PolySense: Reverse Engineering Flex Sensors, and Destroying Your Kitchen With Chemistry for Electrical Functionalization of Everyday Objects
3:08PM Aug 2, 2020
Measure when seniors are fixed
pressure at the intersection. And we played to a virtual reality game. So you can imagine something of motion capture
And as for all of our projects. This one is open source, and you have repository again, in most of the slides. So far, another conference called augmented humans, we presented this project, which is using all the shells. They have it here
normally comes in, in long bend,
and you stick it on your skin when you have
when you damage your joint BGP. So for example, you should slip it
to maintain. It has this interesting property that it's stretchable in one direction,
but not in the other.
normally comes with a beautiful white paper on the back, but this one we we died a we functioning nicely with our organization. And at the end, the paper is kind of dark, but it's still in that. So, as you imagine if you stick it on the screen, he has some kind of glue on one side, and it's just some classic fixate on the other side. So in this case we this. What we're seeing on the screen. we're not using the stretchability, we show another one nigga. What is interesting here is that we use the finger, as analog input. I don't know if you see your screen but there is a ring, made of copper connected to a cable that goes to an Arduino it goes to the analog input, and then under two ends of the kinesiotape stick on the hand, left hand of our test subject. We have this thing that is basically behaving a bit like a linear slider it's a it's a potentiometer. And so the hand comes in by getting close to the red, you would get something that is close to the voltage input and the red crocodile clip. So this gives five volts. And then on the other side it's ground. So when you are close to the red you have five and when you're close to the black you have something like zero or something that you type into. And we can see the visualization that shows how to estimate the position, according to this footage. super simple processing sketch. Again it's available
here basically how to make a human potentiometer.
We took some other applications.
And those smaller embedded applications that we just saw, one might wonder how scalable is this process. So this insulation uses several meters of this textile. The light is not amazing but actually you can see. So, this is a polyester textile that's somehow repeating. We functionalized it after doing a kind of shibori technique. Yeah. And then after functionalizing for analyzing it was not so red anymore, originally was read, and then it became became very dark where we functionality. And then we printed it. I don't know if it's visible but it has this interesting kind of squishy folded structure that will create an amazing passive active feedback. This application is basically functionalizing feathers in. So in the previous example we were doing visual
artistic visualizations shaders. And
in this case its own sophistication so when you touch
a feather. It's behaving like capacitive sensor. So a bit like sensors in our screens off our phones, the things that you touch it, and you can use it with an Arduino to trigger sounds. That's basically what happened. This example shows that you can functionalize, other things that excite. We. Yeah, we try to, and most of the things that are posed are cybers. They should work. Except when they are very synthetic like a
piece of very stick.
Synthetic textile usually doesn't work with. Anyway, you might recognize Hannah Winston, so just after it became last year we went to this conference in Germany consistency is the biggest hackerspace conference in the world at 18,000 people in the winter and in the summer it's, it's every four years it's smaller, and it's, I think was 5000 people or something. It's a big thing it's really worth it if you have the opportunity to actually if you have to get the ticket but it's worth taking. Anyway, and I was showing her slides, by folding her knees up now. So we fictionalized this tent and, and with some connections. Not enough. In this case, kind of a little bit No, It's, it's, it's another kind of gone above, you can send signals in and use them for sure changing your slides on visualization or whatever. It was really one, and we realized, kind of, with a lot of stress. I think for the links in the basically the humidity suite is changing the values that should be. So you have to think about it when you make your own tents or if you make it wherever we didn't know at the time we discovered it's a great community center. Could we discover that thing free. And on the other frontier applications, we realize that you can also one, or you can hit your user by putting electricity on the, on the tech side. So in this case, it's basically as anything resistive if you put electricity we cheat because it doesn't conduct very well. And so it kicks in when you start hitting, you can do that very fast. The same connection can be used to measure the resistance, after you hit it. It happens that the resistance changes with the temperature. So you can actually use it as a temperature sensor to. You don't even have to eat it but it's interesting because if you heat your user, you can avoid burning him or her. As you can see here we have some smoke, and it was pretty fast if you put it off current. So if you want to make something from burning mind and make it burn perfect it works really well. We had a few guns on flames. You can see we have this infrared visualization that was fairly hot. And we characterize the temperature versus resistance. So we can create these kind of crunchy add to loop. Lows regularly with like your thermostat at home with the laws regulating the energy that you've sent to your site to your heater, to keep it at a regular temperature.
So we don't want your user to
be put on fire.
That was white originally zipper. So if you measure the resistance between those two sides. You have this kind of constant resistor. That is for the value something that is shown Okay, 600 k something like that. And this is metal. So when you close it, you shorten the path of the, of the electricity. And you get the skin off your slider, or potentiometer. So you can you can use it as a fusion this new kind of music application, you can use it as a potential method that you can that can keep its position. Whereas, the normal pressure sensor are more like for. It's a Huawei shake off. This one would be more likely to change the volume or change a control signal that would be wrong. But yeah, being stable. So this is an off the shelf.
Zebra that just was white.
Made out of
cotton. And they all work. One trick if you do your own is that the second step of the polymerization is oxidizing the metal so you need to keep after the process with flux and media.
If you have a heat gun
to dissipate the heat.
So, we saw a few pretty variation that we deployed in various festivals and conferences. Now you might want to know how to make your own. We have this certification that describes all the steps. But let's do it step by step together. So in our case we use the washing machine to make a lot, but as I was saying earlier, you can do small mixing things by hand. And I did a lot of this by hand she was not really a fan of the machine. She's kind of a wire.
So, yeah. First, you put the water.
And you should use a bit less than twice the amount of fabric that you're going to need, knowing that the thing will move around and it's nice if, if, if you have a bit, a bit more. But yeah, it isn't like something between 1.5 and two times. Just so the whole thing and just the water access. It's important. Then you put the monomer. So in our case it's pure oil, including to buy the things you want. So, this is a bit toxic before it reacts with the second chemical product is hyper super lutein it's a bit similar to. If you're using like a strong alcohol for like, I don't know the cleaning stuff or you don't want to put it in your eyes don't drink it within your mouth. It's not nice. It stinks. So in our case we use gloves and goggles and even masks to get in shape but I do that, fairly often in the kitchen or in my address space time. Once you have this mix, then you have the dictate you soak it for like 10 minutes. And we could call it again on Monday, Mr. Mundo mirror ization, which is basically, bringing all those Mundo mer disease. Can we get radicals everywhere in the favor. And then you add the oxidizing agent, which we create here the printer ization. So hints are incorrect. In our case, it comes as, you know, powder or crystals, you break them and you deserve them.
And then you meet
again this is kind of toxic you're alone, and
it's meta is not super nice on your skin. At the end if you put it on your screen which I didn't mean that you feel like super dry hands. So that's about it but it's not. So, that we trigger the, what is called the Institute for memorization. I'll explain that a little bit. But concretely, if you want to make kids resistive sensors, so the one that measure the pressure, about half an hour works most of the time, before that 45 minutes for simplistic materials. You want to make the plastic sensors, as the big one that I was showing earlier. That was like seven meters. Two hours is really the minimum for capacitive sensors. In our case, we were making several meters. Several yards of these textile for vegetables, maybe six by 10 or something like that. And we folded in. So it's not exactly a lot of time. And we need several machines of this seven, seven nights. Yeah. Every time when they go free meditation, but basically the longer you put in their eyes, the better the conductance will be. Yeah, chains of monomers, so the monomers which we see on the left is just one element. In our case, and the putting there is a long chain here we just so far them but it's really long, the chains of monomers that are built around each fibers, almost like if they were merged with the fiber. So it's more than the coating. I like to call the original every day. You can imagine that the fiber itself becomes conductive, and this is called institutionalization because the mother Mary is already everywhere around the chamber, already soaked in the fiber somehow. And we trigger the pre memorization if we couldn't act with all of its friends and create these conductive particles that are very long. And this is the trick. We tried so many approaches and we couldn't imagine that first, there was this thing purely that existed we didn't know the way of making it in two steps like that is the thinking. So, if there's something that you have to remember in this process is that it's Institute, and it's this two step process that creates this conductive change. So this polymerization is mostly creates chains of carbons, at the end, and actually not toxic, we did some tests with the methane club, and the technical way of doing is that you take some skin cells that are still alive and you mix them with the result of the polymerization, and you are still alive 24 hours later, it's considered as non toxic. And this is fine. This process for another application which is basically using these functionalized materials as what they call mixed aid batteries. So what they do is they use another layer of another chemical and they put it underneath the armpit, and they use the sweat to generate the chemical reaction, which creates energy. So it's an executive battery. I would say it's more harvesting energy but in a way it's also the battery because it's maybe itching a little bit, some part of the methods. It's not an infinity generator. So we create this change of carbon, which explains why the result is so dark. But it also explains the electrical conductance because the carbon chains are fully connected. So I think this might knock out a few people here, and I will probably skip most of it, but what do we do as a scientific team, we had to characterize it. So we did different tests, basically measuring three pressures that we can see here over time. And so we measure the pressure with a Newton meter. And we measure the resistance. And so we need three pressures that we can see the three peaks here, and we will look at the memory effect, because some pressure sensor would stop working over time.
And so some of the materials would not work as well over time. And it's normally, we have to find the right materials for example she has this natural way of bouncing, but other materials don't have that the stretch sensor is restricts Dixon's lycra for example is quite nice. But some like t shirts, do have this very good behavior over time. So we characterize these. And we found that we actually have a better range of resistance in different cases so the woman felt the paper stingy has a nice. So it's not the nicest behavior. This one is just as a switch of fabric. This one is interesting because it has a high resistance which makes sense or uses less, less electricity to to work. So there is a question does over time to account, like the accumulation of body oils in washing. That's a very good point. We talked about that a bit later, but. Indeed, we have to wash the materials so they are not soft before. Otherwise if there is any sharp oil on the textile. The preliminary ization won't work. In normal scientific lab they will use ultraviolet methods before so I've gone on so. So if you want to improve your material you can do with
pure water that it was distinct
in the process that I just mentioned like UV and alcohol
So the degradation over time can be stabilized. So, if you wash it a few times I think it was something between five and 10 times. There's a moment when, so it will decrease the, the behavior of
overtime after a certain amount
of washing it will stabilize. Now only, I haven't tried but we trained attacking the thick side with UV like sun exposure. And again, this is, like, kind of preparation so you have to, to attack it with UV and then it stabilizes, and then you can release it to your customers if you want to say.
Also the museum exhibit. A lot of.
That's true. That's true. So, I guess, Shuman oils didn't really affect it so much. Free schrenk t shirts, I'm not sure what you mean, are like examples of material that we could use Yes, yes, pre wash jeans. So jeans would not be very precious entity or stretch sensitive.
But I guess you should take it from diagnostic.
I won't go too much into details of these characterizations. If you're interested, ask me more questions. Let's talk about design methods, because I think it's the most interesting part for craft making. Let's question. The same process, having to treat the material, so that the fabric is stable between washes. Yes, that's the idea. I guess when you buy a textile that is already died like probably this, it was pre processed before it was really, it makes it so this is him idea.
we made this interesting way of giving a functionality and electrical functionality to fabric in 30 reading, but we just became mostly what exists, made by Nuneaton and other people also by NGO staff sometimes, which I don't think he's as good as the as Unix, which is really the best in the industry. Our material. But
you could say, all right, it's a bit like copper
blank PCB materials, copper is great for electronics but it's better when you get the structure, like, you can create circuits, or in our case we can even create sensitive paths in the city. So we look at the possibilities. And we're inspired by traditional techniques such as shibori here, some kind of a tie dye and traditional taking from Japan. It's interesting because Lowe's very simple ways of hiding the textile from the premier. And you can see the white part here is not functionalized, and the black part has this superpower of conductive
conducting electricity. We can also make ingredients as you can see, it's kind of nice.
So we got there. Again, I won't go into details on that.
But yeah, you can see the gradient at the top right.
Another interesting traditional technique could be back three or so from, from Indonesia was another one from India. The one is interesting because you can functionalize just fibers. And there are some cities. Some people, is when you could predict in varying who found a very smart and technique to use this kind of mix of fabric that is functionalized or residual ridges, dark or white. And, and you can, if you have a machine that weaves sorry the skin stick style.
You can have for example, a white car, and a black
thread that rotates around it. That is turned around, into densify. The black thread, you can have dark areas. And you shouldn't be in sci fi you can have mostly white area. And if you do that in a smart way.
You can anticipate
where you should have been areas always shouldn't have an image in this picture here. You can see as a thread in the rest they're all there. So these things actually, and you can say I want this zone to be dark, these need to be dark and so on and you can create this triangle. So they created the software they lose, making your own thread and controlling the weaving machine. So that you can create patterns, and you could create functionalized patterns. So we tried different fibers like lean in sync and few others that work really well. But it's not super easy to friction I said that you need some checks to some techniques to make it without the dreadlocks, because then your favorite is Mr. or Miss. But, yeah, we worked with those guys project Hilo is the full name. If you are interested.
They have is to do Hello, sorry, so you would say.
Another interesting technique from Indonesia quick bandage basically uses wax to mask the polymerization. And this wax can be removed after one, so you can do several steps of a functionalization. So we did some simple tests with this you know shows exactly, and looked at how the conductance works between two points and how it's isolated between two areas that should not be connected, and we can see that we have mega ohms. We really didn't do that very quickly but when you have mega ohms, it's pretty much isolated. Whereas, if you have two zones that are in the same area like into. They are kind of conductive so you can use that as an essential
is, what is different between them.
So this is not exciting. It's a plastic that happens to have this high conductance I think it's originally made from.
Because that institution.
Isolation so it's basically conducting electricity but not very well.
care. We just
uniques is really nice just to finish,
which is really nice because it has this. They have two main
kinds of accent one is stretchable, which is also pressure sensitive but not as nice. The other is woven. So the first is kind of like our lycra, and the other is moving. Sorry non woven, it's actually. And it has this nice z pressure and behavior, kind of bouncing and have nice range of resistance, a bit of range, then we'll start. We'll start as it's not exciting, but you should worry is really wearing plastic you don't really want that. Second question.
Yes. So, somebody says, I really don't like
their idea ization we traded
share some more details if interested.
So you so you always have to try it a few samples at the beginning and then you see what works, and you see how each method, you'll need usually something between 30 minutes and 560 minutes. If it doesn't work after 60 minutes and it would work. As you can see visually if you start from a white material it should become almost black or dark gray.
One more question. I totally understand.
So, we actually haven't. This is a work in progress we haven't been to the studio, you lose. And basically making patterns of woven textiles, that have either functional or non functional fabric in for example here you can imagine this triangle could be a pressure center or capacitive sensor, but you can make any kind of
shape. So we saw with
betting that you can make fairly random kinds of shapes. But if you want to make it in a scalable way. You could also have that kind of approach, where you even prepare your paper in advance, and then you we with Dell machines that do it really fast and really well. So I think I was here. This is a little bit outside of the, of the functionalisation that we're talking about, but this is still a country. If you take the classic copper textile that you can buy, which is usually made from isolating like electromagnetic waves, you can find that on less EMF Comm. I think the TMS people who want to isolate electromagnetic frequencies. And so this takes time is pretty much a quarter. On top of polyester. and you can fix it the same way as you would a normal PCB. So in our case we used. And we found that if you do that, you can create secrets. And once you have your electrical circuit, then you can add the
I see that I speak a lot. I'm going to go fast. In this case it's a quick. It's a simple stretch thing sauce if you stretch this resistor when we change any trace. That's key, the other one. Should we need a precious enzyme which is somehow taking advantage of the, of the fact that you can further your circuit that is being of hydrated. So it has this conductive part that we come in contact with this season. And then you shuffling again is another conductive path will be on the other side so you have the sandwich. That is entirely made of textile that is funded with conductive and the resistance the resistance can make a precious metal. So we looked at the Microsoft results but I will macro results, presents, but always keep that. Just to say that some of us really functionalize Well, and simply don't
talk about that.
some of the philosophical take away from this project is. I didn't have the background to understand how to do this kind of functionalisation, and I've been just asking many people, people in hackerspaces people in academic conferences and basically This one wasn't impossible if I didn't ask people are in the conference who told me hey yeah I know someone who does matter and saves they can help you. This is a quote from irini. Another friend from the bush from the, from the Excel community who say in a tip for, if you can. No, sorry. This one is
from me, she can open it, you don't own.
Again this is abundance open source movement, we go.
And that's the quote from uranium
tools and materials are never neutral, and this is surely of use when you look at those machines, let's say, for people who make music like a Pro Tools setup cost, lots of money.
Whereas open source tools can do the same.
If you look at scientific tools like,
oh, embroidery machine right now, you should look at on Twitter this guy could go on living living, he's not he's a guy from the processing community and open frameworks. He's been making an embroidery software that allows you to make custom shapes from processing. The number of tools costs thousands to make custom patterns, and he's basically making it open source and accessible. And so if you make tools that close sources, not everyone can use it and so this is not neutral in that. So, I'm so sorry I speak too much. Now, is going to be showing a little
application that we did together. And he was to share your screen, even if
we don't have a lot of time, three minutes.
So, really quickly what we did with.
Today Show. I think people are going to be for that as well so basically what we did was we attached the strip center to in Arduino. And then when I moved my hand up and down the front a nice visualization will move accordingly. Center and the information pipe and how that works is we take the analog signals from the textile. We read it through an Arduino that I'm actually holding in my hand in the gift but you can attach it in another form, if you prefer. And those signals are connected via a serial port to a f running, no GS on my computer, which then converts those into something that can be read via WebSocket into p five Gs. And that will control this and you can see that if moving over here. What you will see with the whole system setup is this Perlin noise judaization actually moving like waves, and you can check it out at this thing. I'll send it in the chat later
can also show tomorrow.
I think you actually had it in your choice. An alternative to this that bypasses this p five zero Ctrl F would be just use something called whip with me and my friends created a tutorial that you can check out later if you're interested to basically skip from using the Arduino to getting the signals read in p five.
It's quite nice. You can go faster in, but it has this problem that not every kind of works, you have to use an Arduino to know these kind of all those zeros, zero, allow you to configure the processor, as you can. Those that have a real hard, use the hardware, physically, so not the Arduino Uno, for example,
I have yeah the GIF is here.
Yeah, this is the visualization. So basically, the stretch changes the angle of the, the orientation of the Perlin noise. And, yeah, just send the link. So, the link will be around the sensor connected so it's, it's not moving. But you can use a, an Arduino and BFP five to play with. It's known. I'm hoping to do alone and. Did your question. If you haven't.
This is Polly sense reverse engineering flex sensors and destroying your kitchen with chemistry for electro electrical functionalisation of everyday objects, Cedric on a thank you very much for joining us today. You know, we will have time for just a couple of audience questions maybe one or two. When one member of the audience asked about the washability of this, of this material.
So, I somehow replayed on the radio
channel. I think the, the way I mean the simplest answer is that when you make your own material, you have to wash it at the end because there's still a lot of chemicals, which are not too toxic. Again, but you still need to watch them, otherwise you get some kind of gray powder
on your hands
and I usually wash them like five to 10 times. Something like that after, after functionalizing the material, and then it gets stable. So, it's fairly reasonable after that. Obviously if you watch it 1000 times super hard and time, every time but it's reasonably stable.
Thank you. We have a question from the audience asking, Can you obtain the heating effect at a lower voltage, I'm thinking of the concept of plugging in a USB power bank on a cold day to heat clothing.
So that's a
very useful feature I did. I would say that if you have a power bank that can provide reasonably good amount of current which is the case engineering, and depending on the way you organize your. You're pretty memorized circuit. It can have, for example, a long path that has a lot of zigzags, or it can be just a one big pad, depending on the structure you can adapt the resistance of the system and allows it to work with the lower voltage but it will consume more current, it has to be somehow proportional right so it's possible, and it would meet some tests, I haven't
done, and we'll have to leave it there because we're just about out of time but very quickly Cedric, where can people find more information about your work and get in contact with you.
I can send this link somewhere everywhere, it's the simplest is counter chemists on GitHub. So if you look counter can use the GitHub that IO. There's all the documentation. If you look up also petitions on hackaday. And we have a good documentation there. I think those are a good starting point.
Excellent. Cedric on a thank you very much for joining us today. Thanks, Linda. Just one more quick announcement for attendees watching, you're invited to take the stage yourself and 1500 EDT this this afternoon. You have up to five minutes and up to five slides to share knowledge of interest in our lightning talk session. Go to the help announcements channel on matrix just before the start of the session to get the meeting link and give your talk. In our lightning talks no pre registration is needed. Cedric thank you very much indeed once again for joining us today at hope. Did you throw it right back to ground control.
Hello World Ronnie wrote from Hawaii, I just wanted to send a great big aloha to everyone out there
well wishes of healthiness
I'm a cybersecurity Project Manager for a government contractor company here in Hawaii. I'm also a Polynesian and choir dancer, so I live a well balanced life in technical and personable work, hope you're enjoying hope 2020
is a little sample of my fire arts. Enjoy.
Hello, all hope lovers out there.
I'm Rafa Ninja, and this is my buddy Jason. We're based in Sweden, and we wanted to send out some thoughts to our fellow hackers out there.
Make sure to take care of yourself,
sleep properly, eat properly. Drink a lot of water, exercise, and take care of yourself.
This is hooked 2020 and 1500 hours EDT attendees are invited to the hub stage, you have up to five minutes and up to five slides to share your knowledge of interest to other attendees.
Go to the help announcements channel and matrix just before the start of the session to get the meeting link and give your talk, no pre registration is needed,
school districts throughout the United States suffer from notoriously poor information security, even the school district spending on tech is at an all time high.
Our next speaker finds that understaffed and underfunded districts are either ignorant of the risks, or simply choose to accept them, and there is no penalty for either.
We present Dr Travis Paki with the privacy of 100 plus million children, families and young adults is unprotected.
In particular, k 12 education. During the day I'm a leader in a K 12 School District. I have been an avid cyber security researcher and advocate since the age of 12, and I teach cybersecurity at the Masters level, my study is called an exploration of the strategies Information Assurance technologists need to improve information security practices in a school district. This mouthful was my attempt to bring light to what I feel to be a growing risk, either ignorant of or abdication of responsibility for information, infosec, and privacy protection that affects 10s of millions of individuals, not only by school officials but also by law makers and auditors. For the purpose of this study, a strategy was defined as a method or plan to bring about a desired future, which is very general but it could be culture laws influence of the CFO or CEO, cio budget audit pressure cloud posture and outsourcing preferences. I decided to focus on the idea of trying to find out what is missing in K 12 education infosec. Other studies and media had established a clear gap in security posture, which is unusual in such a large industry so large that it was 700 and 6 billion in 2015. Yet from my experience the industry has informed dedicated and committed individuals who are relatively effective at wearing many hats. So why is infosec in the field, generally an underfunded afterthought. To be honest, I approached this study with a lot of bias. I thought the answer would overwhelmingly be low infosec budgets, and in some cases contradicting previous studies. It turned out to be much more than that. Like all things in the real world, the answer, and the remedy is complex nuanced and will require investment commitment and yours to correct. Before we get started, some terminology ADM is average daily membership for this talk, we're going to use it as a measure of District size in terms of number of students enrolled. It is essentially the aggregate average of students in attendance over a given period of time, a school district. This is an organization that manages the operation of a collection of schools. This is usually governed by an elected board, who then appoints a superintendent to manage the organization, all the other positions are usually appointed by the superintendent and their delegates. Key 12 or now, often referred to as pre k 12. This is the range of grades that the district serves typically between the ages of three and 18, or the grades of pre kindergarten through 12th, and there's obvious exceptional programs that go beyond that. Education service district. This is an organization that operates shared services on the behalf of many districts, offering services such as payroll Benefits Administration it nurses and other services that may be too expensive for the member districts to operate independently. Also, the terms information security and privacy are used together a lot in this talk. For the record, you can have security but not privacy, but you cannot have privacy and no security. I use the terms interchangeably, but to indicate that both are necessary. The problem statement. I work for a district that has a little less than 50,000 ATM, it's fairly big as far as districts go with most measures of a large district being 15,000 students or more. At a line of about 15,000 ATM or less, organizations usually have an IT director. This is a kind of jack of all trades with maybe one or two technicians. The staff are often scrambling to keep up with what is asked of them in terms of end user support information security and data privacy are just one of the many hats they were putting that into context, my district, at 50,000 students has an operating budget of 650 $5 million in almost 10,000 employees counting part time staff that translates into an IT staff of roughly 65, and an information security staff of one presently open since last November. As a general rule, the top 150 or so districts are large enough to have dedicated security staff. After that the next 350 may have staff that are differentiated enroll and have someone who has multiple duties, one of which is information security. After that, the United States has another 13,000 districts that are small enough that they may have no one responsible for information protection. If someone is, it is doubtful that they have the resources to perform the task. So how did I get to the hundred million number at the top of the talk. Well, there are 50 point 8 million students in the United States public education system of those approximately 3.3 million graduate every year, those graduates have at least one contact double doubling that number to 6.6 million, take that time the seven graduating year since the rise of the student information system. And you have 46 point 2 million added to the 50.