 Okay, so if anybody can get on the Wi-Fi and wants to follow along and go to hypothesis slash alpha today is the first day that we're pushing something that people can actually use out there. It is a Chrome extension, so if you click on that link it will launch the little interstitial page from the Chrome store and you can install it into your browser. Also there's a bookmarklet here that you can grab and put up on your bookmark bar if you are not using Chrome or if you want to use it. Also you'd be able to embed this as an embedded JavaScript on a page, so bookmarklets are important. There's one there if you want to use it. So basically what I'm going to do is just take you through some of the little pieces of the application and really just talk about design decisions. This is early prototype code, it's buggy, it's not perfect and hopefully it will get better. The important thing for us is to try to think about how to solve challenges that we might have as we try to annotate all of human knowledge. So let's try to break those down into some different components. So the first thing that might be, so let's try to find a web page here on the net. There we go, our friends at PureJ. The first, let's pop this, extension, it should show on our page here at some point, so that will speed up. So I'm signed in to the extension right now. Our goal is to implement annotation on the web in layers. So there's a user interface layer that might be in your browser, there's also a storage layer that should be able to be accessed through an API. You might want to use the user interface and access multiple storage providers. So let's say that you're inside of a company and the company annotates the documents, corporate documents that are not available outside the firewall. You might actually be signed into multiple storage providers at the same time. You should be able to do that through the user interface. Right now we only allow accessing a single provider at a time. But the provider that you're signed into is visible up here. So Hypothesis is a project, it's not a brand. There's no branding on the user interface. This is open source software that you can download. You can run as a layer on top of your own content against your own storage layer. And if you're signed into a store, that store shows after the username up there in the upper right-hand corner. So let's just create an annotation here on this page. I can just select some text and say something pithy and then save that. And then that annotation is visible if I roll over the tab here on the right-hand side. So what are we trying to do here? A couple of things. One is that annotation up till now has often been kind of implemented as a sticky note kind of user experience paradigm. So all the annotations are competing for real estate on the page. If we imagine 100 years in the future with 10 million annotations on top of the Bible, and it can't all be sticky notes, it just doesn't work. So how do we disintermediate the annotation layer from the page layer? Good question. So that's a user interface challenge that we have explored. As of other people with a sidebar approach, you can pop it out, see what the annotations are that are on the page, and scroll to find additional ones. I can reply to this here, so there's threaded conversation. We've chosen a threaded paradigm right now. We also think flat conversation is interesting. There's no reason why you can't take this same conversation paradigm and implement a flat view. We're experimenting with a couple different user interface approaches for that. You can toggle between the two or choose whether the annotation that you've created should enable a threaded conversation after it or a flat conversation. If I want to share this annotation or this subpart of the thread, I can click the share icon. There's a URL. All annotations are first order objects, so all parts of the reply chain are also annotations, and they all have their own unique URLs. Cool. So if I wanted to tweet this annotation in my like this, then I could do that. Hopefully at some point here in the future that will go out to the universe once the Wi-Fi kicks in. If I go to a web page and I type that URL in, as though I've clicked on it on Twitter, and I hit return, then that pulls up an interstitial page right now that looks like this. It says hypothesis at the top of it only because it's coming from our servers because our storage layer is the one that we are signed into. Pardon me while I kind of do that, I'll only do that once. We will be implementing a intercept mode by the extension so that if you have the extension installed and you click on a link that is an annotation link, then the extension will intercept that, take you straight to the source page, pop the sidebar out and actively scroll to the annotation without having to go first to that interstitial page. We wanted to have that working today. We weren't able to get it done, but that should be ready and available on GitHub in the next couple of weeks. Right now this is, in order for you to see, and if you go to this peer J page and you want to see this annotation that I've just made, and you have the extension installed, you have to reload the page in order for those annotations that I just made to refresh so that you can see it. We are just almost, we wanted to have it done for today, but should have shortly implemented an actual live update capability using PouchDB and CouchDB, Pouch in the browser and CouchDB on the server so that the annotations that I may show up live on through the extension if you're on the page that I'm annotating. So that stuff should be done in about a week or two. So if we go to a page, so let's go to another page that has more annotations on it. Here's one of our friends in the room. So here's, we can see that there's already a bunch of annotations. There are 10 annotations below me. I can use the up and down tabs to scroll to them, pop the sidebar out, let's see what's there, click, there's the threaded conversation. I can collapse that if I want to create more room or quickly go to the next batch of threads. The other really interesting thing, especially in the scientific world, in the world of journals is that a lot of times you have an HTML page and a PDF page of the same article. So ideally I'd want to annotate the HTML page and have the same annotations show up automatically on the PDF and vice versa, I'd want to be able to annotate that PDF and have the annotations show up on the HTML without having to do anything special. So we're working on that too, we have an initial implementation of it. So this will work. So right now the PDF component works in Firefox because Firefox just pushed PDF.js out about a month ago, which is their native PDF renderer inside the browser. So if I go to the PDF link on this, cool, let's try PeerJ, because PeerJ does PDFs too. Ian, you might check your PDF server. Okay, so let's do this, let's create an annotation, make sure I've got the right book mark, so let's say I've created it two, five minutes, I'll try this, naturally we have demo problems, so I apologize for that, it's working, I promise, it was working last night. So let me just explain kind of why this is cool and why it's important. Content exists in many different forms and we want to be able to annotate the same things wherever they are. Also things change, so the same article on the New York Times webpage may get updated throughout the morning with additional content. The annotation that you made to a sentence may have a spelling error that gets changed or the sentence on either side of the annotation that you made changes slightly or the paragraph gets moved later in the article. So we want annotations that are robust to changes and also able to exist and to span format types when we're annotating the same thing. So this is kind of a grand challenge of annotation that people have been working on and thinking about for 10 or 20 years. There's papers on it, it's kind of a whole scholarly focus area. And some interesting work has been done recently to think about how to create kind of fuzzy annotations by using the content itself as the anchor as opposed to using structural elements of the page like XPath and elements of the DOM. And so what we've done is taken some of that best thinking and implemented a prototype toolkit library which is available on GitHub now called Fuzzy Anchoring to basically take a sentence or the selected bit of text in the framing context on either side of that and use that as the thing that references what's being annotated. And to use fuzzy logic on those context anchors that frame that text to be able to stick to that document and that text. So that's just been finished, the early alpha version of it, Kristoff has been working diligently on that and it's live, it's in the application, I wanted to be able to show that to you. But all of the annotations that are being made right now through the system are using that logic. So even though we haven't got the PDF thing just working just right yet, the fuzzy anchoring is built into the system. If the page changes a little bit, actually the annotation that you see, the selection that you see actually shows strike-throughs where what's coming back from the page is different than what the annotation originally pointed to. So there's actually some diff logic in there too. So I'm going to stop there and just say thank you to Josh Greenberg who's our, who at Sloan who's been an extraordinary supporter of ours over the last year for their confidence in us and turn the mic over to Nick and then we'll have Paolo just after that.