and hoping that blind people willjust sort of come along.So hoping that the public runs with failed technology projectisn't a great option either.The second way that we deal with failureis we're closing data up.We build regulations to promote innovation and protect people

  • 07:09

    KEITH PORCARO [continued]: from data breaches, and we should.But that means that organizationsbuild data forts to protect their IPand to avoid liability.In other words, we're staying on our islands,and we're building walls around them.And this warps the relationship that wehave with data in strange ways.It means that organizations who hold the most datawill be the ones who can afford liability risksand that makes collaboration hard.

  • 07:31

    KEITH PORCARO [continued]: It's more difficult for organizationsto collaborate on things like biomedical researchor for local governments to collaborateon technology procurement.It took a massive breach of trust on Facebookfor the company to even agree to workwith researchers and academics in the first place.And who knows how long that will last.So by filtering data through these forts,we're sort of bypassing our opportunityto build independent relationships

  • 07:52

    KEITH PORCARO [continued]: with the communities of the peoplethat we're actually researching.In these relationships online, are reallyfiltered through these business relationships,so the public domain may not be the best place for a projectbut locking it away isn't doing it any better.So our work focuses on stewardship.In other words, as we're doing more and more,how can we ensure that promises about data and code are kept?

  • 08:12

    KEITH PORCARO [continued]: How can we prevent undesired uses?Can we build safeguards that let ususe sensitive data that we might not have had access to before?And how can we give communities voice about whathappens to their data?And again, we're doing this because data and technologyis valuable.We're collecting and creating more and more of it,and more and more of it is sensitive.And we have the ability to use itfor more and more varied applications.

  • 08:34

    KEITH PORCARO [continued]: And when these-- projects fail, thereare more and more severe consequencesfor the communities who rely on them.Put it another way we're rewriting relationshipsbetween people and organizations and governmentsbased on data and code.The frameworks for managing these relationshipsare quickly growing obsolete, so weneed new tools to handle these digital relationships.

  • 08:54

    KEITH PORCARO [continued]: And trusts might be one of them.OK, so onto the good stuff.We're saying that we're trying to create stewardship.How can we do that?How can we do that right now?In law, there's something called a fiduciary.Basically, it's when someone has a legal dutyto take care of something on behalf of somebody else.So your lawyer has a fiduciary duty to you as a client.A company has a fiduciary duty to shareholders.

  • 09:15

    KEITH PORCARO [continued]: A lot of times, these duties regulatehow a profession works.Legal scholars Jonathan Zittrain and Jack Balkinhave suggested that this is kind of onepath to regulating companies thathold lots of personal data--so if you say that Facebook has a legally enforceable dutynot just to its shareholders but to its users--to the people whose data they hold.This is kind of an interesting idea

  • 09:36

    KEITH PORCARO [continued]: that somebody can have a legal duty to take care of somethingon behalf of somebody else, and thereare lots of examples of that come from government fiat,but we can also form these duties privately with a trust.So a trust is a legal agreement-- a wayto own something for the benefit of somebody else.Trusts start with a grantor.A grantor has an asset--

  • 09:56

    KEITH PORCARO [continued]: money, code, data, whatever.A grantor writes a trust documentto put an asset in trust, essentially declaring that itis so, and it is in the trust.And this trust document has a couple of ingredients.Obviously, that's to say what the asset actually is--the thing that we're holding and trust is.Without the assets.There's no trust.It also is to designate beneficiaries-- a person,a group of people, the public, in the case

  • 10:17

    KEITH PORCARO [continued]: of a charitable trust.In other words, it has to answer who isthis asset intended to benefit?Who is it for?It to designate a trustee.This is the person of the organizationresponsible for taking care of the asseton behalf of the beneficiaries.And finally, we have to articulate the trust's duties.How should the trustee steward and steer this asset?

  • 10:38

    KEITH PORCARO [continued]: And depending on where we are, wecan customize these duties to fit the needsof our specific project.In fact, we can customize kind of a lothere within this triangle of duties, trustee,and beneficiary.We can even do something kind of crazy like givethe trustee a board of directors or have community electionsor have multiple trustees.In other words, this is a really useful framework

  • 10:58

    KEITH PORCARO [continued]: for sussing out what duties we ought to have about somethingand to whom those duties are owed.And so we see this framework kind of a lot.Charities are often organized as trusts.In the UK, civic trusts are used to help communitiesmanage local resources.In the US, land trusts aren't trustsin the strict legal sense, but they'remanaged by a third party and based on sortof conservation easements.

  • 11:19

    KEITH PORCARO [continued]: But they borrow it from this basic structure.There's a trustee.There's duties and a beneficiary to help preserve and conserveland.So as we're thinking about protecting public interesttechnology projects and making data sharing easier,we might start thinking about how we canbuild on a structure like this.So all right, how can we use it as a wayto deal failure or transition-- something better

  • 11:40

    KEITH PORCARO [continued]: than abandoning a project on a desert island?In a way, it sort of feels like estate planningfor a technology project.So we might write a trust that says the goal of this trustis to help find a new caretaker for this such and such softwarethat continues its mission in the eventthat the current organization goes under, abandonsthe project, and sells out, et cetera.So how would we put this trust together?

  • 12:01

    KEITH PORCARO [continued]: We need an asset, which might be a copy of the project,along with an unlimited license to use it.And of course, technology products change,so we might need to upload that asset to a trustee serverand update it from time to time.We need to be able to articulate beneficiaries--who this project is supposed to be for.And we need a trustee with duties--somebody would hold a copy of this projectin, say, cold storage and then find a new home for it

  • 12:21

    KEITH PORCARO [continued]: if something went wrong.So for a charity running a technology project,this looks a little bit like insurance but for your mission,and it could be relatively inexpensive.OK, so we can use a trust as a wayto protect a project's mission and doour best to find a new home in the eventthat something goes wrong.And this could make something else that's kind of cool.

  • 12:43

    KEITH PORCARO [continued]: So if I'm a trustee who has one of these trusts,I might need a way to find new organizationsto take on a project and vet those who are interested in itbecause, as a trustee, I'm duty boundto protect the beneficiaries and continue the project's mission.And so I might start soliciting applicationsfor adopting the projects so that Ican vet requests and ensure that the organization taking overwill keep their promises.

  • 13:05

    KEITH PORCARO [continued]: And if I managed more than one of these trusts,I might set up something like a hub for adoptingthese new projects.And this might be a better place for organizationsto come and help pick up technology projectswhere they left off.And this idea that a trustee could solicit and processapplications for data and code is actuallykind of interesting on its own.So we could have a data set or a code base managed

  • 13:27

    KEITH PORCARO [continued]: by a trustee, who's obligated to protect and supporta community.And that trustee could decide whichoutside requests for that data are going to be fulfilled.So this sounds like a potentiallyuseful and interesting structure for governing data sharingbetween people who have data and people who need it,especially when we aren't sure who's going to be requestingthat data up front.In fact, the UK government is using a similar structure

  • 13:49

    KEITH PORCARO [continued]: like this in order to start thinkingabout how they can govern AI and the data that it needs.We can also think about this from a social science researchperspective.So there's a wealth of data on poverty prevention,on the effectiveness of legal assistance,and how people move through government services thatare hiding in the data sets of charities and social serviceorganizations.But this data is also really sensitive,

  • 14:10

    KEITH PORCARO [continued]: so surfacing those insights mightrequire some kind of trusted intermediarywho could approve or deny requestswhile ensuring client anonymity and enforcementagainst bad actors.We could keep building on this idea.So if trusts are a way to help build relationshipsbetween data providers and data requesters,it doesn't have to be just one provider and one requester.

  • 14:30

    KEITH PORCARO [continued]: We can help collaboratives--groups of organizations-- find ground for sharing codeand technology.So in the US, local governments are contributing codeto a trust in order to help improve technologybehind local elections.And in a field like, say, cancer research wherethere are universities and startups and large corporationsall with different data protection standards

  • 14:51

    KEITH PORCARO [continued]: in different countries.A trust could be a way to resolveall of these a little bit more easily.It could also make a framework like this greedy sothat if somebody takes data out and uses it to developsomething else, they have to put the results of their researchback into the trust.And far be it from me or anybody to hold themselvesout as a GDPR expert, but for organizations worriedabout sharing data with US companies,a trust could be the sort of intermediary vehicle

  • 15:14

    KEITH PORCARO [continued]: for enforcing third-party data processing.And so if we were to really go crazy, in the UK,there are some cities-- some cities have civic trusts.They're volunteer organizations with missionsthat center around protecting civic lifeand influencing the planning of their city.So we might think about a digital civic trust as a wayto empower communities to decide what happens to their data

  • 15:37

    KEITH PORCARO [continued]: with a digital public advocate backing it.So in a nutshell, these are the thingsthat we're trying to do at Digital Public.We're trying to connect together this archipelago of isolateddata and technology projects.We're trying to keep mission-driven technologyproducts alive and make sharing data easier.And so Causeway, our platform for-- to help

  • 15:58

    KEITH PORCARO [continued]: stewarding public interest technology with trusts--will be ready later this year.We'll make it easier for organizationsto secure their project's missionand help set up structures to share sensitive data.So the last part of this presentation,I want to close by sort of zooming out a little bit.I said at the beginning of this that wearen't very good about facing the ideathat technology or really anything may fail.

  • 16:20

    KEITH PORCARO [continued]: It's not just technology.We're bad at confronting risk head on with intention.In other fields, we try to account for that.We set up rules and guidelines and frameworks and insurance.And we do all this before something goes wrong, and notafter.These rules and these guidelines and these frameworkschange how we behave.They change habits.And so what I'm saying is that weneed to prepare for the change that we might see in the world.

  • 16:43

    KEITH PORCARO [continued]: And because technology is rewriting our relationshipswith each other, with governments,with companies, the idea of public service,we need new frameworks for grapplingwith these relationships.This trust is one.It may not be the best framework for everything all the timebut that isn't our goal.Our goal is to help people and organizations and governmentsreconcile with these new digital relationships

  • 17:04

    KEITH PORCARO [continued]: and build governance frameworks that are suitedto their mission and purpose.So whether it's a trust or a contract or a license,these are foundations they're rules of the game.Digital relationships are built. Digital culturesare built on how people behave, on what people understand,what duties people owe to each other, and what people value.And these relationships-- these cultures of data--are going to evolve on their own.

  • 17:26

    KEITH PORCARO [continued]: We're just trying to help nudge them in the right direction.And so there's an interesting design challenge herefor us is how do we help people plan for failure,to think about stewardship, to think about governance,to consider their mission without askingthem to become lawyers or data scientists or programmers?I think it requires us to think a little bit differentlyabout how we're building software.We need to help people understand

  • 17:48

    KEITH PORCARO [continued]: the rules of digital relationshipsand what their rights are.And so our goal when we're writing softwareisn't just to improve somebody's experienceand make it delightful.It should be to improve their understanding.So for the most part, this goes against howsoftware is designed.Most of the websites that we use onlinehave two elements in common.One is kind of a slot machine interface,so you type something into a form.

  • 18:08

    KEITH PORCARO [continued]: You hit a button, and you get something random back.And then you might do it again.And there's nothing to really helpyou connect the thing that you putin with the thing you get back.Government feels a little bit like this too.We send off a form, and we wait.And then we get something back.And then we might have to fill out another form,and then we wait.And then we might get something back, or we might not.The second thing that we deal with hereis sort of a novella of legalese.

  • 18:30

    KEITH PORCARO [continued]: So trying to understand the rules of your relationshipwith a service is meant to be as difficult as possible.So you have two choices.You can either ignore it and hope for the bestor get lots of training in how law and technology works,so you can understand what's going on.That's expensive.And this is just kind of what we're used to.It doesn't actually have to be this way.I think there's a better way, and we're sort of

  • 18:51

    KEITH PORCARO [continued]: interested in finding it.So one place that we're looking for inspiration is game design.Will Wright, the creator of Sim City,talks about how when you're introducinga player to a game this complicated,it's so complicated that you can't justdump everything on somebody, or they'll be totally lost.So the beginning of the game focuseson helping people build this sortof simple mental model of what's going on in the world.It might be wrong, might be incomplete,

  • 19:13

    KEITH PORCARO [continued]: but it gives people a toehold and a wayto sort of bootstrap a progressive understandingof the world.And so part of what we're trying to figure outis how do we help people build these intermediateunderstandings of the world, even if it's wrong?So maybe a piece of it is making language that's simplerto understand or using icons to help people keeptrack of abstract concepts.

  • 19:33

    KEITH PORCARO [continued]: And maybe another piece is taking advantageof the fact that online, we're workingin an interactive medium.So rather than a slot machine, wecan make explanations that are interactivewhere people can play with the settingsand see how things change.And maybe this won't be enough, or not everybody will use them.Or not everybody will care.That's OK.The fact is people are forming mental models

  • 19:55

    KEITH PORCARO [continued]: about this technology, anyway.Every interaction that you have with a piece of software,with a bureaucracy, or a form changes your understandingabout how it works.We build models about the world all the timeconstantly without knowing it.And so if we're trying to empower peopleto take control of their data and of the technology that theydepend on, shouldn't we be helping them become experts,

  • 20:16

    KEITH PORCARO [continued]: just a little bit at a time?In sociology, there's a theory of social movementsthat says that a movement is an ongoing dialogue,a negotiation, a way for people to agreeon a common meaning for words and ideas.So in that sense, we're trying to build a digital stewardshipmovement as best we can by tryingto help people understand what that might mean for them.We're developing model trusts and model modules--

  • 20:37

    KEITH PORCARO [continued]: little LEGO blocks for people to understand piecemeal and thenfit together.And we're doing research to find outhow people think about data in code and how they value itand what their sort of intermediate understandingof the rules actually are.And we're working on interactive explanationsto help people understand the peculiarities of someof the legal structures that we're throwing at them.

  • 20:58

    KEITH PORCARO [continued]: Our goal at the end of the day isto make digital stewardship more accessible and moreparticipatory, not just for experts but for everybody.And as data is more and more a part of our everyday livesand as our public spaces are becoming digital as muchas they are physical, the decisionsthat we make about how we govern them are ever more important.And that's really the challenge that underlies all of this

  • 21:21

    KEITH PORCARO [continued]: is how can we build that participatory digital societythat's really worthy of people's trust?We think it's a challenge worth chasing.Thanks very much.[APPLAUSE]

  • 21:35

    LOUISE COADY: Thank you so much.That was really interesting.So we're going to do questions now.Yeah, yeah, if you keep it on and I'llrepeat the questions after they'reasked so for the cameras.

  • 21:46

    KEITH PORCARO: OK, I thought you were going to stall.It's a good stalling technique too.

  • 21:48

    LOUISE COADY: Yeah, yeah, yeah, yeah, yeah.OK, awesome.So why use a data trust rather than a data repository?

  • 21:55

    KEITH PORCARO: It's a good question,and I think the answer is that it's not really an either/or.Like a data trust, what we're focused on here,is sort of how decisions are made about whathappens to that data.And so here where there might be some assurance in a datarepository if it's government-backed.But if it's not government-backedor if the administration changes,those sort of additional wins that it might be privy to.

  • 22:17

    KEITH PORCARO [continued]: And so you might think about this for a data repository.They might use a data trust for somethinglike archival purposes.So there's sort of a backup to the backupand that trust is only activated when--if something catastrophic happened to the repository.

  • 22:31

    LOUISE COADY: Do you have any other examples of companiesusing UX for legalese?

  • 22:37

    KEITH PORCARO: There aren't many great examples.The banner one that I used to use for a long time--and I don't know if it's still true--was Tumblr did a really good job of explainingtheir terms of service in sort of like regular humansentences.But then the problem with that isthat you see like this sort of we'll explain what this means.And then at the bottom, you're like by the way,this doesn't actually-- like noneof these other human sentences mean anything

  • 22:58

    KEITH PORCARO [continued]: about what we're allowed to do and not allowed to do.And if you misunderstand it, it'sjust sort of too bad for you.Companies, I think aren't really focusing on it.I think where you're seeing a lot of interesting thingshappen is in government and in public service.And the reason, I think, is that if you're a company--when you're saying-- when we talk aboutdesign for a company, what you're really talking about

  • 23:19

    KEITH PORCARO [continued]: is sort of designing your user base.You're designing something that people--some group of people but not everybody will--look at it intuitively.And with government doing public service and with charities,like they can't pick their user base.They've got to be accessible to everybody.And so that allows for a few more interesting things,but it's hard.

  • 23:37

    LOUISE COADY: How do you decide who'sgoing to be able to execute it?

  • 23:40

    KEITH PORCARO: I think it's a little bit like askinghow are you going to decide who you'regoing to sign a contract with.So I think that it would be kind of communities will eventuallyhave to sort of pick this.But what I'm also suggesting maybe implicitlyis that they're sort of maybe a new professional classbrewing of people who had these skills to be

  • 24:02

    KEITH PORCARO [continued]: sort of data stewards.And I don't really know what sort of beginningsof that might look like.You might sort of look at the beginnings of bankingor the beginnings of law or the beginningsof specialized professions, and they get offto kind of rocky starts.There's kind of old adage about how regulationstend to be written in blood--something horrible tends to happen,and then we decide how to fix it.

  • 24:22

    KEITH PORCARO [continued]: And I'm not quite sure that we've hit that moment yet.I don't quite get to your question.So one of the examples that I gave hereis a bunch of state governments in the US.So in the US, elections are administered at a local level,not at a national level.And so each state is sort of responsiblefor its own election infrastructure, whichis exactly as much of a mess as you might think it would be.

  • 24:44

    KEITH PORCARO [continued]: So there is an organization calledDemocracy Works, that works with a bunch of state electionofficials.And they do a bunch of different things.And they are working with state governmentsto set up a trust where they are the sort of trustee--this nonprofit is.And all of the state governments are putting--who signed up for this trust-- areputting their sort of election tech in into the trust.

  • 25:07

    KEITH PORCARO [continued]: And the role the nonprofit then is to sort of be the stewards.So they'll maintain the software.They'll keep it up.They'll manage that sort of membership,and then everybody will get the product of that back out again.And so for state governments who have a lot of restrictionsabout open sourcing software or about howto dispose of software or how to dispose of an asset,this is really convenient for them.And because they're also not very good

  • 25:27

    KEITH PORCARO [continued]: or don't have very good resources with procurement,it's a useful way for them to get technology.And for the community, it's a useful wayto sort of not have to reinvent the wheel all the time.I think in a lot in almost all the examplesthat we've worked through with people,there has to be somebody in the community who's kind

  • 25:48

    KEITH PORCARO [continued]: of able to step into that role.Like this isn't going to be a good structurewhere there's sort of an absence of expertise at some level.This may be a good structure for bridging expertisefrom an adjacent field, but it's not a substitute.

  • 26:03

    LOUISE COADY: Thoughts around the UK government's useof data trusts?

  • 26:07

    KEITH PORCARO: So I think on the use of data trusts,it's still--it's relatively early in the game.I think that most of what I've seen out of the UK governmentis focused on industries that are not quite nascent yet.So the convenient thing about sayingwe're going to regulate AI is there'snobody on the other side to complainabout the regulation of AI.I think we'll see how that actually

  • 26:28

    KEITH PORCARO [continued]: gets put into practice.I think with something like the NHS data sets--and I don't know very much about it.So if I go totally off base, you can fill mein with more contexts.But it strikes me that the challenge with the really largedata sets like that is basically gettingadditional consent for the thingsthat you'd like to do to it.And I think that there is nothing sort of--

  • 26:51

    KEITH PORCARO [continued]: there's nothing inherent about a legal structure of a trust thatwould make that go any faster--like it's still a lot of data.It's a lot of consent to get.I do think that what it does provideis sort of a vehicle to do that.So in the same way that like NHS or government or corporationsare organized and kind of data governancereally means like compliance and that really

  • 27:12

    KEITH PORCARO [continued]: means preventing data breaches.There's not anybody who is set upas an intermediary to help researchers, for instance.And so this might be a sort of a placewhere you could put those resources and make that happen.But I don't know that there's anything sort of particularabout this structure that would make that go faster.

  • 27:27

    LOUISE COADY: Just asking if there's a place for blockchainwithin data clouds?

  • 27:31

    KEITH PORCARO: I think that there's sort of applesand like wheelchairs a little bit.So you could see the governance of a decentralized organizationlive in a trust.And but as you start kind of getting closer and closer

  • 27:52

    KEITH PORCARO [continued]: to there is somebody who has legal duties who'sresponsible for this thing, the closer and closeryou get to what looks like a lot like justa centralized organization like a company, anyway.The thing about blockchain is that we'd like a copy of--if everybody had a copy of the database livingon their computer or their machine,and it needed to be good about keeping up to date.

  • 28:13

    KEITH PORCARO [continued]: Those are the solutions that I thinkblockchain is really good at.The solutions that trusts are really good atis how do we help organizations sort of deal with risk and dealwith creating sort of expertise for maintenancein the event of failure?So I have other complaints about blockchain,but that's after a couple of beers, so that's fine.

  • 28:36

    KEITH PORCARO [continued]: It's a good question.I think there are two pieces to it.There's one piece around sort of success storiesand the second piece around financial incentives.So the first piece, you're right.There are successful examples.They're good examples.I think that my argument isn't that this is a replacement.It's that when a project gets open sourced,and it gets picked up again.It's a little bit like whether that happens

  • 28:58

    KEITH PORCARO [continued]: is sort of like a lottery ticket.We tend to read a lot about the people who win the lottery,but we don't read as much about the people who don't.And so I think this is kind of a complementary step.I'm not saying don't open source it and put it in a trust.I'm saying that what a--what an organization might do is open source itand also put it in a trust to make sure that somebodyis taking it forward.And so I think to your point about markets, the reasonthat I'm talking about it in terms of public interest,

  • 29:18

    KEITH PORCARO [continued]: I don't think a company would care about a trust.I think that might be a violation of bankruptcy law,anyway, is that like we form charities,we form public services because there'ssome kind of market failure.And especially when you're working with nonprofitsor with specialized communities, theymight not have that sort of technical expertise.And they might not be that right match of capital and talent

  • 29:39

    KEITH PORCARO [continued]: and need.And that might come eventually.It seems like bad form to wait for itlike the market inefficient.If the market is efficient but it's not alwaysefficient on time, if that makes sense.

  • 29:54

    LOUISE COADY: How do you make sure the right projects geta trustee?

  • 29:59

    KEITH PORCARO: It's a good question.You try really hard.I think I would frame this as a product a little bitlike insurance, in that case.Not everybody has it.It's helpful in the event that you need it.

  • 30:14


  • 30:15

    KEITH PORCARO: Yeah, it's about sort of thisis about kind of another tool to catchwhat falls through the cracks.

  • 30:21

    LOUISE COADY: How does this translate into the real world?

  • 30:24

    KEITH PORCARO: I think the translationis that it's already happening.So what I'm talking about isn't a new legal structure.It's actually one of the oldest legal structures in the world.It's origin I think is horrificallyduring the Crusades.When people who held land in the UK were going off to fight,they left their land in the care of somebody, a trustee,who was sort of their agent.

  • 30:44

    KEITH PORCARO [continued]: Their duty was to care for it on behalf of somebody else.And so really what we're doing hereis adapting a structure that's alreadybeing used to care for land, to care for other people's money,to care for a range of assets and adding a couple of tweaksto say that we should use it for digital assets.And so in terms of enforcement, really, you've got--

  • 31:06

    KEITH PORCARO [continued]: you've got kind of similar rights in a court of lawif you really need to do it.A trust isn't going to prevent your sort of disasterlike argument or lawsuit from being a mess.It's just going to sort of frame the mess in a different waythan a contract might have or a bankruptcy might have.Yeah, so the question is, is therea risk that a trustee will kind of not serve the community's

  • 31:28

    KEITH PORCARO [continued]: best interest or make decisions that withhold data from the--from the community?Is that a fair representation of it?

  • 31:33

    AUDIENCE: Yeah.

  • 31:33

    KEITH PORCARO: OK.There's totally a risk.You're dealing with people.Sometimes people are bad at things.In trust law, what you would do issort of take advantage of sort of the mechanisms of trust law,which is that a beneficiary of a trust-- so a memberof their community can challenge the decisionof a trustee in Chancery Court.And so that's what you would expect that somebody to do.

  • 31:54

    KEITH PORCARO [continued]: And if there were really lots of kindof-- if this was a sort of a pattern of behavior,then you start going after the trustee for dereliction of dutyand then for breaching their fiduciary duty.And then you try and find a new trustee.So the nice part about it is that it sort of makesit not easy, but it makes it relatively straightforwardto swap out a new trustee if it turns outthat that's what you need.

  • 32:16

    KEITH PORCARO [continued]: It's totally at risk.I don't yet have a legal device to prevent peoplefrom acting poorly.Question is about sort of the internationalizationof this concept.I was going to say her wait how long you stand up here.So this is like a side note for law and notary in the US

  • 32:39

    KEITH PORCARO [continued]: and the UK are what's called common law jurisdictions.And the idea of a trust is based on the ideathat ownership in something is actually a series of rightsthat you can separate out.So like I can have the ability to control something and ownit, but I also have the ability to benefit from it.And you can split those two things up.In civil law countries like France and Spain and Italy,that concept doesn't exist.And so the idea of trusts doesn't really exist.

  • 33:03

    KEITH PORCARO [continued]: But it's kind of getting there because the sort of againbenefit of piggybacking on a solutionthat rich people use to hide their moneyis that there are rich people in lots of countries,and they have need to hide their money.And so in Italy, there's a really confusing structurebecause this is how Italy works is that you can set upa trust that has Italian law for all of the bitsthat aren't related to the trusts,

  • 33:23

    KEITH PORCARO [continued]: except for the trust bits, which can be English or Jersey law,which doesn't make any sense.And in France, there is one structurethat's like a fiduciary structure that they've set up,and they get really angry if you call it a trust.But then they're also importing some bits of trust law, anyway.And there's an international conventionthat's signed by maybe 12 countrieson trusts that I think starts to get at some of this.

  • 33:47

    KEITH PORCARO [continued]: I think that there are interesting sortof unexplored uses of kind of trusts thatgo across borders, especially in countries where governmentsmay be repressive.And people may be running technology programsbut projects that may be or have sets that are kind of put themat substantial personal risk.

  • 34:07

    KEITH PORCARO [continued]: And so you could kind of conceiveof a trust that breaks projects up into piecesand breaks an encryption key up into pieces and hideseach one in a trust in a different countryas a way to sort of prevent a really sensitive data setfrom being accessed.So the sort of short version is it's hard, and it's weird.

  • 34:27

    KEITH PORCARO [continued]: But it may be getting better.All right, so the question is about incentivizing peopleto care about stuff?

  • 34:35


  • 34:36

    KEITH PORCARO: Yes, this stuff in particular--OK, good.Other stuff, I'm not as good at.I think that you're right.There is a sort of incentive issue here.And my hunch, my suspicion is that you sort ofhave to attack it from three different angles.Like one is from kind of a challenge

  • 34:58

    KEITH PORCARO [continued]: to sort of mission-driven researchand mission-driven programs.Like if you're really trying to do thisand if you're really trying to achievesome sort of public purpose, then like thisis the way to sort of further it even after you lose interest.The second is, I think, like directlyfrom a financial incentive perspective.You need to make this inexpensive.You need to make this sort of like fire and forget.

  • 35:20

    KEITH PORCARO [continued]: And so the idea that you could dothis kind of archival cold storagefor fractions of a penny per gigabyte, which is possible.Then you could say, well, put it inand then you can stop paying for itas soon as you lose interest.And then we'll hold onto it for a while,and it's relatively cheap to hold onto it.So like what you're paying for is notthe sort of paying for somebody to find a new home the instant

  • 35:41

    KEITH PORCARO [continued]: we go away.You might be paying for somebody to hold on for another decadeuntil researchers come back aroundand decide they really look at this data again.And then the third, I think, is that you need to create--and this is, I think, part of our mission as well.So you need to create a culture where this is somethingthat you just do--like there are lots of default actions in an organization

  • 36:02

    KEITH PORCARO [continued]: or a project.You get liability insurance for directors.You get liability insurance in case people slip and fall.They're just kind of the responsible practices.And so I think the point that we get to hearis that this is a responsible practice.

  • 36:18

    LOUISE COADY: So can you get the fundersto like if you put pressure on the fundersto mandate this as part of giving fundinggrant or something that help get it to be sortof normal, required practice.

  • 36:31

    KEITH PORCARO: Yeah, I bet if youwent through some foundation, like grant histories,and found how many times they funded the same type of projectagain and again and again and again, that might sort ofturn the light on.And the foundation-- they probably won't at this stage.But at some point, they might consider actuallybeing the fiduciaries of this project--the idea that I'm putting money into the worldto make something happen.And I should be responsible for itother than just sort of hoping that it learns to fly

  • 36:54

    KEITH PORCARO [continued]: when I throw it out the nest.[APPLAUSE]


Keith Porcaro, fellow at the Harvard Berkman Klein Center and co-founder of the Digital Public Project, discusses some of the current risks to public-interest technology and how these can be reduced, including the relationship between business models, public-interest technology, and failure; the limitations of open sourcing products; effects of data silos, privacy protection, and data breaches; application of fiduciary duty to public-interest technology; and the benefits of a digital trusteeship program such as Digital Public. A question-and-answer period follows covering data trusts versus data repositories, deciding on a trustee, blockchain within data clouds, and incentives for establishing public-interest data trusts.

Looks like you do not have access to this content.

How Technology Fails Us, and What We Can Do About It

Keith Porcaro, fellow at the Harvard Berkman Klein Center and co-founder of the Digital Public Project, discusses some of the current risks to public-interest technology and how these can be reduced, including the relationship between business models, public-interest technology, and failure; the limitations of open sourcing products; effects of data silos, privacy protection, and data breaches; application of fiduciary duty to public-interest technology; and the benefits of a digital trusteeship program such as Digital Public. A question-and-answer period follows covering data trusts versus data repositories, deciding on a trustee, blockchain within data clouds, and incentives for establishing public-interest data trusts.

Copy and paste the following HTML into your website