This seems problematic. :)

One of the most important questions. We wrote about it here and here. Many people voiced their feelings about this in the six months or so that we spent investigating it and exploring it.

The tension is between what users want and what authors want. The way the web works now, authors control what is delivered by the web server, users control how they consume web content. If I want to install the "Drumpf" chrome extension, I can, and web sites can't (easily?) block me from doing so. Greasemonkey is another good example of this.

If the browser comes w/ the native ability to fetch, anchor and display annotations, but as the user I have the ability to decide which services I subscribe to-- then should page authors be able to block me from that? And if you think they should, should that blockage only be for public commentary, but not personal notes or private group annotation?

One particularly thorny problem is around how governments and other public entities might want to use that same blocking power to prevent you from marking up their sites and documents. Should your government, or any government, be able to block its citizens or others from critical analysis of documents that it hosts? And if not, then how do you distinguish them from just any old page author.

For the moment, a countermeasure that authors can employ is to add shrapnel to their web pages which blocks some of the strategies that are used. Wordpress plugins are available that do this. In a crude sense, this may be a reasonable compromise.

Should a user be able to override the page author's preference? Probably.

A single ... client? Yes! A multi-service capable annotation client is one of the most important things we can build next.

There may need to be a couple different approaches. Certainly the scroll-to-text work is a great example of URL-decoration, which could (and should) be extended to enable pulling individual or collections of annotations from services. A potentially useful approach we created is what we call the "bouncer" URL, which sniffs whether a browser is "annotation capable" and if so, passes a decorated URL to the browser, and if not attempts to proxy the page on the fly w/ an embedded annotation client.

The bouncer URL for this annotation is here: https://hyp.is/VKkMzrkhEeuzahdJq5gLPw/github.com/bokand/web-annotations

Pages should definitely be able to announce that there are definitive or relevant annotations about their contents from specific services / communities, users should have ways of automatically or manually subscribing or listening for those annotations. They should be able to announce that to either browser clients, or locally embedded clients or both-- preferably in ways that will become standard.

Clients should find elegant ways to let users know that annotations from services and community groups that they're listening to are available for the page that they're on. Doing this in a way which is helpful but not distracting, especially as the number of annotations increases will be an evolving problem-- clients will probably be noisier about it initially, and much more restrained later.

Moderate them at the service level, either through community moderators, or service-level systems / filters, or both.

Services should prevent or disincentivize spam through a range of different strategies. Some services will do better than others. Users will tend to favor them.

At the individual annotation level, I'm not sure it should moderate at all. However, there may be levels of UX that it can implement which aid in moderation, for instance, the way potentially low quality comments sometimes are collapsed by default on reddit or other places.

imho this is the responsibility of services that store and deliver annotations to clients, and the operators and communities behind those services. users should choose services, and they'll choose those services based on their quality. some websites [insert the worst most awful website you can think of] we find offensive, but our user agent still delivers them to us faithfully, unless the website is spoofing a certificate or is included on community lists suggesting it contains malware. it's important for many reasons that the web operate this way.

Indeed :)

Interesting, because you're using the word link two ways here. One as in hypertext link, which is a "pathway for traveling from one document to another", and the other is more as in a "connection between a comment and the passage in a text that its about". For this latter concept, perhaps anchor is more appropriate?

Also, if a preprint is published, and that's clearly visible, then it may become obvious that the utility of a review is naturally lower from the author's pov-- and maybe you stop pinging them.

Good point!

The availability of overlay systems like Hypothesis suggests that review infrastructure could be separate and distinct from the preprint service, even though visually it appears to be "over top" of the content. If the review functions are also then carried out by different communities, then maybe this mitigates the sense that the preprint services are encroaching-- rather they're just facilitating the discovery.

So, here's another bit.

So here is a useful piece of instruction about the Pythagorean theorem.

Hi Mark. This is a link.

here's another annotation.

Hi Ben

Can you expand your thoughts here?

Discussion of Kraken rolling fees,

This seems like an important point

Yes!

Hi Geo!

Test

Like this.

I should have known this would work-- but it's kind of cool that the youtube track playing in the background here is also getting transcribed in THIS video.

Isn't it amazing that YouTube itself doesn't really expose this incredibly important part of their data in a UX which is user-friendly?

Actually the docdrop referred to our original service, which just allowed you to drag and drop a PDF and get annotation added (while hosting the document) :) But I love you riffing on the concept.

You didn't show off the cool search feature at the top of the page! :) for instance in this video, if you search for the word "highlight" it looks like this:

I personally think this is one of the most powerful things-- and it's not even annotation related!

Whoops you didn't save that annotation. #UXfail on our part. :)

test

The search is my favorite feature. SO useful even separate from the annotation!

I'm curious-- did this ever get addressed? Can't find any subsequent reference to its resolution.

It's a slash or "forward slash" technically. Backslashes are only used regularly on windows machines actually, because apparently bill gates decided back in the day to invert a tradition that was already very well entrained in the unix community.

Change to: "selectable text layer".

Nope, OCR is a process by which the pixels of an image of a page are converted to actual semantic (symbolic) text. For instance the system can read the text in a native (true) PDF but it has never been OCR'd. Here's a good primer on the three different kind of PDFs, In short, they are image PDFs, native PDFs and image PDFs that have been OCR'd (searchable).

It's not called an OCR layer when it's there natively, it's called a "text layer" or perhaps a "selectable text layer". In fact, it's always just called a text layer.

I think you want to distinguish, by analogy, between the Web and a website. Is the Agora equivalent to the Web? Or is an Agora equivalent to a website?

Is it really more beneficial to know? Let's separate two things: 1) the ability to know, and 2) the ability to amplify. Trump has a website for instance at whitehouse.gov. He can make statements and post them, and other people who care to know can discover them there. Trump's fundamental speech is not denied. But just because he has a right to be heard, and we do want to be able to know what he's saying, doesn't mean we want to force a private company to give him unrestricted access to one of the most powerful social platforms in the world. In other words, he can open his mouth and say the words, but does that mean he should have unrestricted right to the highest form of social amplification? No. In fact, one could make a strong case that it has been social media's reluctance to restrict his bullshit that has indeed encouraged him to take it to the n-th degree.

But not incitement to riot, which is illegal, and which is arguably what the president did yesterday prior to the march on the capitol by the crowd. https://codes.findlaw.com/us/title-18-crimes-and-criminal-procedure/18-usc-sect-2101.html

Glenn doesn't bother talking at all about whether private social platforms should have any rights to enforce their own rules or decorum on their platforms. Hypothesis would have banned Trump a long time ago-- not because he's a republican, or a conservative, or even an idiot, but because his participation on the platforms has not been constructive so much of the time.

He hasn't been barred outright from using social media. Presumably most social platforms in the world are still available to him. However, he was suspended from Twitter for 12 hours for 3 specific tweets which were violations of their terms of service-- and which directly led to or followed the riot. True, facebook banned him till the end of his presidency for reasons not having to do with his use of their platform. While within their rights, this seemed much less defensible. However, neither represents a complete ban on all social media.

He's an amazing intellect, but he's not infallibly right in all things!

Well certainly I agree intellectually with the fundamental point here. But, in the case of what Trump did yesterday, is Glenn arguing that Twitter let its emotions get away with it? If anything they've been incredibly restrained, no? Glenn is suggesting Twitter was overreaching, but he doesn't go so far as to draw the hard line. So, hard to tell.

Barbara Lee is a total hero in my book for this vote. Here I completely agree.

Again, this is where Glenn loses me the most. There's no reason Twitter needs to allow Trump to use their service, which has certain properties that make it in many ways much more powerful than Fox or Newsmax or a white house press conference, just because he also has those outlets.

Again-- the government can't regulate your speech, your political speech, or whatever, unless it violates key norms which are well known through case law, but private companies bear no requirement to be the vessels through which that speech travels. This is important, because it's the basis by which some services might impose their own community guidelines to more narrowly constrain the speech they will permit.

Wait-- we're not empowering social media platforms to increase their censorship powers, these private companies-- which created their respective platforms out of whole cloth-- have always been able to deny speech to whomever they want. They are not regulated entities.

trump is essentially ceding the dominion point here.

In other words, Brad is saying they did a hand recount so it doesn't matter about the Dominion machines. All the machines do is make counting faster initially on the day of the election, so if you count the votes by hand (they did it twice) then it doesn't matter if there are issues with the machines.

Brad Raffensperger's beginning response

This is flawed for a number of reasons.

First: branching, threaded forums like Reddit and HackerNews are happening in "the real world".

Oh, perhaps you mean like when people are hanging out together in the same room? What about when groups of them have side conversations-- or maybe when someone's talking two people in the audience will whisper to each other.

Even more to the point: in a discussion we often naturally conclude one line of thinking and then pop the stack to pursue other lines immediately after. This is branching. Threaded conversations naturally accommodate this, where flattening this might in fact make it more confusing.

Indeed online and IRL conversations are quite different because they have different properties-- fundamentally that live conversations have to be synchronous, whereas online ones don't. If online paradigms give us advantages shouldn't we take advantage of them?

The question really be: Which is best, when? Not: "Because it seems unfamiliar we should dismiss it."

Let's state this a different way. "Branching allows multi-party 'conversations' to naturally explore the logical space around a topic." Sure, it may require time to go read all of the 100s of branches and 1000s of replies in a well-trafficked Reddit post, but that's only because you're a human with limited time and attention. At least those sub-conversations were able to happen in the first place between their counterparties, who seemed perfectly happy to have them, and not "discombobulated" in the least.

Note that each branch becomes a referenceable object that can be linked to and which others can discover later as a useful exploration of its own topic space.

For instance, this one.

One might even go a step further and say that the largest discussion forum on the Web, aka "the front page of the Internet" was only possible because branching let discussion flourish in the way that it did.

Demo

demo

Further down

Hi Fed Wiki Folks. Here's a note.

Beginning of Ian Linkletter

Beginning of Audrey Waters

Beginning of Jesse Stommel

Beginning of Cory Doctorow

"both Arab and Israeli" FTFY.

I remember that Lab! The corks were all over the ceiling and it was such a fun idea. Plus those scientists drank a lot of champagne!

I actually had one of my first jobs in that lab-- titrating samples from the sea floor w/ acid to see how much CO2 would be released from the calcium carbonate, as a proxy for biologic life in the form of shells of phytoplankton.

test

Not in percentage change. Denmark increased by 44% or more? Sweden increased by about 28%.

This is a phenomenally important UX insight and affordance that has become a foundation of how all modern wiki-linking knowledge graph tools work today. Kudos to Ward for this!

This section 6 is a key bit on how Roam can function as a much larger 2nd global brain.

Here

Hey, this is cool.

Hello

This is one of my favorite projects! Here's the background on it: https://qdr.syr.edu/ati/ati-initiative

This is one of the first high quality examples of civic annotation focused on the critical analysis of news--here wrt Climate reporting. (And they happen to use Hypothesis to do it) https://climatefeedback.org/

Here's the list. Getting a bit out of date. I didn't really even set out to create a list, but people kept telling me about more and more annotation projects and eventually it found it's way into a doc, and then a spreadsheet. A lot of the early efforts are in here, maybe not so many of the more recent ones.

I agree! Here the archive.. https://web.hypothes.is/liquid-margins/

Here's the Liquid Margins 008 Math episode..

https://www.youtube.com/watch?v=CfHy3WN1XGE&feature=youtu.be

Science in the Classroom is a AAAS project here.

Here's the page at MIT Press for the book

Welcome aboard Remi! Here's the full announcement: https://web.hypothes.is/blog/researching-annotations-power-with-our-first-scholar-in-residence/

This one I think: https://www.emerald.com/insight/content/doi/10.1108/ILS-04-2020-0117/full/html

Here's Marc Andressen's post calling for some guinea pigs to test it.

Here's the preprint of that research. Also, a preliminary article on the findings.

Hi Mark.

Here.

There was this group of republicans in suits running around the conference (Young Republicans maybe), it was obviously a clash of cultures. I think the speaker before me had made a fairly obvious political statement of some kind and I just wanted to acknowledge them.

This is related to his discussion of the Ouroborous.

https://youtu.be/NLc6NEuec3E?t=13

I think a really interesting question is: What is the genius of using an ouroborous to frame this conceptualization of the extreme scales of the small and the large? It wasn't obvious, but I think it's beautiful. I'm still trying to figure out exactly why.

Make an annotation.

Like so.

test

I've always thought so too! Right William?

test

Hello world

Still working on this part.

This part of this etrajkc;kjfd

test2

test

test

test

test2

test

One such place is docdrop.org

But the key question might be another: "In contrast, for those same papers, what were the degree of changes accumulated between the first posted preprint version and the published paper?" Perhaps the rapid feedback of the early preprint process resulted in something more substantive, and perhaps as a result the obvious weaknesses of the papers were corrected before they ever got to the publisher.

This reminds me of Churchill's attribution (apparently said as a quote of someone else) "Democracy is the worst form of government, except for all the others." https://richardlangworth.com/worst-form-of-government

I'd heard of ComMentor before, but I hadn't noticed that Terry and Martin cited Vannevar and mentioned the notion of trails here. https://www.theatlantic.com/magazine/archive/1945/07/as-we-may-think/303881/

Hi

Hi Steve

This feels significant enough to have its own WP entry. Without true writing you cannot precisely convey knowledge to people who aren't within earshot.

Note: This assumes you have installed and are running your own instance of Hypothesis. If not, you'll need to have a Hypothesis team member do this step for you.

Make an note.

Oh, you mean this one? https://uxmovement.com/thinking/the-aesthetic-accessibility-paradox/ I'll just go ahead and link it to make it ... accessible ... for everyone. :)

There is a fair amount of theming already provided for and described in our client side docs: https://h.readthedocs.io/projects/client/en/latest/publishers/config/

Backed up at the Internet Archive, here: https://web.archive.org/web/20150711145316/http://ca-political.org/what-we-do/

A subtle but important point here is that when the manuscript is a preprint then there are two public-facing documents that are being tied together-- the "published" article and the preprint. The review-as-annotation becomes the cross-member in that document association.

Would be great to get a few citations for this.

Presumably a key question is why a personal attorney would be involved in a state matter.

Test

Why the hell not!

You've probably hit on our most secret desire here. That we could have this kind of cross-cutting impact.

Like this

Like this. And type whatever I want.

This was cited by Justin Amash recently in his decision to leave the Republican party.

Here is the passage in the farewell address.

Hey, the point you should have made here.....

We have new group versions that will allow student's annotations to be public in their own layers so that other people can read them, but cannot respond unless they're invited to the same group ("restricted groups"). We are also creating another type ("open groups") where anyone can participate-- but you'd still get the benefit of the "clean page".

I assume you know you can create a fresh group for them and direct them to that, so that they get a "clean page" instead of teaching in the Public layer. You don't need to fork the content to do this.

I'd be interested to know if there are public statements or studies that more formally characterize the issue w/ textbook expense. (Not that it's not obvious, but I'm always curious to see data and different perspectives).

Test

test

No... that was my Dad's home. :)

Actually no, we trained in TKD for 2 years before I moved up to Ashland, Oregon for a couple years. When I came back I didn't reconnect w/ Bruce for over a year, because I was feeling a little sheepish that I was still living at my Dad's place and hadn't really established myself yet. I finally called him up one day w/ a very early demo of the ITN concept. To his credit, he saw the potential in my crude prototype and morgaged his house (together w/ my credit card debt) to allow us to get started.

Well here we are, talking about zoom annotation! So, here's an annotation on a zoom transcript where we're talking about zoom annotation!

Meta!

Hi Craig-- here's a public note to you that any one else could see-- but we could also create a private group here and have a conversation just between ourselves and others.

fljsdlfjf

Like this.

Ann starts speaking here about her experience teach with Hypothesis in the classroom. Close this sidebar and click on the text to advance the video to this point (53:29). Ann speaks for about 4 minutes. Worth watching till the end.

blah

Yo.

blah.

Annotating my mention of annotating transcripts. Meta!

Like this.

Like this.

Does this jibe with this other fact?

Blah

This contradicts what we know.

What if another SAFE happens in 2020?

Why talk about a series of SAFEs?

Here.

Hi Dick!

It's interesting how true this has been over time-- certainly in the US. Could it be true that it's because mothers and (most of the teachers) had no vote? Are there other countries or cultures where this dynamic has not historically dominated?

This is really spot on. It's not ideological, it's simply business. It's clear that Elsevier is going to have to adjust to lower margins in this part of their business. Whether their CEO and board are fully cognizant and clear eyed about that yet is definitely a question.

Our open letter to Genius and Marc. https://hypothes.is/blog/a-letter-to-marc-andreessen-and-rap-genius/

.

When we were thinking through how to do fuzzy anchoring effectively for annotations, we looked to Sebastian's research here. The working string length of Google's bitap fuzzy search implementation in their diff-match-patch library is 32 characters, which sits in between these two values of context 20 and context 40. It turns out this is happily a great tradeoff between the relatively high selectivity (URatio) and the relatively good stability over 100 random changes (Stab 1...100) (since as selectivity increases with string length, stability decreases).

This research over the wiki corpus gave us confidence that on a fairly large corpus of english text we could be relatively sure that the prefix/suffix searches would still select fairly well and yet be as robust as possible against breaking changes in the context strings themselves.

Great paper on orphaned annotations in H.

The UI is fully available now by the way. Not experimental.

What are the things that would make the biggest difference to you in terms of ease of use? Would love to see if they jibe with our priorities.

Our default is really targeted at users from the Chrome extension, who are carrying it around the web with them, and turning it on or off whenever they like.

I can imagine that we would have a different set of defaults for embedded use cases. And more documentation and design examples. So much to do!

As you say there are a complete set of client configs options that allow you a wide variety of design choices-- including toggling all visible parts of the UX from view for those embedding like you did on your demo page.

Small correction: Outline.com embeds Hypothesis, but we're different organizations. One of the great things about Hypothesis is that you can embed it anywhere with a little javascript. Another interesting side project is docdrop.org, which l created to let people easily host and convert document formats to something annotatable.

test

test

test

test2

test

test3

test2

test

test

test 4

test

test2

test

An even smaller neighborhood are the folks lurking in an annotation over your post, since they're disconnected from the comment roll below. Possibly an integration opportunity?