{
  "metadata": {
    "transaction_key": null,
    "request_id": "metagov:kuhn-metagov-20240320",
    "sha256": null,
    "created": "2025-10-27T23:38:01.023309+00:00",
    "duration": null,
    "channels": 1,
    "models": [
      "metagov-manual"
    ],
    "model_info": {
      "metagov-manual": {
        "name": "metagov-manual",
        "version": "2025-10-01",
        "arch": "manual"
      }
    },
    "warnings": null,
    "summary_info": null
  },
  "results": {
    "channels": [],
    "utterances": [
      {
        "speaker": "Speaker 1",
        "start": 0.0,
        "end": 0.0,
        "transcript": "Cool. Yeah. So I'm excited to welcome Tobias. He's a professor in VU University in Amsterdam and founder of Knowledge Pixels, which is a start up. And Tobias has been developing nanopublications for more than a decade now, I think, both in academia as well as in startup. I came across nanopublications about a year ago. It really resonated with a lot of my own research. And I got this I don't know if you've had this feeling, but it was like finding something that I've been thinking about for a long time and then seeing someone who's already done that and, like, a lot more, so a few steps further. So, yeah, I've been inspired a lot by by Tobias' work, and I see a lot of connections with what Medigov is doing, where we're also getting more interested in, like, collective knowledge management. There's an experiment with Black Science about connecting the Medigov Slack to, like, LLM and AI systems. And, yeah, nanopublications are about this larger story of how communities can effectively control, organize, and share their knowledge. And, yeah, if you guys yeah. If anyone has questions, you can also feel free to type them in the chat, and I will moderate the discussion after the presentation. And, yeah, go ahead, Tobias."
      },
      {
        "speaker": "Speaker 2",
        "start": 15.0,
        "end": 15.0,
        "transcript": "Thank you. Thank you all. It's it's an honor to be here, and I'm very excited to be able to present some of my work and my thoughts here. I have to say I'm I'm really really new to to this topic of Metagoff. So I I did not try to make kind of connections to to this topic too much, but I understand we have a lot of time for discussion and questions. So I'll be really interested then to hear how how it connects, and I hope I hope it does. I'm also yeah. If at some point, you want to interrupt because it's kind of a a question on something I'm showing, please feel free to do so. I would like to talk as Ronan mentioned about this concept and technology of nano publications and how it can contribute to a rethinking of how we do knowledge sharing in a global and in a general manner. By the way, oh, I wanted to to bring it over with this. I wanted to share the link here that you can also see, but so that you don't have to type it. Oh, now somehow it has disappeared. I'll just copy it out here again. Because because I have a number of links there, and maybe you wanna follow the slides as I talk about them, and maybe you wanna visit some of the links as we go. Or, of course, after the talk as well. So the very quickly, the type of problems we are addressing is that current knowledge sharing, imagine or think about scientific knowledge sharing, but it it's not necessarily restricted to that, is in many ways very inefficient, I believe. Scholarly knowledge, when researchers find out something, is mostly published as narratives in English, paragraphs, longer texts, and in in papers, articles that are published in journals and then as PDFs or HTML files available on the web. But not as something like database entries that we could just directly process in on a larger scale reliably. So that's the second point. There's no automated reliable processing that you could automatically aggregate, do a meta review automatically in an up to date manner, could immediately find out what's the latest state of science in this course on a particular hypothesis, for example. And on top of that, many contributions"
      },
      {
        "speaker": "Speaker 3",
        "start": 30.0,
        "end": 30.0,
        "transcript": "some Yeah."
      },
      {
        "speaker": "Speaker 1",
        "start": 45.0,
        "end": 45.0,
        "transcript": "Or maybe you could mute. Maybe I can mute. Next."
      },
      {
        "speaker": "Speaker 2",
        "start": 60.0,
        "end": 60.0,
        "transcript": "On top of that, many contributions to scholarly discourse in a broad sense go through other channels. Also because the channel of centering papers is super slow, So you don't wanna wait for papers to get accepted and published before you can make a kind of a quick comment on something. So many of these go through social media or blogs or other types of channels. And the problem is that these are disconnected. You cannot find all the things that have been said about a particular hypothesis, for example. And on top, they're also not rewarded. So as a scientist, you get you get recognition by publishing by publishing things that can be measured in one way or another, mostly papers and journals. So the we can probably expand this list quite a bit, but I'm gonna present a counterpoint technology that is nano publications that I've just mentioned already that are intended to solve this these problems. Now what are nano publications? In a nutshell, they are tiny packages, so that's why nano, of knowledge graph contributions. Alright? So contribution is because you publish them, so you contribute them to the to the broader public. They come with provenance and metadata. And what that means in a concrete case, you can see in this concrete example. It's it's a simplified one, but it shows that it's a real one, but with I I cut away some parts to make it to make it fit the slide. So this is an example of a nano publication where somebody stated a particular protein interacts with another particular protein. And what you see here is from a user interface called Nanodash, I'm gonna show you a bit later too, that tries to find nice labels for all these different things. But in the back, these are all formal identifiers. So it's not tumor protein p 53 is not stored as a string, but as a formal identifier. So these are all in a way sentences in a formal logic. Specifically, if you know about these languages, we're using the semantic web language called RDF. And the same holds for the for the parts for the next lines where we say that the assertion above, so the yellow part, was derived from a particular place, namely, from this paper that is a link to a journal paper. And then we also see that the nano application as a whole, that refers to the whole thing here. The whole box that we're seeing was created by me in this particular case, a particular moment in time, and there's a bunch of other such data items that I'm not showing here for simplicity. But you see the important part is that on this level of a single assertion, which can be multiple lines and not just one as in this case, but it's a kind of an atomic piece of information I wanna share. And on that level of granularity, we have the provenance and metadata. A few properties about nano publications. Firstly, they can be about anything. You can publish domain level links like what I've shown you between proteins or linking organisms to drugs, to environments, to social phenomena, to anything, basically, but also more low level data entries, metadata records. You can have reviews where somebody assesses some other piece of of knowledge. You can have opinions. You can define workflows in them. Read anything you can represent in such a structured form. And they come with reliable hash based identifiers and digital signatures. So they apply these kinds of cryptographic techniques that they will come back to. They are linked to their creators, for example, with identity schemes like ORCID, in case you know about that. So that provides formal identifiers for people. And vocabulary terms can be defined on the fly if needed. And we can use any type of ontology vocabulary that is out there as long as it's expressible in in URI. So in web or semantic web identifiers, then you can use them. So that's kind of the concrete technology of nano publications that's already available and we can use. But before I show you more concretely, I want some hands on, demo what you can do with them, I'd like to quickly step away back, or step take a step back and, talk about the broader vision that we have developed around this, which is called the knowledge space. So in a way, this is the vision where we wanna get to and also that the plan that we have to make this happen. And we basically rephrase this in in in more general terms just to distinguish the the vision that we have from the kind of partial and so far incomplete current system that we have."
      },
      {
        "speaker": "Speaker 1",
        "start": 75.0,
        "end": 75.0,
        "transcript": "So Tobias, maybe could you zoom in a little bit if you if it's possible to Oh, wow."
      },
      {
        "speaker": "Speaker 2",
        "start": 90.0,
        "end": 90.0,
        "transcript": "I I wasn't expecting you to to be able to read all that. I can't Okay. Please. Now I think it's better if I if I don't go into details. I want to say that in this link, you have kind of also a long white paper that that we have written. And just so you see, there's kind of a lot of aspects that we have thought about. But I will touch upon just a few parts, very high level parts of that. So sorry if you cannot see the details. You feel free to follow this link and then dig into it more deeply yourself if you're interested. But yeah. Sorry that I'm just showing you a high high level overview without going into these details. But I'll show you a few things that we'll also find in this document to just give you a quick idea what the knowledge space is. So in case you are familiar, and I'm sure most of you are, with some of these pink terms here, then this comparison might give you kind of a quick idea of what what I'm talking about here. The knowledge space is a bit like a knowledge graph, but one that is open, that is decentralized globally, and that's collaborative where everybody can contribute. It's also a bit like the semantic web in case you know what that vision is, but robust and scalable and trust aware. So we can actually seriously rely on on it and build stuff on top of it. It's also like a blockchain, but with an important difference that it doesn't actually have a chain. And so you don't have a chain of transactions, but rather in in the blocks, what you have are assertions or statements of any kind. It's also a bit, you can say, like a container environment like Docker that you use for software. But not for software, but for knowledge. So you can build stable applications on top of that knowledge. Oh, yeah. It's not like a large language model actually at all. But, yeah, just so you know. So at the high level, the principles are the following and only that's only as deep as I'll go with explaining the broad vision. Everything in knowledge space is communicated in formal logic. So concretely, in the concrete implementation, that will be RDF, this language where you can make statements out of any type of identifiers you want to. That humans can understand, obviously, but also computers can interpret and in a reliable and accurate way using a universal and extensible vocabulary. So you can state things about anything and if something doesn't have vocabulary or a vocabulary term yet, you can define it and then use it. And these statements are expressed in small, you could say, in atomic smallest possible knowledge records such that each record is individually reusable and referenceable. So you have all your knowledge and you break it down into the smallest bits. So each small bit you can reference and you can reuse just the bits you need. So that's the intuition you have. Yeah. And there is more thoughts as you saw in this complicated pictures around that. But the important part here is that nano publications are basically a first a first version of an implementation of knowledge records that contribute now to this vision of the knowledge space. So let's go back to oh, no. I have one more. Sorry. Some properties that were that or you could also say design decisions that are the foundation of the knowledge space according to which it was designed. So the knowledge space is intended to be open and inclusive So everybody can publish these small knowledge records. Everybody can add servers to this decentralized ecosystem. There is detailed provenance. So I've already talked about that a bit. Then importantly, we the knowledge base has the principle of publish first or let publish first, worry about quality later. So we decide whether something is good enough at the time of use, so the application can decide what level quality it needs, and not at the time of publication as we do it now in science with peer review where you have to get through a very rigorous or depends, sometimes not so rigorous process before your thing is even published in official sense. And the last principle list here is that it's bottom up with no central authority. So who is trusted is in a way community defined. And in case you know the term or the concept of web web of trust, It's a concept from cryptography. If you know that you can trust this other person with his or her credentials and that person trusts a bunch of other people, then you can basically trust these other people as well. And you can define different types of algorithms how to decide who to trust and who not to. And if the knowledge base, you can do that, but it's not that the system tells you what type of algorithm or what kind of set of trusted people you should use. But you can define your own your own web of trust and use another space within that. I will, unfortunately, not be able to go too much into detail, but if there are questions in this direction, I'm happy to elaborate. So now going back to the concrete ecosystem that is kind of an imperfect but already significant step into this direction. If you follow the the link here, if you want to, you see you're gonna see a map like this one where you see this decentralized network already running. Let me just quickly show you. So you can go here and you see this number of servers. So it's not huge yet, but it has a reasonable size that that deal of applications allow you to publish them, they allow you to query and retrieve. And, importantly, yeah, the this network is decentralized and open, so new servers can show in by anyone. You could set up one and it will show up on the map. And also the data is fully open. So that's our current yeah. This system as it's designed only deals with fully open data that you release openly under permissive license so it can be replicated without restriction. We are not dealing with any data that has any type of confidentiality or restricted access around it. If you wanted to design such a system for confidential data, it will be a much more difficult problem or a very different design probably would have to to do for that. So just as a as a as an explanation, we're dealing here only or addressing only open data. Now in order to make this network both scalable but also reliable, it is this it is split into two layers. One is a publishing layer of servers that only deal with allowing you to publish nano publications. And if you're looking for a particular one, to retrieve it back based on the identifier. So they basically just have a huge list of nano publications that they they have accumulated, and they let you access that list so so you can actually see which one is the last one, for example, and you can get the notifications if you know exactly what you're looking for. But you cannot do anything more complex than that. For that, you have the querying layer that gets the notifications from the publishing layer and then can provide you with different types of preservices. So most of them or actually at this point, all of them are based on the semantic web language SPARQL, but it there's no restriction there conceptually on that. Now, here's an example, a simplified example again. What you can do, I'll later show you one live little live demo. But in this case, we ask the the network, give me all LAN applications that have an assertion, where the assertion so this this whole thing is the assertion or the the the query we do on the assertion. It needs to be about tax some organ organism tax into environment association. That's a bit of a long concept name. So it links a species, for example, to an environment. And I wanna know what is the subject that would be a taxon and what name does it have, or is the predicate, or is the object. And here, see an example of what come out. Here would be musmusculus, so the the house mouse as a taxon, the predator has habitat and the environment city. So somebody, I think there was me, stating that mice live in cities among other things. To to deal with this network, the idea is that you have clients that sit up on top and gather the data from the network and provide you with feature as end user to deal with them to publish them also. And Nanodash is such a is such a tool. I will show you what it looks like, what you can do with it just now, basically. So this is the the live demo that I have. So just quickly walk you through some of the things and, well, you can do it yourself as well and you're very much invited to do so either now or later. Let me get to the down page. You see there's also a video. So I also recorded this demo roughly the way I'm gonna give it now. So in case you wanna rewatch it later, you can do that. The NanoDash interface, you can find if you go to this link, but there's there are other links further down. I'm already logged in here with my ORCID, but you're probably not. So the first time you get there, it will ask you to log in with ORCID. If you don't have an ORCID account yet, you can very quickly get one. They're intended as identifiers for researchers, but you don't actually have to be a researcher to to get an ORCID. It's really just a way to to to get an identifier. And now once you've done that, you can now try out Nanodash by publishing your own nano publications to give you a better idea what they can do. And this first one I will show you now is I'll just show it right away. Shows you a template where you I hope this is visible. I'll make it a bit bigger even. You can state an opinion that something is over or underrated. So this is kind of a half fun, half serious example to have something that you can easily publish and but also shows you the the the power and the universality of nano publications. Because I'm logged in, it already knows that IMDb is coon. And so according to this template, I can only make statements in the assertion that start with my own identifier here in subject. And then I can say, think this is overrated or think this is underrated. And then I can specify something here that I think is over or underrated. This template is defined in such a way that it queries Wikidata, which you might know, kind of this off shot of Wikipedia, for identifiers. But that's not nano applications are not restricted to that. You could configure a template to to do different to go to different place to get the identifiers or use any other kind of identifiers. So for example, I might wanna say decentralization decentralization. Let's see what comes up. Yes. This. I think this is underrated. And now in provenance, I can say, what is where did this assertion come from? In this case, it's my personal opinion so it just came from me and full stop. But for other things, I could say, for example, so this list here is not very well organ very well sorted. So many parts of the UI still need improvement, but you could, for example, say, derive from an existing entity. I could paste here, for example, the URI or the DOI of an article where I read this, for example. Or there are many other things, many other templates I could be using. Attribute to myself is the right one I wanna use here. And then as publication info, it tells me that I create this, it wasn't to choose a license. And I could also mark it as an example in a publication. So if I don't want this to be taken seriously. If I say something here that I don't actually think, I could just say it's an example. So don't don't take the content seriously. But actually, I do think this centralization is underrated. So I can publish it like that. Click publish, and done. Well, and you have to check here the box that where you state that you understand that this is now really published in a strong sense. So as long as the network of application survives, this statement will survive too. Which means others can depend on it, can be sure that it remains out there as well. And now to to see where this comes up, you can go, for example, now to my channel and it will not immediately show up. Oh, now it does already show up. So this is kind of like a a sort of tweak defeat but with nano publications instead of just tweets. And you see here that it shows up already. But more interesting, of course, is and I will now jump a bit down where where you have the section check out published nano publications. I can now go, for example, this first one. This is now a different service that oops. A little bit larger. That in this picture left something different. So another publication that I just published was pushed to one of these yellowish yellowish orange notes, but now it is kind of moved to all different types of places or been replicated all different types of places. And now I'm asking this network on a different from a different point. And now here, I can get the latest under overrated entries. So it turns out the last three were all from me as such. Product demonstrations underrate and applications overrated underrated and decentralization, underrated. But we can also have another one. We can look at the top underrated things. So this is not just mine, but other people. And if you do yours, if you publish what you think is over or underrated, they will show up within, like, half a minute or so. It'll show up here too. Yeah. You see open data is the most underrated thing. The most yeah. Is it underrated? Oh, yeah. Top underrated. The most underrated. And pizza, why, interestingly, is the second most. And while some of these links also show you all latest to nano publications have been published by anyone. And actually some of Ronin's might show up here as well. One last thing I can maybe show is that when you get here and you log in, then at some point it will ask you to whether you want to publish an introduction. And it looks a bit complicated in my case because of kind of a complicated setting. In your case, it will look simpler. But it will ask you to publish basically your identifier and your name together with the public key, this user interface. So NanoDash here uses to sign your nano publications and I have a number of others here. But importantly, this thing is also just a nano publication. And another thing that's a nano publication is that all these templates, you can actually find more of them here. If you click to publish, it'll click just on any random one of these. This nano publication at this template is also published as nano as a nano publication which you can see here. So to define a new template, it's just about publishing a new nano publication. And then you and others can use it as a template to create more nano publications. And the the last thing I wanted to show is that one for users to have published these introductions, they will show here on in the user tab. First, they will show up here at the very bottom under other users, but then they can get approved by people who are approved already. So I can paste in here the nano publication identifier of the introduction and then this person is considered approved too and it will show up up here under approved users too. So we have this first simple version of a web of trust, but at the moment, it's just one hard coded way of doing that. Well, it's not really hard coded. It's well, hard coded in the client side of NanoDASH. It's not hard coded in the network. But this is just a first step to make this more general. So in NanoDASH, for example, you could choose to use a different way to calculate the the the web of trust and to just decide who is approved and who is not. Yeah. So I think that is the end of my presentation. I think I'm already a bit over my time. Sorry for that. But so I'm I hope this was interesting and I'm very interested now to discuss any topics around this for the review. Thank you."
      },
      {
        "speaker": "Speaker 1",
        "start": 105.0,
        "end": 105.0,
        "transcript": "Thanks, Tobias. Yeah. This is super cool. Yeah. So people"
      },
      {
        "speaker": "Speaker 2",
        "start": 120.0,
        "end": 120.0,
        "transcript": "follow any any comments in the chat. Was I saw there was some activity. Were there any questions there?"
      },
      {
        "speaker": "Speaker 1",
        "start": 135.0,
        "end": 135.0,
        "transcript": "Yeah. We'll we'll probably surface it now. We can just ask live, I guess. I see Steve. You have your hand raised."
      },
      {
        "speaker": "Speaker 4",
        "start": 150.0,
        "end": 150.0,
        "transcript": "Yeah. I mean, I really liked the r idea of using RDF. And but, yeah, but your central knowledge space idea, I think, is very similar to what I put forth in my white paper."
      },
      {
        "speaker": "Speaker 2",
        "start": 165.0,
        "end": 165.0,
        "transcript": "Oh, interesting."
      },
      {
        "speaker": "Speaker 4",
        "start": 180.0,
        "end": 180.0,
        "transcript": "Yeah. What what I really noticed, though, in absence in this demo is there's no sort of, like, numerization of anything. There's no, like in other words, yeah, you say something is overrated, but you're not actually giving anything a numerical rating. And in my system, it's all about numerical ratings. It's all about sort of using a Bayesian approach where you say, I believe in x because of x one, which I believe y one, and x two, which I believe y two, or y one and y two are sort of my confidence percentages."
      },
      {
        "speaker": "Speaker 1",
        "start": 195.0,
        "end": 195.0,
        "transcript": "Mhmm."
      },
      {
        "speaker": "Speaker 4",
        "start": 210.0,
        "end": 210.0,
        "transcript": "So in other words, so I believe in x y rather because I believe in x one y one plus x two y two, etcetera, etcetera. And so I was just wondering if that's, you know, in your forward plan or if where you intentionally excluded that type of stuff or whatever else."
      },
      {
        "speaker": "Speaker 2",
        "start": 225.0,
        "end": 225.0,
        "transcript": "Am I okay. So it sounds really interesting. So is it like these numbers kind of with, I don't know, type of confidence intervals or whatever you call them, that would be part of the data you're operating on, part of the pay part of the data that you're operating on that you publish, or it's part of your"
      },
      {
        "speaker": "Speaker 4",
        "start": 240.0,
        "end": 240.0,
        "transcript": "Yeah. You'd publish it to you data. Either or. You could choose to publish or not. You could publish your interior criteria or not."
      },
      {
        "speaker": "Speaker 2",
        "start": 255.0,
        "end": 255.0,
        "transcript": "Yes. It's definitely covered. I mean, you can easily do this in RDF, and you can also easily do this in nano applications in Nanodash. The user interface for having, let's say, different types of numbers is currently not very well developed, but it's it's it's definitely possible. Yeah. I guess the reason why we don't have it that much so far, we do have there are numerical fields out there. There's also yeah. The the they are there. But, yeah, I guess so far, for for most of the applications where you kind of wanna get some quick output and with these numbers, they're probably more pay off when you have larger"
      },
      {
        "speaker": "Speaker 4",
        "start": 270.0,
        "end": 270.0,
        "transcript": "Oh, oh, yeah. Yeah. I mean, it should"
      },
      {
        "speaker": "Speaker 2",
        "start": 285.0,
        "end": 285.0,
        "transcript": "So so short answer, yeah, it's perfectly possible to have numerical things in there."
      },
      {
        "speaker": "Speaker 4",
        "start": 300.0,
        "end": 300.0,
        "transcript": "Okay. Cool."
      },
      {
        "speaker": "Speaker 1",
        "start": 315.0,
        "end": 315.0,
        "transcript": "Cool. Yeah. Any more questions people wanna raise or comments?"
      },
      {
        "speaker": "Speaker 3",
        "start": 330.0,
        "end": 330.0,
        "transcript": "Yeah. I I have a quick comment, and I'm guessing the answer is, like, yes slash this is easy, but I'm I I maybe missed the sort of the the technical requirements right now to spin something up like this myself. Like, sort of I'm guessing that it's pretty easy to run this across many nodes and sync them. But what what does that look like right now? Like, if I wanted to sort of run my own little mini nano nano publication universe for, like, my lab or my class or something like that."
      },
      {
        "speaker": "Speaker 2",
        "start": 345.0,
        "end": 345.0,
        "transcript": "Okay. Well, if if you wanna if you wanna use them just internally and and not share them to the outside, then while this whole server network infrastructure probably doesn't help you or, I mean, you could in a way make your own small yeah. It could probably quite easily apply just for your own university kind of a single instance and not have to do anything else."
      },
      {
        "speaker": "Speaker 3",
        "start": 360.0,
        "end": 360.0,
        "transcript": "Or I I guess when"
      },
      {
        "speaker": "Speaker 1",
        "start": 375.0,
        "end": 375.0,
        "transcript": "oh, sorry."
      },
      {
        "speaker": "Speaker 3",
        "start": 390.0,
        "end": 390.0,
        "transcript": "Go ahead."
      },
      {
        "speaker": "Speaker 2",
        "start": 405.0,
        "end": 405.0,
        "transcript": "Or or where you add or, I mean, you could, of course, also use this one even I mean, it's maybe for for you to organize yourself, but if you're fine with all these things being public and possibly reused and referred to by others, then you could use this one right away. And,"
      },
      {
        "speaker": "Speaker 5",
        "start": 420.0,
        "end": 420.0,
        "transcript": "yeah,"
      },
      {
        "speaker": "Speaker 2",
        "start": 435.0,
        "end": 435.0,
        "transcript": "there's not a requirement that the data must be useful to others or anything like that. So as long as you're fine with making it fully open, then you could actually build it right on top of this network, basically, right away."
      },
      {
        "speaker": "Speaker 3",
        "start": 450.0,
        "end": 450.0,
        "transcript": "Yeah. I guess I'm wondering about, like, a mix of that. Like, let's say that so I have a bunch of sort of, like, things that I do regularly that I would love to be captured via nanopublications. Right? So this might be, like, me reading a paper or telling, like, my grad student to read a paper, and, like, the lab is sort of, like, sharing visibility into the stuff we've been doing via this. And maybe they're all the grad students are all super happy to share with the whole world about the papers they read in a given week. Right? But I also wanna use this for students. Right? And there's, like, sort of in in a lot of places, there's very specific sort of data sharing and publicity requirements about how student data is handled, which is really special, where my grad student can share they write a paper publicly. But maybe my students my students sharing the fact that they did their assigned reading might technically be a violation of of data sharing, right, in, like, Canada and and and The US, depending on the exact context and how and if you've gotten approval before and things like that. So, yeah, is there, like, is there a way to sort of, have half the people running on a more internal network and then these things could be shared or or changed down the line? Or or is that, like, a big headache?"
      },
      {
        "speaker": "Speaker 4",
        "start": 465.0,
        "end": 465.0,
        "transcript": "Yeah. Can you control stuff by provenance? That would seem it to be the question he's asking. Because I know you had a provenance solution in there."
      },
      {
        "speaker": "Speaker 2",
        "start": 480.0,
        "end": 480.0,
        "transcript": "Yes. Yeah. Well, the the question with this is, like, for for what purpose? Right? So if it's just kind of to get students some, experience of doing that, then you could do it internally. Yeah. Much of the benefit would would kind of like, the practical benefit that you can now write one query center network and it it gives you the answer. Right? Basically I do have the slide. Basically, this one, you wouldn't have because yeah. Well, you could either you could either have this query to your network or to the public one, but not kind of do any combination of that. Or there will be some extra thing you'll have to implement to have that done. So that's why I bet what I'm say why I was saying the the the real benefits or or the the biggest benefits from from nano publications come from when you're able to share them publicly. So they can be kind of distributed at different places. They can be loaded. They can be queries services specifically focusing on particular type of nano application and load them all in and provide the queries service on that. As as soon as you have even a tiny bit of a restriction on who is able to see that, then much of that doesn't work anymore. But if it's just to get students a bit of exposure to these kind of technologies, then it's, of course, another thing. Then you might, I don't know, even throw the data away every year once the the academic year is over."
      },
      {
        "speaker": "Speaker 3",
        "start": 495.0,
        "end": 495.0,
        "transcript": "And Yeah. Yeah. Be actually Another context would be doing studies that are covered by, like, a research ethics board with with this as, like, doing user studies on this. Yeah. Sorry. Sorry to interrupt."
      },
      {
        "speaker": "Speaker 2",
        "start": 510.0,
        "end": 510.0,
        "transcript": "Yeah. No. So but so guys, it's actually something that that comes up again and again that people kind of like to apply this also to cases where the data is not fully open. And I think node case can also help with that. Just so far, yeah, this network is not designed for that. I think many of the things would have to be basically redesigned from scratch to make that possible. If you wanna have some sort of way I don't know. To to say I have now, I don't know, there's a database over there and the oh, there's kind of yeah. And that and NetApp server over there and one over there, and I have access to both of them. I can log in and and be there. But now I wanna create them together. But now somehow I need to be able to access them both at the same time. So somehow the systems need to know of each other or I need to somehow have a a way to authenticate more than one system at once and and make a query over them. So a lot of the things that that we can currently provide wouldn't work well in that setting or at least you wouldn't get kind of many of the benefits that we're now claiming the network as a whole has. But it's kind of interesting. It comes up and up again. So I I think, indeed, it's a it's a it's a thing that that that applies to many situations that if we have if we were to be able to come up with a solution for that, that will make it even more powerful. So just a student case, student want to kind of try it out, I think that that is easy to cover."
      },
      {
        "speaker": "Speaker 5",
        "start": 525.0,
        "end": 525.0,
        "transcript": "And how do you connect those nano publications? Is there any layer of operating with them? For example, if you want to have something bigger statement combining several."
      },
      {
        "speaker": "Speaker 2",
        "start": 540.0,
        "end": 540.0,
        "transcript": "Yes. So the different ways how to do that. Well, you can you can refer to not sure which slide this should show here. I It just is a very simple one. You can refer to nano publications in general. Right? So that's one of the benefits, very kind of trivial ones that each single piece of of data, each nano application has its own identifier. So you can refer to that. So and then applications can use these identifiers to refer to each other. So now if you wanna make kind of a bigger statement out of a smaller, it depends in what way exactly you want to do that. So for example, you could have kind of a a more general hypothesis that is kind of supported by a number of more specific findings that that have been kind of found. Right? I know there's some social phenomenon and has been shown to be true in Japan and Germany and in The US and a couple of other countries. Now you wanna make a more general statement that this is actually true in general in all societies from that. So in that case, it could be as it could be in the provenance that you say this is a general statement. We say it because it seems to be supported by these and these and these. You can of course also have different types of statements that in the in the assertion you state something like, I don't know. Some for example, a study, this study confirms this hypothesis and this hypothesis is linked to another publication where the hypothesis is defined. And in that sense, it's also kind of a compound statement that then makes a bigger statement, if you wanna say it bigger. But I think in general, it's more helpful to think of it as a kind of a complex network. So if things point to each other in in in different ways rather than something kind of compositional where things are packaged together in bigger and bigger structure. It's not entirely wrong to think in that way, but I think it's more accurate to think in the network."
      },
      {
        "speaker": "Speaker 4",
        "start": 555.0,
        "end": 555.0,
        "transcript": "I was I was gonna say that. In fact, like, you there's definitely hierarchy to the sort of, the nodes in my paper where you sort of have at the bottom, you have the kind of the experimental nodes, which then, you know, combined with the hypothesis nodes to make the theoretical notes, and they all have different types of debate going on with different types of participants allowed to, contribute to that debate, you you know, once again, determined by personal trust networks. But, eventually, everything aggregates up to the dot fact come question mark node, which is sort of these quotable statements, assertions that letter, like, hydrogen has one proton that everybody gets on to. And the idea is that you could sort of cut and paste these into web pages anywhere, and then you would have then an entire, you know, almost, you know, provenance of the idea down to the basic experiments of everything that's related to that that contributes to that one fact."
      },
      {
        "speaker": "Speaker 5",
        "start": 570.0,
        "end": 570.0,
        "transcript": "Yeah. This can also help to find conflicting nano publications. For example, they say"
      },
      {
        "speaker": "Speaker 4",
        "start": 585.0,
        "end": 585.0,
        "transcript": "Yeah. Correct. Exactly. That I go all all in that stuff in my paper dealing with conflicts and so on and so forth. Though I think the RDF approach would allow a lot less of those conflicts to occur on first blush. So once again, I haven't looked at this paper in, like, two years. So but yeah."
      },
      {
        "speaker": "Speaker 1",
        "start": 600.0,
        "end": 600.0,
        "transcript": "I wanted to add a a quick comment about that, the sort of ability to find, like, agreement and disagreement. Yeah. I just I just saw someone on Twitter. The the head of bioRxiv was posting that there was a study that showed a bunch of papers with methodological issues, and they need to be retracted, or at least that should be signaled that they have issues. And he was asking, like, what's the best way to signal this? We don't have a good way to do this. Like, we have to wait till someone is gonna write a review article and, like, mention all those papers and go and do all that work. Word-of-mouth, like, what do we have? And and it's like, yeah. We need we need data publications for this kind of stuff. And and it also connects to what you said, Tobias, about them being like, they could be part of applications, and I think that's a point that, like yeah. Maybe if you could say a little more about that. Like, there can be applications that use nano pubs as, like, the inputs, and then they would, like, have the signaling thing over a paper. And you're looking at the paper, and it's like, oh, there's actually Nanopubs that signals that it should be retracted or has an issue. So, yeah, just just a thought about, like, how we can use them in in downstream applications. And, yeah, do you have idea do are there any applications built on Nanopets currently or it's still in progress kind of?"
      },
      {
        "speaker": "Speaker 2",
        "start": 615.0,
        "end": 615.0,
        "transcript": "Yeah. Well, Nanopets is kind of a special case because it's kind of an end application, but rather kind of a tool you can use for all kinds of applications. They are they are built into the this fab no. The data stewardship wizard is a tool that allows communities around FAIR, in case you're familiar with that, to make research findings findable, accessible, interoperable, or usable, and the whole community movement around them. And they allow their communities to define in a fair way, with their applications, what kind of tools they use to make their research data in general fair. So it's kind of on several meta levels at the same time. So anyway, that that visit, that's basically kind of a questionnaire type thing. And whatever you fill in, the result is expressed in applications and published to to the network. And several of the drop downs where you can select stuff in the visit is also fed from the applications. So that's one example. Another one is now being built. I haven't seen the system yet, but it's from colonial heritage in The Netherlands. It's like a collection of museum that will use nano publications to allow native communities to kind of make their own statements and comments about colonial objects that come from their from their native community. And they were really kind of the decentralization aspect, which for us at that point was more like a a technical nice thing to have, but not something that you can sell. And they for them, that was actually really important to not give any impression that they're now colonizing knowledge on top of the colonized objects they already have. Yeah. These are the most important exams that come to my mind right now. So and yours, of course, Ronan, is would be another one worth mentioning. So, yeah, there are a couple, but not not lots of them yet."
      },
      {
        "speaker": "Speaker 1",
        "start": 630.0,
        "end": 630.0,
        "transcript": "Cool. Yeah. I can I can plug our our project? We're working on application that helps that uses, like, AI to publish the to create nanopublications from natural language. So making reducing the barrier to to creating nanopublications. Instead of form filling forms, you can type it, and and it will try to fill everything out for you. And we're seeing this, like, in the social media setting where a lot of scientists are already basically nano publishing. They just don't know about the concept yet, and they don't know about RDF, and they probably wouldn't have the time to to bother with it if they did. But AI can already bridge some of these gaps and actually make it painless to nano publish in addition to posting on social media. Yeah. And are there any more questions? We have maybe time for one or two more before we talk about"
      },
      {
        "speaker": "Speaker 2",
        "start": 645.0,
        "end": 645.0,
        "transcript": "some messages on the chat, which I haven't managed to read through. So if there's anything unanswered there, maybe you should just speak up and then so I can answer."
      },
      {
        "speaker": "Speaker 1",
        "start": 660.0,
        "end": 660.0,
        "transcript": "Yeah. Nick has some really cool ideas just about the using, like, in the more university setting, like, a a more like a less public or kind of dev dev repo as opposed, like, fully public repo. So we were talking about that a little bit. Cool. Yeah. So if there's if there's nothing else, we can end early, and thank, thank Tobias."
      },
      {
        "speaker": "Speaker 4",
        "start": 675.0,
        "end": 675.0,
        "transcript": "Well, I I I did but I did Go ahead. You know, discussed the signing and so on and so forth and and his, the, you know, the data, the structure, the blockchain without the chain, go into a little bit more that aspect of things. Could you talk a little bit more about that? I'm reminded again the work of of Udi Shapiro, which who come from a couple of months ago here who's doing stuff like with this. But, anyway, go on."
      },
      {
        "speaker": "Speaker 2",
        "start": 690.0,
        "end": 690.0,
        "transcript": "Yeah. So the so nano applications in a way, if you wanna, yeah, compare them to to blockchains, you can see them as as blocks. They also have this hash based identifier. So that's very similar conceptually. So So for the"
      },
      {
        "speaker": "Speaker 4",
        "start": 705.0,
        "end": 705.0,
        "transcript": "If it's not a change because it's not append only, is that the difference is that you can go in and you can change things later? So everything is being shared, but it's but it's not a permanent record. Everything there's not a a permanent past state recorded."
      },
      {
        "speaker": "Speaker 2",
        "start": 720.0,
        "end": 720.0,
        "transcript": "It it is append only, but it's not forcing the things into into an order in the form of a chain."
      },
      {
        "speaker": "Speaker 4",
        "start": 735.0,
        "end": 735.0,
        "transcript": "Oh, okay."
      },
      {
        "speaker": "Speaker 2",
        "start": 750.0,
        "end": 750.0,
        "transcript": "Which"
      },
      {
        "speaker": "Speaker 1",
        "start": 765.0,
        "end": 765.0,
        "transcript": "Okay."
      },
      {
        "speaker": "Speaker 2",
        "start": 780.0,
        "end": 780.0,
        "transcript": "Makes things much less, let's say, computationally intensive or there's you don't have the cost of maintaining a chain. But then you don't have kind of a very easy check that you have exactly the same nano applications. Right?"
      },
      {
        "speaker": "Speaker 4",
        "start": 795.0,
        "end": 795.0,
        "transcript": "Okay. So so so once you sign something and append it to whatever this, you know, Merkle treed, you know, object is, it's it's stuck in there forever. Okay."
      },
      {
        "speaker": "Speaker 2",
        "start": 810.0,
        "end": 810.0,
        "transcript": "It's not even Merkle Tree ish. So it's not a so basically then the the nonapplication servers that just deal with this publishing and index based retrieval, they just keep each of them a long list of all applications they know. But these lists don't they are similar, but they don't need to have the same order. So your nano application might up at different positions in different lists, but the kind of the the the the permanence and the kind of long term availability is simply guaranteed by having it at as many places as possible. So even if one of these servers makes a mistake at some point, first, it will there will there's still the others around, and second, it can be corrected by this server at some point being reset or figuring out it missed something. So it's a more, let's say, dynamic or more yeah. Not purely technical approach in that sense. And also servers are not kind of committed to follow one exact protocol in a way. You could have a different server that only that only keeps notifications of a certain type and kind of doesn't bother recording the others, which would be perfectly fine. Right? The whole system would would still work. And in a way, the the fact whether something is published or not, I've defined this now in this in this white paper more like as soon as it's as it's available on a number of independent places, and this number you can maybe now set to three and later maybe you wanna set it higher, then you consider it published because it's now a places where it can further be copied. So and they're independent, so there's not kind of one player who could take it down unilaterally. So that's kind of, yeah, kind of as strong a permanent statement as you can get. And with blockchains, you probably have a bit more, but then comes the whole cost of maintaining a block blockchain. So but then comes the whole cost of maintaining a block blockchain. So blockchain in a way then"
      },
      {
        "speaker": "Speaker 4",
        "start": 825.0,
        "end": 825.0,
        "transcript": "Yeah. You just have a simple distributed ledger system. It's it's it's it's fine. I I use these type of things all"
      },
      {
        "speaker": "Speaker 2",
        "start": 840.0,
        "end": 840.0,
        "transcript": "day. If, yeah, if ledger is still a proper term in the setting, maybe it is. I don't know. But, yes."
      },
      {
        "speaker": "Speaker 4",
        "start": 855.0,
        "end": 855.0,
        "transcript": "Cool. Also, just another note, you know, if you when you do add numerical options, you now have the ability to do things like make ratings for products independent of any platform with this sort of thing. You know? You're so, basically, a testing that I this is my my rating for this, and you can then share that just like any other nano publication around."
      },
      {
        "speaker": "Speaker 2",
        "start": 870.0,
        "end": 870.0,
        "transcript": "Yes. Yes. I I actually have one of my to do is to do that for for books that I've read to kind of publish. I've read this one, and for me, it's a four out of five star book, but I just didn't get around to do it. But, yeah, it's all there, basically. It would just be about making a template and a few kind of decisions on how to model it, but I think this is all there. Yeah."
      },
      {
        "speaker": "Speaker 4",
        "start": 885.0,
        "end": 885.0,
        "transcript": "I mean, just because I mean, product ratings now are just total shit, and they're gonna get worse and worse as as, you know, these platforms get more bot infiltrated. So this is really gonna be the only alternative to, you know, whether there was something as good or not as to use your trust network."
      },
      {
        "speaker": "Speaker 2",
        "start": 900.0,
        "end": 900.0,
        "transcript": "Yeah. That's a yeah. Maybe if this whole if our focus on scientific knowledge doesn't work out, maybe we'll go into product ratings."
      },
      {
        "speaker": "Speaker 1",
        "start": 915.0,
        "end": 915.0,
        "transcript": "We can do both? Yeah."
      },
      {
        "speaker": "Speaker 2",
        "start": 930.0,
        "end": 930.0,
        "transcript": "We can do both. Yes."
      },
      {
        "speaker": "Speaker 3",
        "start": 945.0,
        "end": 945.0,
        "transcript": "Maybe it's"
      },
      {
        "speaker": "Speaker 2",
        "start": 960.0,
        "end": 960.0,
        "transcript": "not so different."
      },
      {
        "speaker": "Speaker 4",
        "start": 975.0,
        "end": 975.0,
        "transcript": "Alright. Yeah. Good stuff. I look forward to exploring it more on my own here."
      },
      {
        "speaker": "Speaker 1",
        "start": 990.0,
        "end": 990.0,
        "transcript": "Cool. So, yeah, I guess we'll we'll wrap up, but this was, yeah, this was great. It was a great presentation and discussion. Thanks again, Tobias, for coming around, and thanks."
      },
      {
        "speaker": "Speaker 2",
        "start": 1005.0,
        "end": 1005.0,
        "transcript": "Do you"
      },
      {
        "speaker": "Speaker 4",
        "start": 1020.0,
        "end": 1020.0,
        "transcript": "ever have any community events besides this, Tobias?"
      },
      {
        "speaker": "Speaker 2",
        "start": 1035.0,
        "end": 1035.0,
        "transcript": "We have if you go on the oh, no. You don't see my screen anymore. On nanopop.net, there are we have nano pop sessions. Mhmm. So that's kind of the equivalent of of these, but in in our on our nano pop topic. So, yeah, anyone Alright. Happy to go there. There's also a mailing list if you wanna just kind of Yeah."
      },
      {
        "speaker": "Speaker 4",
        "start": 1050.0,
        "end": 1050.0,
        "transcript": "Yeah. Yeah. Of course. I'll do that. Alright. Great."
      },
      {
        "speaker": "Speaker 1",
        "start": 1065.0,
        "end": 1065.0,
        "transcript": "Cool. Well yeah. So yep. Thanks, everyone, and see you around. Thanks, Tobias, again."
      },
      {
        "speaker": "Speaker 2",
        "start": 1080.0,
        "end": 1080.0,
        "transcript": "Thank you, Ron. Bye bye."
      }
    ],
    "summary": null
  }
}