Introducing *DAS: A Framework for Certifying Hacker Knowledge

10:53PM Jul 30, 2020

Speakers:

Keywords:

certificate

learners

prerequisites

learning

people

system

reviewers

trust

teaching

question

grades

knowledge

award

accreditation

makerspace

review

certified

certification

startups

case

Welcome back to hope 2020 really glad that you could be here with us. We have a great session coming up right now with our next speaker who's Dana Retton was a maker, programmer, artist and informal learning enthusiast. They integrins talk is about introducing stard das, which is a framework for certifying hacker knowledge. And so please remember that we're going to have the host of live q&a in a matrix chat, please go ahead and chat within there while his presentation is on the live stream, and then post your questions there and we'll give it to him at the q&a after the live stream. So coming back to you with hope 2020. Here's Dana Gretton.

Oh man. Oh man. So excited. It's time to press enter to help. Hello HOPE. Welcome to certifying hacker knowledge with startups. I'm Dana. I'm an informal learning enthusiast. I'm a musician, a singer, bio robotics engineer and a DNA screening engineer. I'm also a known explorer of non conventional uses for drill presses, and a computer bender. So here at hope 2020 I think it's safe to say that we are all here together growing hacker community and sharing hacker knowledge.

Here's some hacker knowledge being shared right now.

This is a makerspace in Beijing, where I was lucky enough to be a professional in residence in January of 2019. People here are cooking they're sharing knowledge. Stories skills. Here's a group of everybody together, eating in the makerspace, which is whatever. But anyway, this makerspace obviously produces some quality learning. We have people coding we have people working with Arduinos. We have people making little robotic projects, sewing, assembling all kinds of amazing skills. So this is a learner run makerspace. It's unique in that the makers themselves make decisions about how it should be run. In this space, I saw people learn lessons about governance and the meaning of leadership, how to run effective meetings, even how to invent their own organizational schemes.

This is stuff you don't get in school.

So the question is, can makerspace learning be certified?

For one thing, it's completely impossible to grade.

It's also difficult to capture on rubrics And it's intensive to collect and organize the evidence, you'd need to prove that you learn something in a makerspace. So this is traditional certification. It's a story everyone's familiar with grades, go to transcripts go to certificates. That certificate could be a high school graduation certificate, or a degree. So this whole process is based largely on grades. It celebrates rule following and good behavior. But as we all know, it deeply suffers from bias. As you can see, if you look at the statistics for how many people of color graduated in different PhD programs last year, it can't possibly serve everyone. It's for certifying everyone in knowledge that's standardized across our nation, but it can't possibly be relevant to everyone and the specific knowledge that they need to succeed in their communities. It's not accessible at higher levels. In fact, it benefits from exclusivity at higher levels that actually increase its economic value. It only tells one story, it completely lacks intellectual diversity. Now, people have been saying things like learners need more autonomy, or non school learning is the most important. Both of these things are domains where hacker knowledge excels, where all learning is self directed. And obviously, it all happens outside of school. But these statements are coming from years like 1903 1899. The non school learning in most important can be traced to john Piaget and his formulation of non school learning as a developmental theory. And also the free school movement of the 1970s which popularized many of these ideas. So why should we certify hacker knowledge? Maybe it can't or shouldn't be certified. So I think that it should be certified. Because certification limits what students can learn it limits what they can afford to learn. In fact, it limits But anyone can afford to learn. And the reason is that credited learning must always take priority over non credited learning. I would say to someone in the maker space, do you want to work on building a cool project? And they'd say, Well, I have to do my homework. It's just an insurmountable obstacle. And something has to change in order to allow people to take on hacker knowledge on an independent and universal basis.

So the issue here is practicality largely.

This is flow grades to transcript a certificate has its flaws, but it's very efficient. Institutions responsibility that they take on to teach learners means that they must enforce some kind of standard of achievement. and that in turn means that they need to have assessment. Reviewers can delegate the valuation of millions of students to trusted high schools and standardized tests by reviewers. I mean people like employers and college admissions Reading a certificate in order to deal with this many learners must be absolutely instant. And that is a benefit that we see with high school transcripts and degrees. So of course, we've seen the development of some very powerful new technologies in the last few years things like the extremely fluid sharing of media, between people, ease of capture and distribution. And also, this computational power. Of course, computational power is always much, much greater at any particular moment than it ever was before. Thanks, exponentials. But we don't see either of these forces utilized to make our accreditation systems and therefore what we can afford to teach and learn better or more advanced. So how can we turn these potentials into new trustworthy certifications? So the purpose is to introduce a new learning certification concept and this is certificates are made up and created by learners to certify themselves and their own knowledge. And references in between the related certificates can be interpreted by open source software to give insight. Many of these certificates are connected together. And we're calling the system is start off for reference rich, decentralized accreditation system certification by interrelated certificates and computational tools.

So here's the start off certificate.

This is a view of

its content.

You'll notice here at the top of the certificate is this kind of cryptic code which is unique to start off certificates. It's called its tag, and you can see zero Qd on the left is mirrored as the first three characters of the tag on the right and we'll use that abbreviation throughout.

So let's just explore this certificate.

This certificate is called drying the walk. It was made by a learner named Bella. And she observes that it's very important to dry out a walk before cooking in it with oil, because as she says it will explode in your face and hurt a lot. I'll start off certificate has a description and a title. And then it has a case that shows that the learning happened in a way that is unequivocably tied to that particular learner. In this case, she accomplished that very succinctly by creating a video of her drawing a walk

and also demonstrating what can happen if things go wrong.

So these are the basic parts of a start up certificate, title, some metadata about who created it and when a description of what's covered in the certificate and a case for the learning covered. So now we can ask ourselves how much wood Trust the certificate, and the learning that it certifies. And just to note, this is experimental software. And of course, the user interface could use some work. But this is basically how rating or trusting a certificate works. So this is the zero key D certificate.

And the slider at the top here represents a trust input.

And this represents a trust readout. So you can see as you move the trust input, the trust readout moves as well. For the creation of these certificates, I mean, they're obviously much richer than you could ever do on a graded assignment, A, B, C, or D. You can see exactly what the learning was. And very much like a portfolio, you can get extreme detail on exactly what a learner knows. And it's enabled by a modern fluid media sharing that enables really fast creation of certificate content. You can sync this media between devices that are used to capture it and that are used to compose the certificates and you can even collect evidence on behalf of other learners and share it with them. Immediately. So here's another certificate. This one's abbreviation is XP one. You can see it's mirrored here on the top. So I'm going to open this one. But actually, this is just a plain old Word document. So I'm going to open it here on my computer. I want to emphasize that this is really all there is to a start off certificate. It's legitimately just a doc x Word file. And it has the same parts as before this title description case case for the learning. So this certificate was made by me. I don't actually happen to know how to cook with the walk. And so this is not a very trustworthy certificate, and I hope that anyone viewing it would be able to see that I don't even appear in any of the images. So this is not a trustworthy certificate, it's important to be able to tell the difference. So these two certificates should be rated independently, XP one has its own rating, zero, Qd has its own rating. Start off certificates in summary are self contained proofs of specific learning made by learners for themselves. And in this way, they're very much like portfolio items. So the challenge comes when you face this, what do you do as a reviewer, when you receive 50, or 100 certificates or even 1000 from all the different corners of someone's learning experience, and then from many different learners? Well, this problem is solved all the time, but in an unfortunate way. So this is the typical traditional approach. We split subjects up into kind of tracks or groups to simplify the review process. So modern western style education splits different learning topics up into familiar subjects like English, math, social studies and science. And here on this topic, I've organized all of these symbolically from beginner on the left to advanced on the right. So you might learn to solve equations before you do any calculus. And then at the end of all this learning, you receive one document, which is your transcript as you exit High School. So this is a plus D, C, and a plus on these four subjects. And what a reviewer understands from the certificate is that you've got an A plus and everything across the board and English, D, and everything in math, and so on. And obviously, this misses an incredible amount of detail. It's not that reviewers don't know it. Of course, they understand that there was some variation in what you've mastered and what you didn't. But all that detail is lost and this is terribly reductive. Grading has an implicit prerequisite structure. Reviewing just one grade thing on the right can tell you a lot about all the certificates on the left, but it is extremely reductive start certificates have a similar mechanic, they have an explicit prerequisite structure. So reviewing one certificate can still tell you a lot with the help of some math. So here are the same two certificates. One of them is a Word document. The other one is an online Dropbox paper document, both have the same format. And they're linked together were the relationship. And this is an arrow and it means a prerequisite. So lifting the walk here is a prerequisite to driving the walk. Now you can see, as I manipulate the trust input of one of these certificates, the trust readout of both of the certificates changes. So now the trust readouts of these two certificates are linked. And as you can see, the more you trust zero Qd, the more you should trust it's prerequisite XP one. So let's have a look at the certificate again. I'm going to open up drawing the walk on Dropbox paper once more.

So as you can see down below, there's this section called prerequisites.

And if I click on that,

here's an online

version of lifting the walk, which is the same certificate.

The real power of this system comes when multiple prerequisites are chained together. The left for certificates here represent just general computing knowledge, things that anybody who is going to do anything related to programming, or code development, always need to know. On the right here, I have a certificate that represents sharing code. This is actually one that I made. So these arrows again, indicate prerequisites. So everything on the left has to come before things on the right. I want to show you this certificate and what it looks like online. This is again, just a plain old normal Dropbox paper document. So this describes how you can share code using get on GitHub. I've been doing this for years. So it was relatively straightforward for me to make this certificate and actually the This certificate isn't just me speaking about it, it's actually me showing someone else how to use get that person's name is Camilla. And we'll come back to her later. But here at the bottom, you can see that I've listed the prerequisite, the topic is getting around a terminal, that's something you need to be able to do first. And the certificate is why changing directories and Unix terminal. And as you can see, why why is in fact the prerequisite to 443. So now, if I say that I trust 443, because this person seems to be experienced with it, and put it to 100%, that says that everything below it all of its prerequisites must be satisfied, and all of their prerequisites must be satisfied. So we say that, in order to share code, I absolutely must have general computing knowledge. And in general, if I trust a certificate, I trust its prerequisites. So I don't even need to look at the ones on the left necessarily, if I trust the relationship that they represent So when prerequisites are present, a reviewers job facing all of these certificates is significantly easier. Okay, so what do you do if you face hundreds of people? What about thousands of people? Well, so our traditional approach is that we'll sort people by grades, so we don't have to consider so many people.

So of course, this is very exclusive.

Grading just assumes that all the students are independent. And so you have to kind of toss out information about most of them, and just Grade A few that you can afford to give the attention to. But of course, no one actually learns alone. Start off explicitly recognizes knowledge sharing that happened between learners while they were learning. Reviewing one individual can boost the credibility of the entire community, again with the help of some math, so let me show you how that works. So On the top and bottom here are two certificates. The top one is ZSU. And its name in Chinese means making curry dice potatoes. And the bottom certificate is making dice curry potatoes, this one in English. So these certificates are actually composed by two different learners. The one on the top is an experienced and expert cook. And the one on the bottom is just learning.

In this case,

we've granted an award from the expert cook to learn. And we'll represent that with an arrow with this magenta color. So an arrow means an award.

And again, you can

see that as I manipulate the trust input of ZSU it changes how to eat is perceived. Again, the trust readouts are links when there's a relationship between the two certificates. If I taught someone to cook potatoes I simply am must know how to cook potatoes, there's no way out of it. Similarly, if I trust a certificate I trust it's a Warder. So the trust goes back up from the learner to the teacher. So here's a weak certificate. And I say week just because it only has a good case, if you were to review this certificate, you'd say it's strong, but you don't have any evidence outside of that. Here's a stronger certificate, because you were taught by a master, someone else who has a very strong certificate in the subject.

But the strongest kind of cert

is one where you teach others and it's obvious if you can teach other people that you must be an expert in the subject. Here's another example of an award. This is returning back to my good example. And at the bottom, you can see this section called a word and I'll go ahead and open this one up. So this is a certificate that was made by Camilla, the person who I was teaching how to use get in the videos. She describes what it does. And she provides a case that again isn't about her just describing it. It's actually her teaching it to a friend named Megan. So I can give you a sample of the recording that she created as she was teaching Megan

is basically a way of like sharing code that

can work on the same project. At the same time. She gives the basic principles of get.

So if he wanted to look at it and see the changes that I had made,

then you would have to, like do this thing called poll. He like, pulls my code and then he added, you go

and she explains that in order to view other people's code, you have to pull it first.

So because this

provides such solid evidence that Camilla understands how to do this, I was comfortable giving it the award that I gave it. So this actually helps in two different ways. Camilla benefits from my experience in get so that if my certificate is reviewed, it boosts her certificate. If her certificate is reviewed, mine gets reviewed very positively. So when awards are present, a reviewers job is even easier than if just prerequisites are present.

So this, you could say is the strongest possible certificate.

It builds on prior knowledge.

It's the foundation for later learning the prerequisite for later learning.

The person who made the certificate was taught by a master and they became a master and they taught others and of course, it has a good case, but after a point, I mean, would you bother even reviewing the certificate itself? densely connected to other certificates. If everybody involved thought that these were correct relationships, you almost don't even have to open it.

I don't even need to click.

When a certificate strength comes from its connections, it doesn't need to be reviewed. The reviewer can save their time you could review it, but your time might be better spent elsewhere. So start off rewards excellent peer to peer teaching, through the combination of prerequisites and awards, prerequisites happening within the same learner as they build their skills and awards happening between learners as they share knowledge. You can build up a network of certification that gets stronger and stronger and faster and faster to review.

With your requisites and awards, and the way that they interact.

In dense networks have started off certificates. Reading just one certificate can confidently assess dozens of others. like looking at just one grade on the transcript can confidently rate your skills in many different areas. Okay, but how can these certificates be trusted? How do we know just basic things like the author's real, like the prerequisites are correct, that the awards are deserved. So usually all of these functions are taken care of by our central school districts. So we know the author is real because the school districts keep track of the identities of their learners. We know the prerequisites are correct, because we trust our teachers to make correct curriculums. And we know that cheating, and receiving grades that aren't deserved is severely punished. in a broader sense, this is why we trust accreditation. today. We have at the top of the stack, the United States Department of Education, at least in the US, which is in charge of learning of hundreds of millions of people. And we have the New England Association of Schools and Colleges, for example, which is in charge of millions of people. And then if you were to look at MIT Which is accredited by any ASC. They're in charge of maybe thousands of people, and then finally, their diploma after all this accreditation is awarded to you. So there's this very clear chain of command that establishes accountability in the case of a failure.

Of course, any system can be misused sometimes.

If it doesn't meet the standards, if some kind of cheating is detected,

it's very clear that there's a simple path to discover issues.

But what I'm interested in is a decentralized model for certification. Where you're teaching and you're learning are inextricably linked to the Learning and Teaching of other people in your community. And you as a community can gain trust and credibility that accumulates over time. The same mechanism that infers trust on this complicated network can also identify missteps. So what if there's cheating? What if there's just misinformation that people don't know what they're teaching? What if there's bribery? So in the case of a failure, it's not so clear what to do. There are many possible causes for any particular link going wrong. There are so many paths that you could potentially follow to discover issues. It's just mind numbing to imagine it. And I think that's why this problem has never been solved. But there's also many possible insights about exactly who knew what and who taught what, and all the different variations in the knowledge that they have and don't have. So I want to give you a vignette about how startups maintains its credibility. When a yellow says, could you award your son dressmaking certificate to mine, and gives a link to their certificate? When urteil says, Sure, this looks amazing. But then after look at it for a few minutes, gets back and says, actually, could you add something that makes it clear that it's you who made it? Someone might think it's copied and that's because learner teal is really cute. served, that their reputation is preserved and that this award that's going to be granted is actually good and benefits everyone involved. And the learner yellow says, Okay, I'll come back. Thank you. Here's another vignette. learner Arne says, hey, my certificate, how much can I pay you for an award? winner green says, if you don't actually know the topic, that would just make my certificate worthless as soon as anyone notices and learner green doesn't just mean if someone notices that certificate. They mean if anyone notices that someone learner orange went on to teach him certified later didn't understand this topic, that it would be immediately traced back. You know, it can be traced back to you and me right.

When our orange says I don't even want it then.

This is how startups maintains accountability.

In this decentralized model, learners are motivated to carefully verify everything about the certs, including their connections, and making sure that They're formatted correctly and valid before connecting to them. And to make sure that everybody else in their community is verifying all of their certificates too, because all ships rise together, and they all sync together to now solving this large system for startups trust, it's a difficult problem. But it's also well defined and it's definitely solvable using established techniques. And just free information that technically minded among us, we're using vanilla Bayesian inference with loopy belief propagation as our algorithm. And the only thing that really makes this challenging is the scale. This is where the computational power advanced comes in. We have fast parallel computation available on every device Today, many being powerful enough to evaluate neural networks with millions of parameters. So this whole situation with high computational power in parallel was made possible by the adoption of neural nets widely, but it can be used for things other than neural nets. So I'd like to introduce trust, which is a developing project. It is the first startups inference engine. It backs all the visuals that you've been seeing, but it actually only crunches numbers. It has nothing to do with how it looks on the screen. It is a free and open source project. It's not a product, it's never going to be for sale.

And it's also

not machine learning, which actually amplifies bias machine learning picks up patterns in data and simplifies their analysis by compressing those patterns. If bias is present in your data, it will be amplified by machine learning. This, on the other hand, is just math. And you can even write it out in equations and understand it if you study it.

So let's do a demo.

First of all,

let's have a look at the certificates we've already seen.

These four certificates

are the competing certificates we saw earlier.

These are all the prerequisites to sharing code. And you can see if I manipulate them, they behave just as they were shown in the video. If I review this certificate, identifying where you are in Unix terminal, and I think that it includes a reasonable case, then I can read that one highly. And as you can see everything below it goes to 100%. But it's dependent why why doesn't go to 100% it stays at 50%. Because just because you know how to identify where you are. And Unix terminal doesn't necessarily mean that you know how to change directories, it's just a prerequisite.

So let's try adding something else.

This is the same set of certificates with an addition. So these are the four that are prerequisites to sharing code. This one is sharing code with get and this one is Camilla certificate. Sharing code with a get and this is my award from my certificate to hers. We've said that some of the best evidence for learning is teaching and start off. If I set this certificate to a high value, even if we know nothing about the other certificates, you can see that it already rates my certificate 443. Teaching people with get pretty highly even if you have no idea what's going on with any of the prerequisites. If you then start to review some of the prerequisites, let's say you do this one and you see that the prerequisites are satisfied. You can be fairly confident that this 443 certificate made by me is probably trustworthy without even looking at it maybe with 91% confidence in this case.

Here are our favorite walk certificates.

You can see that drying the walk has to come after lifting the walk. And they still have the same relationship. But now we can add in some other things that are related to these.

Here's a network of some certificates that are disconnected and a few that have this prerequisite and a word relationship. So here are the walks again, lifting the walk,

trying to walk. And

this one has been awarded to this certificate in Chinese, which covers a lot of the same knowledge. So again, if I set this one to high value, a lot of trust is conferred upon the other learners. But if I said, I don't know as much about this, when I set this to review this for its for its case, and I see it's not exactly detailed.

I don't know if this person knows how to lift a walk, it's me.

If I set that one to a low value,

you can see that everybody's values go low.

I want to show you some certificates that actually are about startups itself, and about how it works, its structure and what the different pieces do.

This particular certificate eight xR is about specifying prerequisites and start off how to give the most basic possible prerequisites. And you can see that here's my description and case that I know how to do this. But the prerequisites for this are even knowing what certificate structure is at all. And for that, we've listed two different certificates PvE and EDT. So here are PvE and EDT, this one is called Sardar certificate structure, which makes sense. This one however, is in German. And you may or may not understand what this means. I certainly don't.

But start off can potentially help you here.

If you were to review this eight xR certificate and decided that it was really worthwhile and then you looked at the This EDT certificate and didn't understand what whether it was good or not. But you knew that either EDT or PBE will satisfy the start off certificate requirement. review this one sort of certificate structure, you say, actually, that one's pretty trash, it doesn't do the job at all. I don't trust it at all, I put it down to zero percent. And notice that that actually rates this certificate, which you can't even read extremely highly, because there's only one way to explain this learners having such a high trust is that if they have one or the other of these prerequisites, and this one has low trust, then the other one must be highly trustable.

Here's something else.

These are certificates about making paper airplanes.

This one's involve prerequisites like folding paper, certificates, like making a masterful paper airplane, and how to fold a paper airplane with grants connecting them and you can see that you can mess around with these See how trust flows in this system.

You could join

us for an upcoming workshop if you want to see how all this works and make some paper airplane making certificates of your own.

Here's a view of every startup certificate that's ever been made and entered into our system. Obviously, it's extremely overwhelming. There's just too much to take in here. But this computer that I'm running this on is actually pretty strong and it isn't fast, but it is able to do inference over this entire structure simultaneously.

You can see that if this were optimized code

this entire this entire network could potentially be evaluated in the blink of an eye.

Finally, here's our makerspace.

The certificate is using The Quarter power drill. And this one is makerspace, General worksite safety. This g 95 is made by Daniel hood from moonshot Academy where I was a professional in residence in Beijing. And he describes the use of a quarter power drill in a safe way. And gives this excellent case where he just says all this text. Now of course, someone else could have written the text. So this is at best, a medium trustworthy certificate. But I happen to know Daniel, I know that this one is definitely correct, he really knows his stuff.

If I set that

certificate up to 100%, you can see that this other certificate, which is a prerequisite makerspace, General worksite safety is that 100%. And if there were many certificates that are just like this one, you wouldn't even have to look at them to know that Daniel knows how to use a maker space in general safely.

With the centralized certification systems of today, not everyone has access. In fact, it benefits from being exclusive. But everybody has access to decent neutralized peer to peer learning if they have curious friends. Or centralized systems can't deal with complexity of any kind, they're as simple as possible so that they're even possible. But start off and this decentralized accreditation model thrives on complexity because the more connections there are, and the less regular they are, the more information it represents about the details of who knows what and when they learned it. So our objective is to take our collection of diverse people diverse thought, who are usually barred from our centralized systems, and use them to build a new structure that's just so much better and will be much stronger.

So this is

my proposed path to adoption start off.

First of all, start off can be used to certify hacker knowledge. We can teach people to so to solder and to think for themselves. We can actually use start offs within our hackerspaces and act as both learner teachers We'll build the graph and reviewers who view it.

using these tools,

we can build networks of trusted makers between and among different hackerspaces. And then if you get this familiar question, Hi, nice to meet you. Can I use your jigsaw? Well, how can we know that you're actually good at this? Do you have any certs? That person provides you with a certificate. And it's related to and supported by a different hackerspace that you already know? Well, you can trust that they would know how to use your jigsaw safely. So this is great because those learners that I was working with back in Beijing can get credit for the amazing work that they're doing. But it's also good because it can help share hacker knowledge that most non hackers might not really think of. And that includes things like inclusiveness,

understanding systemic oppression,

the existence of multiple world histories and how we all understand our histories differently.

The many varieties of government and government

contagion safety just clearly not quite taught as well as we would hope. And other knowledge that's not taught in school picked up independently or from the community. Number to start us could be used in classrooms. It might sound like a bit of a contradiction in terms, but teachers can actually ask to grade start off certificates that does have advantages. learners will collect their own evidence or artifacts of learning as it's called in progressive education circles. This is actually a principal challenge faced by the mastery transcript consortium at this time, which seeks to provide a more comprehensive view of the competencies of a learner as they present a transcript to one of the colleges that accept MTC, but accumulating and organizing that knowledge in the first place is a huge challenge and start off could potentially help. Number three, start off certificates then stick with the learners after they've made them. So the hackers we've trained and the students who have used start off consulting MIT certificates with applications to things like employers and colleges, and then the colleges and the companies can become reviewers.

Maybe sounds optimistic, but this

is what could happen. an applicant says, Here's how you can tell that I know my stuff. And our viewer says, This is way too much, I can't possibly look at all of this. And the applicant says, Here are tools that you can use to review these certificates and the strength of my own community. And the reviewer says, Oh, I see. So it's actually a huge value offering to start using startups. So we'd love it if you could help grow startups, trusts our inference engine works, but it isn't accessible to others yet. It just works for me, our front end is not useful to new users yet.

And our team is not very diverse at the moment.

So people we need, we need patient users to try out and teach how to use start off. We could also really use some mentors who are experienced in open source software development who can help Give us a boost because we don't know how to do this yet. And we'd also love to hear from activists who see potential for their organizations and want to help spread knowledge that isn't typically covered in school, effectively within their organizations. If you're on this list, if you're a historian, especially one of education, if you're a probability theorist, if you're a lawyer, or in some related field, if you're a probabilistic programmer, or an economist, or game theorist, or a wizard and using Docker, or into front end or back end or side end, or just a curious learner, we could absolutely use your help. Finally, here are all the technologies that are currently in use in startups. So if you're on this list, good at Docker, Python, C sharp, flask, JavaScript, CSS, or dotnet framework, we could definitely use your help. And finally, to learn more about all of this, you can visit start off org where you have much more information and our contact.

Thank you so much. I hope you enjoyed

We're back with Dana data. Thank you very much. That was excellent commentary. How are you today? Fine, thanks. All right. Well, thanks for being with us. We do have quite a few questions from our audience that are that is in the matrix chat. And I'll go ahead and get started right away with our first question. And that is, how do you encourage experts to validate the certificates of learners in this system? Why would an expert take the time to do this?

So it's true that like in any accreditation system, the The true measure of trust is going to be coming from experts. And usually we establish experts in settings like universities, where a professor level degree has a lot of prestige. In this case, I guess if you think about why do professors teach students and why does being a professor confers so much It's actually because the professor is motivated to do good teaching, because it certifies them more strongly, among many other motives. And this is a similar dynamic and start off, we're not actually trying to say that any particular start off certificate can be trusted. And then we have to have people go and verify them to make sure they're right. It's that the dynamics of the system promote people teaching other people very well in ways that will be understood and reviewed positively when they are looked at. So not to say that you can't do anything wrong in the system. Anybody can make any Certificate of any trustworthiness, but the incentive structure is aligned such that people make trustworthy certificates and help other people learn.

Got it? Thank you for that answer. Our next question is, why do you think your certification would be received with any credibility by a society for which your paradigm is way too far outside traditional certification? For example, a person with a bachelor's degree in math Mathematics can present substantial documentation from a credible institution. Also, the traditional paradigm relies on Publish curricula. So what's your response to that?

Well, I don't know. Imagine like, if there's a group of people who are in like, say, Ghana, and they are phenomenal at doing architecture work, and they've at least one of them is self taught. Then they're going to like teach other people because of their, their passion, their and they can actually use start us to get credit for doing that teaching. And then they can submit the pile of certificates that they produce by teaching others in their community, along with their application elsewhere. So start off is going to help create places or kind of pockets where people can be certified just using well formatted Word documents and other really accessible tools they can use on their phone, etc. to produce trustworthy accreditations that can be presented elsewhere. As for why would they be be accepted. I think it's because it actually provides a lot of insight. If you receive a stack of startup certificates, and it's already connected to many other certificates that you know, like, for example, if you're like this math department that you're talking about, and you're taking people are considering applicants, and you know, some star high schools that tend to produce really good students, then you almost certainly have already reviewed people from many different parts of the country surrounding there and also from that high school. So if you get a new certificate, which is like very densely tied to a certificate that you've already seen, or many certificates you've already seen, it cuts down the work that you have to do, you hardly have to look at their stuff at all. So as your as the kind of cloud of certificates that you've received, and judged grows, your effectiveness at finding out exactly the kind of information that you're interested in actually goes up and there's a lot of

extra power that comes from that history.

Sure, absolutely, definitely helps we find all those candidates from that high school. All right, the question is, are there ways to cheat the system, particularly in the context of social networks, validating users and posts and concluding that they are bots?

Sure, um, I want to emphasize that a lot of the questions I saw in the chat, they seem to be surrounding the possibility of cheating and kind of overwhelming the system or 51% attacks, these kinds of arguments that are familiar things that we need to identify when dealing with decentralized systems, and especially things like blockchain. This system, yes, it's vulnerable to that sort of thing. But we're really talking about people submitting a little file folder full of doc files to somebody when they ask for it. And there can be a means of redistributing these things, but I want to get away from this perception that this is a system that involves Like massive servers forming tons of data, it's really just about, like maybe a couple hundred certificates produced by one, each of many different learners in one community, all submitted together, and with some cryptographic controls on them so that they only they only provide the data in plain text that they intend to. So I think that setting, probably the change of setting is the best answer that question where the setting is not actually just broad. public open access and like anyone's popularity can potentially weigh in to change the outcome. It's, it's entirely up to the personal view maintained by the particular reviewer and how unbiased or biased they personally are.

Sure, sure, it's always a balance between objectivity and subjectivity, as they say. So let's go on to our next question. How do you protect your system But from being skewed by the social popularity of certain people that don't necessarily reflect on the quality of their certificates.

I think that that is definitely a challenge. It's not something that can possibly be addressed by any technical system. I think that's solidly in the domain of the social. I think that this certificate system might actually outperform things like grading on, if you can call that a metric on a metric like that. Because in a traditional grading, setting, someone's different elements, their identity, like their race, and their gender will play deeply into the grades that they receive, which has been studied even just because of the expectations that they faced when they're that they face when they're being tested. Like, girls doing worse on math tests, when they're reminded that they're girls, that kind of thing. Yeah. So I'm sorry by contrast, start off with allows you to actually review the contents of each certificate. And so it's much more focused on evidence, although certainly if a person's popularity or other attributes about them are apparent on the certificate, these will weigh in, weigh in as well.

Got it. Ready. We're running close to the end. And we have a couple more questions here. The first question is, How easy is it to set up right now? Maybe he missed that part of the talk? Um,

well, I guess that could be taken a few different ways. It isn't extremely easy to set up right now. Although if you sent me an email right now, and we work together, or work together on it for about five minutes and you know how to use terminal. We could get it working together in a relatively short period of time. As for like more general usage, it's going to take some time, it's going to take some work to work it out. And of course, there are parameters of the system to tune like, how much or how, yeah, how much does it reflect on a teacher that their student did? Well? After all, students can learn quite a lot on their own. That's a parameter that you'd have to fit to see if you get the right dynamics. So there will be some time, just development time in general. And then there's also the, the setup time for an on an individual basis, which both need to be worked on.

Sure. All right. Our next question is,

is the project available on your GitHub or twitter feed?

You can visit start off org as ta rd s dot o RG. And you can get a bunch of information about the project as it's evolving. It's a relatively new website, so please be gentle. Yes, I'm excited to show that off, and you can find more links there.

Very good. Okay, we've got a couple of minutes left. The next question is how do you prevent the abuse or inclusion of activities that would make the system look tainted for general use, I promise Because the question started out with nefarious or anti social behaviors being included in the same system as public education. So how do you make that inclusion of stuff and not make it look tainted on there?

I think since a lot of my motivation is trying to seek out ways to view learners with less bias, it's actually very important to not try to exclude pieces of information or types of learning from the system. I think that information security ends up being extremely important for the system. You can't just submit all of the like, I know introducing this to like sex positive communities and like maybe communities people who may want to have means of spreading knowledge between them. That is very high fidelity but not easily to, to transfer or trust through any kind of normal accreditation system. They might really benefit from this but of course, that will come with really tight privacy controls that are necessary to make sure that if someone creates certificate content and uses it, and it's kind of kind of used as a touch point in this broader system, that it can still function in the system with the right pieces of it encrypted so that they remain safe. That's a kind of long winded answer. But I think that's that's basically what it amounts to is that we have to keep the information security high and don't limit the topics that are covered.

Right. Well, I think we're out of time. Actually, we are out of time. Dana greyton. With your star does certificate project. I really appreciate it on behalf of the whole 2020 attendees, the presenters and the volunteers. Thank you very much for sharing your product with us today. Thank you very much. All right, come back at the top of next hour for our next session. And this is hope 2020 Take it away ground control.

What my

That is 99 rise that many. Yeah, that many. Okay.

Put the plate up.

Thanks.

Remember how sharing things with your brother could bring it closer together? It's okay. Long Distance is the next best thing to being there.

Greetings from Philadelphia. This is Bernie s. And this is my botnet server of 50,000 honey bee bots in my backyard.

Sweet

Hey girls,

how you doing girls

like the paper

check out their friends