 Dipsy is just is one of the educational initiatives that the Berkman Center has We have a history of doing educational initiatives So here's a few of them the H2O platform And youth and media which Sondra Cortesi is the director of Copyright X and Digital problem-solving initiative So the digital problem-solving initiative is currently in its second year And it's an opportunity for students staff faculty and other people affiliated with Harvard to come together to work on some real-world problems You can find our website It'll be at the end of the presentation, but it's blogs.law.harvard.edu slash Dipsy And every week the various project teams write up a blog post about what they've been up to Dipsy has three goals The first one is to promote multi-directional mentorship and learning we believe that learners of all ages can learn from each other And we believe in promoting a community of practice to promote that Additionally, we're pursuing institutional and issue diversity through interdisciplinary work. It's a lot of I words And we are also working to encourage fluency and digital literacies As we know digital literacies are increasingly critical for today's students and young people joining the workforce How students actually have those opportunities to Explore those digital literacies is still a problem. We are working on but we're hoping that Dipsy is one of those offerings Although says its second year it's actually in its third iteration So last year Dipsy ran for a full academic year and this year we've split it up into two semesters So these are some of the projects from last semester and you can see that they pretty much range all over the university Some students were working on sexual assault some students were working on the Harvard food project other people were working with on an open access project and some of those projects continued into the spring and Those projects are the ones that will be presenting today, but we also have a few new projects such as a Team that's working with Professor John Hansen at the Harvard Law School on their systemic justice project As well as a couple of women working with Kate Cronteris who's a fellow at the Berkman Center on 21st century girlhood so Olivia's here to talk about the big data group which is Led by a professor Jim Waldo In addition to talking about the content of her project She'll also be talking a little bit about some of the meta questions are associated with Dipsy such as the skills or the structure of the program And I thought that it might be easiest for each person to present and then offer a few minutes for questions after each presenter So yeah, Olivia take it away Thank you. Hi, my name is Olivia. I'm a senior at the college studying statistics and computer science So today I'm going to be talking on behalf of the big data group It's led by Jim Waldo who's the CTO of Harvard, but also does a lot of work with edX and Harvard X Which is the data we've been working with and one of my teammates Hillary is here right now so the As as the amount of big data has been increasing There's this inherent conflict in big data between the utility of a data set And the privacy of the subjects who are included in this data set. So you can think about Whenever there's a data set that contains sensitive information. So this can be educational data sets This can be medical data sets or and a lot of other fields as well Researchers would like access to the raw data, right? They don't want anything to be censored They would love to know the gender the birthday As much information as they can about these people so they can make the most accurate conclusions, right, but then when you think about the data in terms of the people whose Information is in the data set. They would like very much the opposite, right? They would like as little data Given released about them as possible. They would like their birthday to not be the exact day They'd like the year or even not even that, right? So there's this inherent conflict between the utility of data sets and the privacy of data sets So this is kind of what our group has been looking at So the first useful thing to define in this context is what our notion of privacy has been so we've been using this notion of privacy called k anonymization and This means that each record in the data set must be Indistinguishable from at least k minus one other rows in its identifying features So let's look at an example and what this means in terms of the example. So if this data set containing nine people's records and their grades on a certain test if if this is the data set we're looking at right now This is considered to be k anonymous with k equal to three and the reason for this is that if the state of residence is seen as the identifying characteristic Then if I know that a certain person Bob is in this data set and he lives in Massachusetts I can only narrow him down to at most through one of three people Because there's three people from Massachusetts. So the minimum Number of times that someone occurs with a certain quasi identifying field is three in this data set. Does that make sense? Okay, cool So let's say now that I have an educational law that requires this data set to be five anonymous instead of three anonymous So now this data set is not able to be released to the public yet because I can narrow someone down to less than five people Which is bad for their privacy, right? So there's two things I can do The first thing I can do is I can generalize the values of the state in order to kind of combine the New York and Massachusetts value into one and If I do this now there are nine people with like a New England or eastern US value So this data set is more than five anonymous. So that's good. I can't narrow anyone down to less than five people The second thing I can do is I can suppress the values that are troublesome So I can take the three Massachusetts values and say that Since they're two identifying I just throw them away. So that's the second That's the second method that can be done and that's called suppression or deletion Okay, so now we want to look at this k-anonymization In terms of an actual data set in order to see the effect that k-anonymization has on Biasing a data set. So we've been looking at the edX data. This is a massive online open course platform There are a lot of courses from Harvard and MIT, but other institutions have also joined So we've been looking at the data from fall 2012 and spring 2013 for five different Harvard X courses so the educational law as Dictated by FERPA requires that this data be anonymized with k equal to five before being released to the public so This means that in terms of the quasi identifying fields which in this case there are six So the six quasi identifying fields are course ID, year of birth, gender, country, level of education, and the number of foreign posts So no combination of these six traits can occur less than five times So each each person in this data set must have a combination of these traits that occurs at least five times So interestingly When we looked at the anonymized edX data versus the true edX data we saw a lot of skew in the Certain characteristics such as grade. So we saw that Certain summary statistics like the mean were significantly different in the true data set than the data set that was being released to the public Because the ladder data set had been anonymized. So the big question that we're looking at is why this shift occurred So in order to understand why this shift occurred It's it's important to think about why a row would be deleted in the first place, right? So Going back to our definition of k anonymity a row is deleted if it's too unique in terms of its quasi identifying fields So a row is more likely to be deleted if it has a rare quasi identifying value or a rare combination of these quasi identifying values So this is the first thing that we wanted to look at. Is there a relationship between Variables of interest such as grade and how often certain quasi identifying fields that are associated with it occur and we saw from as shown in these graphs that More rare values of quasi identifying Characteristics are tend to be associated with higher grades So the x-axis here is the number of times a certain quasi identifying value occurs And the y-axis is the grade and we see that Rare values tend to be associated with high grades So this means that rows that are deleted tend to be rows with higher grades Which means that the anonymized data set tends to be skewed downwards in terms of grades So is there a way that we can We can kind of quantify whether this relationship between the Rarity of a quasi identifying field is actually what causes a data set to be More skewed So there are three different things that we can do to kind of explore this. So the first is increasing K So in K anonymity the higher K is the greater of a privacy standard There is so if we only require there to be three people with a certain set of quasi identifiers that's That's kind of That requires less anonymity than if we require ten people to have that right because that'll cause more people to have to Be deleted. So if we increase K, we can see whether the skew of the grade increases Another thing we can do is if there's one or two quasi identifying columns that are highly correlated with grade But other ones are not then maybe if we just completely eliminate those quasi identifying columns and don't report them at all We would expect the skew to also decrease And then finally if we just manually change the correlation of these data set of the correlations Between these the rarity of these Values with the grade then we would also expect to see a difference in the amount of skew in the data set So we look at these three different things. So first we look at increasing K so as we increase K the anonymity standard gets more strict and we saw that as As K became higher The mean grade became lower and the mean performance also became lower And this is exactly because of the observation we had before where rare values are associated with high grades and high performers So as we became more strict with the anonymization We saw in fact that our expectations that the performance of the data set would decrease did in fact decrease So this is this is in line with our expectations We also saw that as we increased K the activity level of the people in the data set also decreased The second thing we did was eliminate quasi-identifier columns So one of the six quasi-identifiers in particular was very highly negatively correlated with grade So the number of forum posts attended that the more people posted the higher a grade they would have and and so when we completely eliminated this column from the analysis and Therefore didn't anonymize on it anymore. We saw that the grade of the resulting data set was much higher than Then previously before and this is because we weren't deleting all of those people who were posting a lot of times and also performing very well So this was also in line with our expectations and then finally we manually changed the correlation between different quasi-identifiers rarity with the grade and We plotted the entropy for these data sets and entropy is a measure of Kind of how much data is missing from a data set So we see that as we made the correlation less and less negative. We saw that the entropy decreased, which is a Which is a desirable quality And we also saw that as we made the correlation less negative the mean grade approached the true grade and the true grade here is is represented by the horizontal black line at the top So overall we found three different confirmations that the rarity of a quasi identifying fields with a certain numeric quality in this case grade Did cause more bias in the anonymized data set. So this means This kind of identified something specific we can look at when we anonymized data sets in the future We can kind of do some analysis beforehand and say is Including this quasi identifier even worth it or is it going to cause too much skew in the data set? so definitely more More analysis can be done into how to solve this problem, but it's valuable to have identified it as a problem overall And then I just like to talk for a minute or two about our experience with Dipsy So it's been great To have this as like a overall structure that has allowed us to first of all meet each other and have a mentor Jim Waldo has been a great mentor. We meet every week He he works on the code as well alongside us just like another Another peer so it's been great to work with him and also to find other people who are interested in the same questions Yeah, yeah, my question would be how do you identify a quasi identifier? I mean is quasi identity is the elimination of of those quasi identifiers really a way to safely Deanonymize the data or if I combine the the data that is published in the end with other data sets that are Available somewhere can I still through correlations? Identify the people who were participating there. Yeah, so that's a great question So to answer your first question the definition of a quasi identifier is a column that can be joined with another data set plausibly in order to Re-identify the data set so that's that's why we anonymize on these quasi identifiers so by by requiring that there be at least K people with a certain Combination we require that even if I have this outside data set that I can join I will have at least K people I will only have a one over K chance of being able to guess that person So that's kind of the whole notion of why K anonymity is useful because even when you join it You only have one over K chance. Yeah. Yeah Yeah, how many students were in the data set does more students make it a lot easier to do this or Are you still have the same problems? Sorry easier to do easier to do our anonymization. Okay, so there were 440,000 rows in the original data set that doesn't necessarily mean 440,000 students because some students take multiple courses but the so I'm not sure if this exactly answers your question, but So the more students there are the less of a percentage will be Will be eliminated due to anonymization just because there's more people So it's more likely that there will be at least five people with a certain combination So definitely we tried doing some analysis on smaller data sets because this Anonymization process takes a while it takes about an hour to anonymize a data set So we tried doing it on some small data sets But these small data sets like the number of the amount of representation of these class identifiers is so much smaller that you end up Like chopping about like 60 or 80 percent of your data. So actually more data is better in this case Because we're looking at like K as a count But you do have the trade-off of more rows taking a longer time to analyze Examples of posse identifiers and in particular ones which have high relationship to grade Yeah, so the five other quasi identifiers in this case. We're all kind of demographic. So Their education level their age their country Gender And the number of forum posts was an interesting choice of quasi identifier So the only reason that that was a quasi identifier in the first place was because these forum These forums were online and publicly accessible to anyone so anyone could easily write a scraper and then Just count how many times each person asked a question and then they could plausibly join that with their identity So that's the only reason it was a quasi identifier if the if this forum was not publicly accessible then it wouldn't have been but It makes a lot of sense that this This quasi identifier was kind of unfortunate because the more times someone posts the more engaged They are right and the better they're likely to do So I think this was actually like an odd situation that there was such a correlation because you would expect there to be a Correlation between such a quasi identifier, but I think that in most in most cases like Like medical data sets, I think there's not such a strong correlation The only other like kind of plausible thing that I could think would happen kind of often is for in medical data sets for Certain conditions to be correlated with age Such as heart disease So but I think in general this was kind of an odd case for education data Hi, everyone, my name is Jen creva. I am in the master of design studies program At the graduate school of design and my focus is art design in the public domain and I am on the safe campus Dipsy group our mentor is professor Diane Rosenfeld who is at the law school Over the past year 64 US universities including Harvard have come under scrutiny by the Department of Education For the mishandling of sexual violence and harassment complaints Well policy reform as required by the DOE and judicial procedures related to dealing with sexual harassment cases are important Societal views and individual behaviors must change as well in order to prevent these incidents from occurring This can begin with bystanders an often overlooked resource and essential element of safe public spaces Who can serve as public intervention mechanisms in the prevention of sexual violence and assault a role supported by multiple studies Which I can provide for you if you'd like We're creating a mobile application that builds a digital coalition of bystanders to empower students to seek help in uncomfortable situations The ops working title is bonobo app and takes cues from the unique social structure of bonobo monkeys Bonobo successfully create a community network of a defense in which victims of aggression are defended by his or her Group until the situation dissipates The bonobo app allows students to take control of their college campuses by taking a pledge to aid classmates in times of need The bonobo app would have dual functioning modes One for if the user is moving locations like walking home and one for stationary locations like being at a party For example, if the user is walking somewhere and would like to have virtual company The walking route can be shared by activating the walking option making the user's location visible to selected contacts Or if the user is at a party and would like selected contacts to know where she is in the event assistance may be needed The location can be shared by activating the destination option and can also be entered in the chat function on the app In both modes the bonobo app produces a network in the virtual public That will be called upon to act within the physical public should the user need assistance This social network is based either on the user's personal contacts or his or her geographical location Thus creating a coalition of contacts both known and unknown to the user Although there are a number of personal security apps already on the market none are compatible with the unique social and spatial dynamics of Harvard's campus or of college campuses in general Circle of six panic button and be safe all create digital networks of support But they are based solely on a user's personal contacts kite string restricts the number of emergency contacts and the number of times you can utilize it each month without paying for it Guardly does not create a network of support rather it connects you to the police by pushing one button instead of dialing 9-1-1 While it is important to call upon those you know and trust to come to your aid in times of stress The bystander network has even greater potential when brought in to include those within a localized vicinity of the user Harvard students are independent self-sufficient and involved in various activities and organizations across campus Most of the time they're not in the same geographical proximity as their best friends and emergency contacts The bonobo app will expand the bystander network to include both trusted contacts as well as others who are also signed into the app And are within a specific vicinity that is determined by the user It will provide incentives to increase bystander intervention And will also create a private support network for survivors and app users who have been in uncomfortable situations and are in Sure of what action to take We envision the app to contain the following key features Setting and tracking location using gps Excuse me Allowing to set a walking route or produce specific information regarding location For example, what floor or room of a specific building the user is in Adding friends to the app making it a social network, but not social media And creating different levels of friends such as closest friends or initial contacts or people within a certain proximal location Allowing the user to set check-in properties and alerting the user's contacts in an emergency It would also have the potential to add specific information Like what you're wearing or the specific room you're in sort of like a facebook status But again, this would not be a social media app And allow the user to message their top contacts And you would also be able to create personal privacy settings and determine who the alert is sent to whether it's friends proximal people or a mix of both We plan to conduct focus groups with various stakeholders on campus and foresee having multiple rounds of these focus and testing groups Throughout the development of the app These groups will consist of several different samples of harvard community members so as to cover many angles Of the app as possible Currently some of the specific demographics we would like to target are male final clubs and fraternities female final clubs and sororities Graduate student men and women freshmen and people involved in sexual assault prevention advocacy like our harvard can do better and harvard students demand respect These groups all constitute different parts of the larger harvard community and therefore we expect to get different constructive responses from all of them We've already been in contact with several representatives of these associations and plan to conduct our first round of focus group sessions soon As we glean information and feedback from the focus groups. We will continue to refine the app's functionality and aesthetic The bonobo app would only be available to harvard students upon its release And users would be required to sign in with a harvard university id number Thus allowing for the pull of data That would show how many people did not check in how many bystanders dissolved uncomfortable situations How many times campus police were called etc? By working through this project as part of DIPSI we have been provided with a platform On which we can make connections and build relationships with groups and individuals across campus who could be important stakeholders in the project We have also had multiple opportunities to get feedback during different stages of the project and in varying capacities From one-on-one meetings to large group presentations like this Because of that feedback we have learned how to mediate multitudes of suggestions in order to maintain the Original goal of the bonobo app We have also been able to think critically about the larger issue of sexual harassment and its causes and have made connections between this issue and products that are currently available as solutions The DIPSI community has provided us with the virtual testing ground to flesh out our proposal The flexibility we have been afforded in combination with the support from the DIPSI staff have been key factors in working through our bonobo app project Thank you Any questions? no As far as like the title nine coordinator in my office is you know We just added the link to the harvard mobile app the share website and there's a bunch of stuff already in the work So yeah, we've been working through a lot of those contacts through our mentor dr. Professor Rosenfeld And we're in the very initial stages of making those contacts and doing the focus groups But we'd be happy to be able to hold people worried about this. That'd be fantastic. Great Sure for questions. I'd ask what the status of this was and is it for Multiple platforms or just uh apple platforms? We're trying to figure out what the best platform would be. We'd probably start with apple Initially, and then figure out how we can transfer it to android And the status of it is the very beginning of Development is not really starting right. Okay. Yeah Yes, we applied to both the dean's challenge and the president's challenges And so we'll hear in a couple of weeks to find out if it's something that We have one and they would be able to help help us actually build the app Do you see what would be the motivation of people to join? I mean not potential People who are in a vulnerable situation that might want to join but those who might be Defenders how you motivate them to join the network? We're trying to come up with an incentive system so that if you Do help someone who hasn't checked in or has found themselves in an uncomfortable situation You receive some sort of a incentive whether it's Points that could be turned into a gift card or some sort of harvard merchandise Something like that, but we would incentivize it to get people to sign up. Yeah I'm curious how obviously it's tied to harvard ideas Serious to preventing bad actors from from five seniors trying to get the bad actors and I wonder what other mechanisms you have Yeah, that's a really good question and something that we've talked about a lot, but don't have an exact solution for yet We started with the harvard idea to to To hope that that could be maybe the biggest thing that would help But yeah, I don't even We don't really know that's that's all I can tell you if you have ideas. We'd love to hear them. Yeah I'm gonna ask along the lines of actually signing up with an idea and a lot of people In these situations prefer to be anonymous and remain anonymous. Yeah, so how do you deal with that with in relation to data privacy? Yeah, maybe that's something we need to discuss with the big data group, but um The harvard id would not be visible to anyone who is on the app. It would only be used to pull information So we could anonymize it somehow it possibly could be where you create your own username But you have to sign in with that id so that there's a way to at least track what's happening Does that answer your question sort of Sure, I assume that you've already dealt with the making sure that when somebody is no longer registered at harvard That they unregister they remove from your system because for instance, you know There's a lot of summer students extended students who come and go Who have id's or you know a summer or a semester or two and then aren't part of harvard anymore Right, that's you want them no longer to be part of your system if you're respecting it to harvard. Yep That's a great point. Thank you Super question. Yeah, I was wondering um You said you were gonna have so much just I don't know if you've actually had any yet I was wondering what student response would has been so far on just the fact that you're kind of tracking your data And they're not going to class they're going to parties where there might be tense social due to Things that you might may not necessarily want people to track where they're going at specific times Right what their concerns have been We've reached out to people and different associations to find out if they would be interested in being part of the focus groups So we haven't gotten into that level of detail yet But it's a really great point and it's something that we've had a lot of talks about But I think a lot of the capability would be up to the user So you could enter as much information as you want to but then it could have It could have effect on how quickly help could get to you So there's a there's a trade-off, but it could totally be up to the user I'm a lecturer and it happened once to meet one of my students during her studio brought period was assaulted And the first thing they took from her was her cell phone. Oh sure So in emergency cases So the the idea behind it is that you would set as the user You would set how often you would like it to remind you to check in And so if you don't check in with it after a certain amount of time it will Notify your contacts. So if your phone is taken it can still work A little bit we're trying to figure out what the the optimum time is between When you don't check in and when help is notified. Yeah This is to a certain extent responding to other questions as well, but Further questions along the line of incentivization It seems like I mean, you know As you say part of the design space that you're working within is the university campus and the unique kind of social Norms that obtain here What's your thinking about how to integrate the app into social life You know, not only to incentivize people once they've joined up but to Notify people that this thing is possible and you know, just maybe to a certain extent Explore the shift in perception that would seem to come along with you know before An event takes place being prepared to act in this way. It seems like You know, it's one thing to be presented with the issue, but it's another to say yeah, I'll step in And and I raise my hand to them. So, you know, what's your thinking about the kind of How you get uptake? That is something that is being dealt with in these cases all Over the place and I don't think anyone really has an answer Something that probably doesn't specifically get to what you're asking that we've been thinking about is hopefully pairing with orientation during freshman week at campus and Letting them know that this could possibly be the the official Harvard app that you could join That doesn't shift the perception but There's a group on campus called the Harvard students demand respect and they're working with training coordinators at the graduate schools. I know And working on Doing new training situations and bystander intervention could be a big part of those and that could help shift the perception as well I know that doesn't really know, but it's part of the broader social space and even the pedagogical Mission of the institution is in are entailed in in this kind of project. So it's good to hear how how you put it on the map Is it work? Good. Thanks Thank you Okay, okay So hello, nice to meet you all. Um, I'm joe steel. I'm one of the kind of founding instigators of uh doc shop, which is an interactive documentary workshop. It was kind of a two-part Provocation that came to us from dipsy And also from meta lab In terms of like troubling this idea of interactivity and documentary. Oops. I suppose I had to go to my slides Is this a full screen? Yeah, so this is uh for lack of um, preparing a slideshow. This is just kind of what we had submitted as a Proposal for the uh, Harvard innovation lab design deans design, I mean, sorry deans cultural entrepreneurship challenge So we thought this is kind of like an interesting way of framing the question of interactive documentary and and Our initial conversations were came out of This idea of like workshopping work by artists and and documentary filmmakers and makers and various people that would come to us But also wanting to get our hands dirty and really work with media and stage encounters with media Perhaps in convivial environments where you get to talk to people and have conversations So We Worked with this, uh, here. I'll go to the artist. So we made a prototype With uh, the egyptian lebanese artist lotta veladi who's currently a open doc lab fellow at mit and We we built this room, which was a sort of like It was meant to mirror uh terrier square because she has this incredible archive of images and Videos and you know some things from the public domain and some things that were from her own archive and we we staged this encounter um in In is kind of like a theater in the round where there's no seats and then the kind of traditional stage and kind of broke outside of that Play does cave, uh, you know conception of the theater is Sitting down. It's not participatory The information goes one way the narrative goes in one direction And just made it so that there was a conversation and then um conversations among the media happening at the same time Um, so it was really useful to prototype this both in lotta's own work with her archive and then in our work as doc shop and kind of defining her problem moving forward Um, and then at an outcome of this is there there's going to be a curated Kind of conversation uh this semester, uh, which we're working with meta lab and and with lotta to stage and It involves some federated research and combining several different archives about terrier and about, you know, the 25th of january and egypt generally And uh bringing together These archives so that it could be searched from a single point and that was that kind of giant like went well with um some work that meta lab had done with archives federate, you know federating archive Which had to do with the tsunami I think and yeah in japan Yeah Yeah, and and this this connection and and some of these, uh, you know other things that lotta Will be participating in including in dc kind of came out of this and her work with this archive and um, there were some articles written about this piece and Yeah, it's it's kind of amazing. Um that we're able to do it in the semester and and kind of make some waves and Start start working with the media So yeah, so we'll have to kind of refine this oops to a pitch but The the sort of pro we we sort of framed it as problem solution and then our formula Uh, so like the problem is that university life is siloed and all these really interesting people come here and they may not produce work I mean they may conduct research and then go on back to their homes and You know make make stuff once they get there but we kind of thought a way of uh Solving this problem and kind of moving the paradigm from competition to collaboration would be to Uh create a space where like a physical space But also a space where people could meet and talk about interactive documentary talk about archive Talk about memory more broadly and and producing narratives and have those kind of organic partnerships that would happen amongst graduate students undergraduate uh faculty fellows and the and a broader public And our kind of like formula is that Our team is very skilled with translational strategies and you know kind of Bringing something from the idea ideation phase to You know a prototype for you know this this room that we created was a prototype. It's essentially a paper prototype or a model Um, and we were able to get feedback from real audiences Based on that. Um, so we have a kind of a multi disciplinary interdisciplinary team of artists technologists archivist journalists and uh humanists and historians and you know, we're adding to that uh and uh, we'll we'll see I have we have to refine this to like one elevator pitch and so the uh The committee looking at the i-lab applications won't They'll only look at the pitch so It'll be i'm interested to hear your thoughts and you know What do you think about the the way that we're framing this problem and then Um, yeah, I guess to answer your question. How did dipsy Help us realize this and it it brought us together even though the question was incredibly open-ended. Um and it uh Allowed us to support Lada in realizing this kind of Prototype with the archive and then it's allowed her to reach out to like institutions and and to kind of start talking with curators and other practitioners and um other folks who have been working with archives across a number of different disciplines and then also There's another point I was going to make Yeah, so we'll continue our collaboration with her at least till this semester in addition. Oh the challenge. That's it So, what's the challenge? So, how do we maintain the identity of doc shop? Well supporting one artist In a really engaged way. Um, so that's something that we're negotiating is like how do our How do we keep our identity sort of from getting too entwined with the work of this one artist? And so we're we're trying to carve out time to do the workshopping and to do the the network building and the collaborations with other institutions Well, at the same time fulfilling our promise to incubating her project Is it part of your concept of creation of a library that you would maintain as well of the work that you do? Um, so it's a it's a kind of two part Uh, I think the archive and the library will exist, but it's not a physical library, but but but yeah Data Yeah, that's that's that's something that make an effort to maintain everything Yeah, that's something that lada is working with. Um, I guess we were kind of interested initially And how do you take archive and then make near like dense complex narratives out of that or to To kind of how to bring together narratives, but doc the mission of doc shop itself is to be The this two-part thing of uh, like a table like this where we sit around and workshop Interactive documentaries, but then also incubating one artist project in a really in-depth way And the the purpose of the workshop is to kind of build our vocabulary and add to the body of knowledge of interactive documentary, but also just to kind of create those connections and and It it also enriches the experience of the artist that's in residence at the time because They might meet people or frame their thinking about archive or about Documentary in a way that they hadn't thought about before So I don't know does that Kind of kind of answer your question. Well, it tells me I'm still confused What it tells me I'm still a little confused. You said you were going to work on your elevated pitch. Oh that yeah, that's that's work in progress Sure, sure um Yeah, so could you provide any practical example how this works or any story behind this just a picture of I don't know just a case study or To illustrate this To illustrate the kind of like a stem just we could do like a step. So one person who's creating a portal to To bag dad for instance like they want to create a portal to the american university there And so they might come in and say oh here are the tools that I've been using and then here are the problems that I ran into And then between the knowledge that we have Between metal lab and Dock shop we could provide some directions and some resources that they might go to And so that's one part of it. And then the other anecdotal part is that we You know, there's the artist Residency program and so that's kind of separate so the So lotters project like she came to us and We built we helped her to build this prototype and then as a result of this She's talking with curators. So that's the anecdotal part that goes with The artist incubator and then the the other part is essentially like a small think tank like a room like this I don't know Matthew. Can you Yeah, sure. I mean just to maybe flesh out a little bit more kind of expository sense of Of what dock shop has been about and sort of the narrative it emerged out of an interest That that we have at metal lab To ask questions about the use of interactive documentary in a variety of situations journalistic situations scholarly situations and In in the arts as well There's a sense that this is a kind of burgeoning You know field of of practice And while there's been a lot of thought given to sort of tools for making interactive media platforms You know programmatic tools technologies There's there's not been a lot of attention not as much attention has been paid either to Sort of the critical evaluation of these kinds of projects How they how they land in journalistic circles and scholarly circles in the art world and the kind of You know the kind of aesthetic dimensions what makes them good what makes them What attracts us to them? It's a space that documentary filmmakers in particular Are feeling both attracted to but also nervous about because they feel a kind of radio You know video killed the radio star kind of Concerned that as interactive media come along will the documentary cinema experience be Beaker tailed. So we began with these kinds of questions in mind Attracted a group of students who are interested in making interactive media And began to workshop their projects into that mix Lara Valati came along as an artist who's a resident actually in the open docs lab at MIT right now Her project really is focused on this. She's a she's an artist With with you know who's shown working galleries around the world. She was also deeply Involved with the tar square uprising In the midst of which she created this archive. So the question of the archive Came into the doc shop It you know we encountered it kind of Emergently it wasn't initially part of it wasn't in the crosshairs. You're interested in interactive media But of course interactive media often crucially depend on collections of media The japan disaster archive that that joe mentioned is one example of a project that we worked on with the Rye showered center for japan studies here at harvard To create this this kind of federated media archive of the 2011 disasters the kushima Daiichi plant disaster in the tsunami that caused it those were Archives that have been created by journalists and media organizations As well as by scholars and individuals By social media platforms and finding ways to preserve those kinds of Media objects are very challenging. So with respect to doc shop. We've kind of carved that that dialogue off That's a separate conversation to this prior conversation about Why we want to do interactive media in the first place to try and answer that question the group devised this notion of a prototype with Lara You know we initially wanted we were thinking in terms of interactive documentary projects that are typically As they often are browser based some precedents include projects like welcome to pine point bear 71 a lot of projects come out of the the interactive documentary lab of the In canada But also snowfall and a whole host of projects that New York Times has pursued as well as scholarly projects So these are often browser based and we began to wonder what it would be like to animate interactive media in physical space Something that happens in the art world quite commonly so to bring a lot of In terms of the translational thinking that joe mentioned to bring a lot of the energy of the art world of performance of sort of convivial encounter in space To the question of interactive media became the kind of Experiment that Lara Balati's project with doc shop was about So the idea being can an artist come work with this group of students who are also designers and technologists To realize a project Does that provide us with a prototype for a kind of iterative model for For cultivating and incubating such projects in the future and and that's what the The ask to the cultural entrepreneurship challenge is Targeted at is that question is there a utility in a space where artists media makers filmmakers could come Meet others with you know, independent skills and sensibilities and devise You know workable impactful Interactive media projects in a context that also attends to their kind of aesthetic qualities I probably never heard about I don't think we have similar projects in Russia. So I do You'll have any other examples. Well, it's a big it's a it's a growing it's an emerging discourse and it's probably worth a lunch talk on it Yeah, it is be sure How would I find out about more events like the one that happened last month in december? Oh this uh You can you can email uh, you can email me. Um Hold on. I'm going to try to pull it. Is there a web page that we have a splash page. Um, how do I uh Yeah, it's it's just a web page a very basic web page And all of the projects um also talk about their various events and Space Just dot space. It's dot space. Yeah, it's like I didn't know there was a space domain. Yeah, we thought that was a fun one. Uh So, yeah, then, uh, you can just contact me. Let's see Trying to think of Yeah, my email is on the side. It's uh, yeah, you can just I'll give you my crap All right, let's go definitely you can check the doc shop dot space website for more metal lab dot harvard edu and the dipsy Yeah, yeah, right, right. We'll make ourselves enough. Yeah