 Ac mae'n ddiogelio i fyny ni rwy'n dod, a ong yn cyfreidio'r ystod i vaeth yn digwydd, felly, os drafodd o'i dechrau, fe ydych yn frysgol. Llywodraeth ni'n ddiddorol o ffocüsol o'r tref, o'n ddiddorol o ffordd o dechrau a'r mynd i rydym ni'n llei i'r ffordd o'r ddweud yn Hopebryd. Rwy'n gweithio i'r ddweud yn gweithio yma. Rwy'n gyfreidio'r hynny yw unrhyw dros cynnwys of getting around the problem. Some of them are just problems that are open for discussion and a lot of this is reflecting what's happened in the other talks today. So I'm going back in time a bit with this one. This was a piece of work as part of the Seaware project and our goal in this was to try and come up with something similar to a credit score but to do with your energy consumption. So credit scores take data from various different sources, different credit cards, different financial services you've signed up to and try and give you a day long or a kind of holistic view of your spending. With energy we have smart meters coming in now. I'm sure some people in the room have smart meters. I was talking about my smart meter and how rubbish it was earlier but the meter we have energy tends to be for one place. We have our home energy monitoring which was great, tells us about home potentially. Some of our workplaces have energy monitors but these are not brought together in one place. So the point of this project was just trying to bring together all that data in one place so you can see your route through the day. This wasn't because we necessarily thought it was going to work it's just we hadn't seen someone else do that so we want to see what it was like to try and bring that data together. In one hand that was a technical challenge so some of us on the team were computer scientists so I'm a computer scientist not necessarily a very technical one by am one. We had sociologists and psychologists in the team as well so part of it was a technical challenge and part of it was a social science challenge to see what happened when we put that data from one place and see how people reacted to it. So it wasn't a very big study this was really a test for a PhD student but it was hard to bring that data together so it meant instrumenting a few people's houses it meant getting access to data from their workplaces and it meant instrumenting some workplace as well to try and create a place where all that data could actually be harvested. It's very hard to do. We found some things out and again luckily I'm not going to focus on the results so much today I've put the links to the papers in these slides and I've sent them out at some other points so hopefully they're easy to go and find if you want to see the results. Some of it was as you'd expected so the top graph shows for one person I think where they spent their energy during the week so that the red spaces in there were communal spaces like workplaces blue spaces were personal spaces was their home and green spaces were spaces in between so this was travelling this was shops things like that so they spent less time in communal space at the weekend so they spent a lot of time at home. The graph below so I've basically said time spent in that top one not necessarily energy spent the bottom graph shows energy spent and the amount of energy you spend in different spaces is quite different and I'll come to that in a second so this is it boiled down to pie charts just to make it even easier so time spent on the left a lot of time spent in personal space this is what it looked like when we divided up the energy based on what space they were in so in the workplace you're using a lot more energy this is if you take the workspace's entire energy and divide it up by the number of people they're in that space now obviously in a workspace you've got a hell of a lot going on compared to the home you've got a lot of these people were office based you've got things like servers sitting there in the background you've got air conditioning systems which don't necessarily have a home you've got a whole load of other systems that you're using as part of your daily work that if you just purely divide up by the number of people you end up looking like you're spending a hell of a lot more energy in those spaces doesn't necessarily make sense but the whole point of this was just to see if we just divided up really simply energy what's it look like on paper because in reality that's what a lot of these services into each other might just do they might just divide things up because that's the easiest way to take this data so you can imagine the kind of things that led to we get nice things here like this was a quote from one of the participants so he said looking at his work consumption he may as well give up a home he's using so little at home it doesn't really matter he can do what he likes there maybe he should make a bit more of an effort work maybe not because it's so much energy consumed at work I can't make possibly an impact on that um it's like this I mean again people will recognise this there's a done work in workplaces people don't feel like they have a lot of power in the workplace so people quite quickly worked out they have a lot going on around them lots of systems things they have no way of influencing we had people working in offices where they couldn't touch the thermometer they couldn't change the thermostat they cannot home they can't at work they're not instrumental at work so despite the fact a lot of energy consumption is going on there I can't do anything about it so why bother why care what we also did was have a few people in the study that were friends of one another um and what we let them do was to see their impact on their friends home so they could see how much energy they spent when they're around each other's houses so we again in the behind the scenes got them to do a time you start and got them to divide up and say when they're at each other's houses and we were kind of intrigued to say well your day long footprint doesn't just involve your home or your workplace it involves energy you're consuming other people's homes and workplace as well so if you go to a friend's house and you turn on you ask them to turn up the heating you're kind of responsible for that okay they're being a nice host by allowing you to do that but you're sort of responsible for that consumption um and people started to say that they start saying well someone's come around and left the light on in the toilet and they've gone to the kitchen and left that light on as well they're ruining my data they're increasing that and I really want them to be accountable for that data this was neat one as well um people started to make really I guess simplistic rational decisions about how they were spending their energy and started to try and draw conclusions from it so I guess my point with this one was we did this as a really simple experiment but people got quite invested in the results and started saying they would change their lives based on these results we kind of had to step back quite quickly and say and kind of explain the simplicity of what we were doing so these are very simple calculations and we're doing it to provoke responses but these are not things that you should be basing life changes around um and it's something that's come up again again in these talks is that context really matters and our lives are really messy and we do things that can't easily be explained in data or you can't apply simple functions to the data to make decisions based on and obviously the psychologists in the study were really intrigued to see where this data could be used to change behaviour and like I've just said on one hand it ended up being quite dangerous and you couldn't you just couldn't do things based on this you could people were starting to contextualise themselves but the amount of missing data or data that didn't make sense because of the way we'd processed it just undermined in those decisions they could have made um and I sort of mentioned this this is a really interesting part of it was we we kind of threw in the ability to see each other's data just to see what would happen um but turns out if you're good friends with someone you can almost read their data as well as you can read their own you or at least you think you can this is one thing we found we had a couple of people in particular who were very quick to say they knew each other very well and knew exactly what the data said and of course if you think you know someone well it it starts to reinforce what you're seeing in the data so if you think you see something you kind of change the way you see the data so that it fits your view of that person and this data was showing them entirely what they wanted to see and they would say exactly the wrong things they're making conclusions that we knew were completely wrong um but they thought this was a great view into this person's house as they expected it to be so I'll leave that one up is a study we did a while ago it wasn't very big one but it it it starts the story neatly um so we I guess thought what can we do what sensors what other data can we collect to uh regain some of this missing context that we thought we weren't getting in that first study so we moved like a few people did I think onto wearable cameras um so if you're not seen wearable cameras I don't know how experienced people are with these sorts of technologies in the room we used one called Autographer and that came from one called Sensecam which is from Microsoft at Cambridge uh and rather than me just talking about them these are some nice product pictures from Autographer a guy on the left is what they want people to look like when they're wearing it these two were the faces we got and people gave them the Autographers really um this was the wearable camera here so this is the Autographer and it is how it looks it's basically a USB stick stuck to a tiny camera and you could open and close the lens how you wanted to so this guy here's got his lens closed that yellow bit is the lens cap on it she's got it open there and the idea is you hang it round your neck or you clip it to you and it's got a PIR sensor there so it detects motion in front of you um it's gone accelerometer in it so by default they can capture a picture of what's in front of you every 30 seconds but it tries to use those sensors to be a bit more clever about it and save battery so if it's not moving if it's not seeing motion it captures pictures less often if it's moving a lot it captures pictures more often now this was sold this product was I mean Sensecam was sold as a research tool Autographer was them trying to clamber on the bandwagon of kind of quantified self and people wanting to log their entire lives it's a life logging tool really it's a consumer product where it was until it was cattened um no one really bought it as far as I can tell except researchers because it was slightly more well packaged than Sensecam and that's this sort of slightly devious thing about not actually being very visible so I know what it is because I'm looking for it most people had this on we could put them in a room and other people wouldn't really know what it was um it's kind of and the marketing around it was sort of done so that it's like it fits into your life and it doesn't interfere with things and people won't know and yeah it's slightly devious marketing around it um so we thought uh let's give that to a few people and see if it can complement time use diaries so we know time use diaries they're not necessarily 100 accurate um sometimes people invest again their sort of view of themselves in the time use diaries so when you look back at the end of the day and you're tired you think oh what did I do oh yeah maybe it sound nice if I did this so you kind of start to to to it no longer is objective so we thought why not put some of these pictures alongside the time use diaries so this is the sorts of pictures that we captured uh these are the non blurry ones hell of a lot of pictures are blurry it's not the best camera in Sensecam so you get hell of a lot of just motion blur and blackness someone this is the person that you like to pretend they read at their desk at lunchtime but actually played candy crush and things like that inside their book that's this person eating at their desk that said they took a break from work while they're having lunch they didn't they worked while they're eating lunch um just two ways that those pictures contradicted completely what people put in their time use diaries we gave people an app to go alongside it so we didn't do paper based diaries um well we did start off with but then we developed this app which you at the end of the day you plugged your autografer into the app did some processing across the pictures that it captured and the data it captured and then prompted you for what you were doing at certain times of the day based on what the app could see in the photos and in the data so it tried to do some processing to pick out activities significant activities different changes in the data and prompted you for what you were doing based on the pictures at a time of day so it's kind of neat the app can say I'm seeing a lot of motion at this time of the day pulls out the picture that's timestamped at the same time and said were you doing something interesting at this point in the day so we did that we gave it to people our focus was on uh hey just because the funding we got was on how we use technology at work and whether this was having an impact on their well-being and their productivity at work so we were using the the wearable cameras and the and the time use diaries as a way of studying that so we gave it to not many people to be fair I think some of us doing 2030 got that many images out of the cameras that wasn't fun we had a summer intern who had a great load of fun going through 160 000 images alongside the data and what she did she coded all those images and went through and tried to pick out what she thought the activities were going on in those pictures this was our this was our attempt at ground truth in those images but given the amount of pictures that were blurry and given the the kind of other factors that are going on that was quite hard to do so again I'm not going to focus on that on the the results I've just put them in here because they're quite fun so we only had people wear these cameras during their working days we said when you start on the road to work turn it on and when you get back on the road the end work turn it off so we don't want any of your home life we just want your working day here um I'm sure people can recognise themselves a bit in this so we found on average people had about an hour of non-work activities and this is everything from eating lunch to going to the loo to going to walk outside to have a smoke like some people did but 30 minutes that was digital in some way or another so that again as you might imagine that might be catching up with people on WhatsApp it might be doing some shopping on the internet now what it'll let us do is to try and cluster people into behaviours so we again not going to go into this in detail but we had we kind of made distinction between people that are grazers and we're constantly doing some sort of digital non-work and this might have been blended quite well in with their other activities so it may have been doing something on their phone while doing work on a screen we had people doing that it may have been people walking around sort of looking like they're taking physical break for again doing something on their phone versus people who batched their behaviours people that would not touch their phone until lunchtime when they would maybe watch a video on iPlayer and then go back to work it said they'd clearly segment their behaviour really quickly to us it seemed like what we considered traditional recovery activities they were disappearing people don't do them so much anymore so how many people in the room think they take a significant break away from all kind of technology and work at lunchtime anymore anyone think they fall into that category anyone take a real break at lunch yeah and the people that did kind of like it was a real life statement they were like I'm going to get away from this and I'm the person that does that most people that seems to have disappeared completely digital things become more attractive as day wears on you get more tired more bored with work you start looking at the internet more you start faffing around with your phone a bit more that's just natural it happens you had interesting correlations between our measures of well-being and how much people did non-work digital things so the more you did the more constantly you were checking your phone again these are sort of intuitive that the greater fatigue you felt at the end of the day and people that spent a lot of time on facebook during the day people were not describing this as a break people were describing this as another type of work they're saying this is me managing my social network it's me making sure the kids doing what they should be doing it's me doing my shopping it's it's not a break um so again i'm not going to push on with this too much i want to get to this as a method um so what we were basically doing was we were turning our participants into researchers there were no longer participants we were sending people out with quite a sensitive device to catch pictures of the world around them this is no longer a study where they are kind of ignorant participants they're doing the work for us really um this forced us into doing a lot of things we don't normally do so walking around a workplace wearing a camera capturing everything that's going on is quite a sensitive thing you can walk into a meeting you can capture who's there you can capture documents on the desk in front of you you can capture things on screen so we had to have our participants knowing quite intimately what it meant to get informed consent and knowing when a sensitive situation was happening and that basically meant doing ethics training with participants because he quickly is moving a long way away from the the uninformed participant into being a researcher um and we had to be quite picky about who we were recruiting to see whether they were understanding that process and understanding why we were doing it something i didn't mention with the app was we kind of wanted to capture some of those images we didn't necessarily want 160 000 images except for the coding we don't want to keep those images they're not useful to us um but before we even got to coding it was the the participants responsibility to censor was in those images so quite often this was happening retrospectively and you could close the the lens cap and we were saying to people if you know you're going to go into a sensitive thing if you're on a meeting with people that that you're not allowed to photograph put the lens cap on we don't care we don't mind that we're missing that data it's fine for a lot of times they'd do it and then they'd remember five minutes later they've been in the meeting that was sensitive or hey where people going into the loo remembering they'd forgotten to close the lens cap when they did it there was a need to retrospectively censor these things that's a really big deal at the end of the day these people are taking tens of thousands of images on this camera to go back and start deleting those was onerous work but it was something the participants do not ask there was this real concern so as soon as people start to feel like they are a researcher rather participant and even if they're a participant i'm sure people have seen this in the room there's a concern that you're producing the data the researchers want is it good quality data is it showing what that what the researchers want to see in the data as soon as you start telling people what the aim is of a study are their head goes into that space and saying oh i should be adhering to these things people are wanting to find so this was a real issue for us and it it but it just meant we really had to embrace the fact there are researchers now not participants so in the end uh just another weird technical thing about these cameras the autographers had a usb slot on the top of them if you put your camera down on table someone else could pick it up plug it into the computer and they had all of your images there's no kind of encryption there's no password protection there's nothing on it it's literally a usb stick so one of our real concerns was people would put it down at lunch and then their whole day was captured there and ready for someone else to pick up and just have a a quick glance through um so we the the kind of coding app or the the time these diary app people had was sort of an encryption app wrapped up into that as well and we had to really get people into a habit of plugging their autographer into their phone and running that app quite regularly just so it encrypt their images over and over it seemed weird to me that wasn't built into the autographer it's literally just an open usb stick that anyone could pick up but that's autographers decision not ours so jumping on again this i think is probably the only thing that hasn't really been discussed today this was a more recent study i was only tangentially involved with this but was hearing the stories quite regularly so it's a nice case study this was a collaboration between computer scientists and social scientists to look at um deploying community-based wi-fi into a into a deprived communities the idea was they'd put together this uh technology package that if you had broadband in your home you could replace the route you had with this router and allow you to share your wi-fi with anyone else that's living nearby the idea was there were some people in that community that could afford a wanted broadband lots of people that couldn't afford broadband this was just a nice way to spread that uh digital capability throughout the community or at least that was the plan anyway um okay this is where so this is all about recruitment again we've heard some nice ways that recruitment has been done today this one probably wasn't well it definitely wasn't successful um so we just worked down the list so we had research students go out and knock on more than 2 000 doors in this community so it's a fairly big area um 730 of those were answered so people actually open the door and talk to them uh of those that that's not a bad number above 10% wanted in some way to be involved in the project um so actually more people asked sharers so we share us with the people that had broadband already and citizens were the ones that didn't have broadband but wanted to to you know start using it somehow um of people that then actually took in one of our hubs not very many that's fine 21 people uh obviously you have to have someone living living near you that wanted to use that wi-fi for it to make any sense so we ended up with 17 pairs of those and only two pairs actually took part in the study so we didn't have many people involved in this and out of more than 2 000 that's not great numbers but and part of the problem here was just it we sent people out basically cold calling on doors and that was a stupid idea in retrospect but it was a fairly short project this was the sort of interaction that happened i'll let you get to the end of that quote happened quite a lot and we we're talking 2017 in this case you'd think i don't know my my observations of the world now is that internet is pretty well embedded in in my everyday life i'd struggle to go without it now but for a lot of people in this community and this was a i mean nottingham doesn't really have inner city areas but it was an inner city area in nottingham and i think back on the previous slide does put the stat in there it's only about 70% penetration of broadband or any kind of internet access in that community so that's that's that's low of the uk now but not i guess not unheard of around nottingham that that wasn't the lowest level of penetration in a in a area of nottingham we didn't pick the lowest because we wanted one that we thought was going to be suitable but but still approachable so this slide's going to be offensive whichever side you fit on one of the main things to come out of this so we forget about the results because there weren't any results we know how to speak each other's language it works great this team didn't um so what you had was sociologists that were no longer being sent out to study they were being sent out to bug fix because this is the thing the sociologists had the great relationships with the a 17 people that wanted to be involved um we're learning about their daily lives we're learning about their struggles uh but we're having to do that alongside trying to fix their router and with all the will in the world we had some of our sociologists going out trying to kind of bug fix security certificate issues and things like that while listening at the same time to what was going on these people lives and it just wasn't working so the burden of that was really heavy and just made it not work other side things this is my side um turns out academic computer scientists don't know much about actually fixing bugs so academic computer scientists can talk at great length about these things and this is a neat quote this was from well this was from one of the computer scientists on the project and they were kind of trying to lay out why they shouldn't be responsible for fixing these routers and this was actually just trying to fix the and I guess everyone here probably knows this you put a wi-fi in your home it's actually really hard to know where you're going to get signal it's kind of a black art it's just it just is but the computer scientists were being held responsible for the fact that in this case someone next door couldn't share this person's wi-fi it's wasn't going through the walls and wi-fi just doesn't do that very well so they're kind of I guess their their response this was not all computer scientists the same even in computer science you get a whole load of people with different skills and no one has all the answers so we tried to break this down in three takeaways so the top one is I guess everyone knows this this is I think this is the Facebook saying that Facebook engine is move fast and break things that's their kind of philosophy why it's not their philosophy anymore because it'll be here but it was until about a year ago was that hey we're this cuskiness company and we'll just do things and some some good things will fall out some bad things will fall out and we'll keep the good things when you work with the deprived community that doesn't work you can't move fast and break things because hey you get kicked out pretty fast almost literally in some cases we've talked about kind of robust technology we've talked about needing to switch things on often people do that themselves often I think the work we do we put in technology that we think works fine in the lab sort of if we don't stress it too hard and kind of hope that people's houses will be that way they're not that way the other thing is that again this the kind of I feel like I assume now that I have all of these technical systems that work around me and they're all there so I can buy new products off the shelf and I have Wi-Fi I have a broadband connection I know it's got good uptime I've got my phone that I can connect to mobile internet and fix things if my broadband's not working there's hell of a lot of digital plumbing that goes around now to make these things work and there are people that don't have that digital plumbing around them so we were making assumptions like there was enough broadband around or there was always uptime or people had accounts on social media and things like that and it just wasn't the case here so we built all these assumptions into this technology that we deployed and then we're having to instill other things for people as well so we're having to go in and give them other technologies to fill in the gaps we're having to set up accounts of them they didn't have so we're having to do a lot of digital plumbing to make this work the final thing on this one so I said we had this sort of disparity between the two disciplines what we actually found was that the sociologist here was starting to buy into what the computer scientists were saying which is really dangerous on the best things on it on a cross disciplinary team is that you have some kind of scepticism of each other so you can say I have much more experience of doing field work your technology is not going to work it needs redesigning in this case that wasn't happening it was kind of getting sociologists that were becoming amazingly excited about these technical ideals that just weren't going to work in the field um that just wasn't a good thing and the sociologists ended up being as surprised at the technologist when it failed whereas if I think they felt they'd had that scepticism from the start they could have intervened much earlier and said this isn't going to work so saying no and saying you're stupid and saying you're impractical is not a bad thing in these projects I think and like I said I think it's one of the great things in having across disciplinary projects is you can be sceptical of each other so very last one this was us going back to energy consumption and actually before I go on to that the last thing about its previous approach I didn't even really mention how this was a sensing project but part of this was seeing how people shared this infrastructure but also seeing how the technology performed while it was sharing so it's one the technical side of this was saying if you've now got a in our ideal world 10 families sharing one wi-fi hotspot how well does that perform and having a discussion with people who are already hostile towards the idea of the internet about a hub that will monitor what you're doing and what other people are doing around you that again is another kicker for that initial discussion it doesn't work because it's about coming with a language in a way to to explain research studies to make it work so very last one c-tech like there's another energy project this was looking at energy consumption in workplaces and these are workplaces at least in the UK there's a it is regulated that that organizations with a certain number of people or a certain size building have to have half hourly metering so it's the kind of thing we're going to get through smart meters anyway which has been the case for commercial premises for a long time the problem with that is that really only mandates that you need one meter on a building of thousands of people capturing data every half an hour and if you have a thousand people's energy consumption on a single line on a graph you can't really tell much you can see when the working day starts and ends but that's about it so this project was all about trying to do some sort of social apportionment of energy consumption was to say right we've got that data great that doesn't really tell us much but can we get the people in the building start to break it down into manageable chunks and say who's responsible for what and how much more technology would do we need to put in to make that feasible so i'll see again breaking down a thousand persons data without any kind of extra help is really hard work but can you put in maybe one more meter or two more meters or 10 more meters to try and make that a manageable task so we took one building this is a council building it was an interesting building anyway so this is a picture of it behind so it's a very old building that's a listed building joined up with a very new building to make one new mega building so this was one of the sort of commonplace uk council things where they'd merged about three different councils into one and decided they need to expand one of the headquarters of those councils to make it suitable for all of them to live together so they've done this crazy new building that looked lovely but had one side that performed really badly one side that performed really well and for some reason it'd been given a energy rating of really well despite the building that was sort of leaking energy out of one half so they were having really bad performance in that building and we're going to lose their energy certificate because of it so they wanted us to step in and aid a research project would be trying to work out what was going wrong so what we did um yeah we're almost there so we kind of just showed them the day to start with and wanted to see whether if we get all the decision makers in the room it's not necessarily office staff but people were responsible for changing the systems in the building and deciding where people sat and what went on the building could they come up with anything so we took data from their building management system so again on the top we've already seen graphs like this so uh blue line is temperature outside green is the internal temperature and red is how much gas they're using so you can see when the temperature goes down they use more gas to maintain that nice uh static temperature which when you start to look at these you already wonder why isn't temperature really going down outside of working hours they need it on all the time but there's the sorts discussions we had uh we then stuck in a few meters and tried to sort of break down electricity consumption by part of the building and again just showed this to people uh it got a bit heated in this room so these are the people sitting around the table looking at these graphs and we tried to get people to sort of discuss with one another what they thought the energy was going on what was consuming the energy who was consuming it and just basically ended up in a big argument with lots of people saying what they thought consumed energy and other people saying you're stupid that makes no sense it's all about kind of these turn to a storytelling session but the end of each story someone said you're insane that doesn't consume energy there's lots of people shooting each other down what we did get were lots of errors in the data um and by errors i mean things like inefficiencies and inconsistencies in the data so we saw that the building was consuming energy at the weekend no one had ever seen that before but they there was gas being consumed so i say for no reason this building wasn't used at the weekend despite people telling stories about kind of legionnaires and things like that there was no reason for that to be on the weekend so they just switched off at the weekend and that was saving them 10 grand a month and it's kind of these things were kind of obvious since you looked at the data but that was the easy cases and with these projects you always get a few low hanging fruit at the start what we kind of moved on to uh what we ended up doing as well as them saving a bit of money we ended up with one very angry person or one very angry set of people which were the facilities managers are these people their remit is to look round and find inefficiencies and solve them in reality their job was to make everyone happy and try and tell everyone the right things they needed to know to keep them happy without ever causing big ripples or making really bad things happen they don't want to get people pissed off at other people that was basically their job was firefighting so all they did was sit in the middle and they had a very different language for talking to the managers to talking to the everyday members of staff and they felt like they'd created this wonderful equilibrium where no one was angry the building was going just enough for it to keep going and that was it and we'd stepped in and we'd shown people the data these people were theoretically building decisions on who were causing real problems for them it's kind of my takeaway for this one is when you get into big systems like a workplace you've got to be really careful about the ripples you make for this kind of thing we thought we're doing people a favour by showing this we thought the facilities managers would love it because hey this was their data and they could show people what data they were making these decisions on it turns out they weren't making decisions based on the data they were making on it who shouted at them the most and who had the biggest clout in the workplace so i think we're going to end there a few more slides but i feel like i've gone on for a while now this project continued and we i guess took a we basically end up identifying that man in the middle that person in the middle and them being the first people we talked to in the project so we went to a building we looked for that middle person and started off with them and the study became more about what they did and trying to build systems around them so i think that's where i'm going to end