New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Blog discovery for the future? #96

Open
scripting opened this Issue Oct 15, 2018 · 40 comments

Comments

Projects
None yet
6 participants
@scripting
Owner

scripting commented Oct 15, 2018

Don Park wrote on Twitter -- "I feel that discovery layer is missing or lacking. blogrolls didn't scale. ring blogs just sucked. SEO is just survival-of-the-fittest money pit.

"Discovery layer is critical. Without it, even recent push to reshape blogs into shops is meaningless. even small towns have a Main Street for discovery.

"Blogrolls worked more like book recommendations. Hard to maintain too. Worked well with new technology. With other and over multiple topics, not so well. We need a more self-organizing and ad-hoc, emergent if you will."

I'd like to do some thinking and experimenting in this area. It's a good time for me. Maybe others who have blogs want to do join in.

@akaKenSmith

This comment has been minimized.

akaKenSmith commented Oct 15, 2018

If active citizens need to find good people and reliable information, to stay in touch with them, to affiliate or ally with them, then the importance of Don Park's thread can hardly be overstated.

Having found each other, kindred parties need a work space where they can develop shared understandings. And they need tools and temperament for building something concrete together in the world. I wonder if Don Park is pointing out a tool or design problem best understood not only in steps or parts (discovery, etc.) but also as a much-needed whole? But maybe discovery is enough of a challenge for the time being.

In bad times, citizens are easy to neutralize if they remain in isolation.

@akaKenSmith

This comment has been minimized.

akaKenSmith commented Oct 15, 2018

I remember the food bloggers used to announce a topic a few weeks in advance, then everyone would post about that on a particular day, and somebody would paste all the links for the new posts into a single blog post. I suppose there was a tag of some kind, too, I can't recall. Far too much typing involved in that approach, but it was lively and fun to look forward to, to participate in. It created some worthwhile assemblies of postings around a topic, by multiple authors with different skills and perspectives, real fast, almost like a small anthology, which is no small thing. That example may be too far afield from the blogroll example that got this discussion rolling, I don't know.

@donpark

This comment has been minimized.

donpark commented Oct 15, 2018

FYI, I brought up blogrolls as a mechanism of blog discovery. While it's a subtopic, we should explore a wider range, including but not limited to fixing blogrolls.

To be most effective, I think blog discovery needs to happen not just on the side, like blogrolls, or at the bottom, like mentions, but in the blog post content, using links. This means discovery needs to happen before or while blog posts are written.

This was relatively easy, albeit time consuming, to do when blogging community was small. Each blogger'd subscribe some blogs and write posts linking to posts we've read. At global scale spanning many languages, it's not so easy but possible. What I envision is what Techmeme already does except we need it before or while a post is being written.

Let's not delve too deeply into this proposal yet. We want to explore more before identifying key areas and start digging.

PS: Yes, I am aware that Techmeme is not fully automated. :-)

@scripting

This comment has been minimized.

Owner

scripting commented Oct 15, 2018

@donpark

This comment has been minimized.

donpark commented Oct 15, 2018

How could a piece of software find communities of bloggers?

Link analysis alone can be used for clustering. A level above that can use t-SNE to cluster blogs.

Aside: I think blog channel or tag may be better for clustering.

For more info on t-SNE clustering, search for "t-SNE cluster". t-SNE is just an algorithm to map multi-dimensional data set into 2D map, mostly used in ML.

@scripting

This comment has been minimized.

Owner

scripting commented Oct 15, 2018

@davmillar

This comment has been minimized.

davmillar commented Oct 15, 2018

I like the idea of something like Techmeme, at least partially automated, but needing that vital bit of curation as well.

Re: blogrolls, I mentioned on Twitter that a blogroll.json type file in the .well-known folder might be a good option for normalizing blogroll format and forming a network of trust. Another option might be using XFN (edit -- sorry about that super dead link) on blogroll links to establish that network.

From there, it feels like feed software might be able to query the feeds' hosts for such a blogroll.json or similar thing that can provide a branching point for finding a community of bloggers. If done at subscription time and periodically refreshed, it would make a good list of potential new blogs to follow as well as content sources in which to search tags to get that Techmeme-like overview of a topic at a community level.

@davmillar

This comment has been minimized.

davmillar commented Oct 15, 2018

Some folx on #indieweb mentioned this page that might spur some ideas: https://indieweb.org/follow

I'll also link them to this conversation.

@donpark

This comment has been minimized.

donpark commented Oct 16, 2018

I don't think search works for discovery because it is a snapshot, meaning new posts in the area will not be found until the next search. What may work better is mapping a given post/blog to a cluster of posts/blogs to subscribe to. That information can be useful to both bloggers as well as readers.

It's not that search is not useful. In particular, search is useful for identifying blog posts reacting to an event. So X happens and bunch of posts shows up within a few hours using words relevant to X then it's very likely they're all reacting to X.

@jgmac1106

This comment has been minimized.

jgmac1106 commented Oct 16, 2018

At same time blogs self-organize around topics. For example yesterday we were looking for examples of top ten list post. We discovered that #TopTenTuesday was a popular hashtag among librarian bloggers.

If we could have a blogroll or webring of librarian blogs and then a way to search that ring for post on #TopTenTuesday, I get on topic post from experts.

@scripting

This comment has been minimized.

Owner

scripting commented Oct 16, 2018

The important thing is to pick up on things people already do and use that info to create associations that might help create community.

Whether you call it search or not isn't that important. Every server-based centralized app with a database is in some sense search. So let's not worry so much about words.

@jgmac1106 -- that's something new for me. I hardly ever use hashtags in my own blog, and was unaware that others were.

@scripting

This comment has been minimized.

Owner

scripting commented Oct 16, 2018

Not sure I want to start using hashtags however until there's an answer to the question -- "How do I use them?"

@jgmac1106

This comment has been minimized.

jgmac1106 commented Oct 16, 2018

I talked about this a back at the last credible web w3c community group meeting when someone was presenting a "new certificate" idea that wwould work like SSL.....like we need to do that all over again.

Granted #CredWeb is looking for a LinkedData solution. I believe in HTML and human solution.

If I was a leading non-profit journalist org and I managed a ring, I could have set criteria for joining the ring (credibility), a logo like a badge, and links to others who met that criteria.

Some folks complained "this could be spoofed" easily. The logo yes, but not joining the ring, especially since we are using indielogin for authentication. If you don't control DNS not sure how you get in.

To me this gets to discovery and curation by experts.

In terms of personal websites I believe a following page, grouped by topics, connected to a read/write social reader is best way to start indexing topics and blogs.

@scripting

This comment has been minimized.

Owner

scripting commented Oct 16, 2018

@donpark

This comment has been minimized.

donpark commented Oct 17, 2018

Blog clusters are emergent. Fake or not, blogs with posts on similar topics will be mapped to same cluster which can be seen as a place in which a blog belongs to. Once we have that, a blog reader should be able to 'pop out' of that blog and see some visual representation of that cluster with neighboring blogs, not unlike a shopper leaving a store will see a street lined with other shops. That's how discovery is done IRL and I envision that may be possible online.

Blogs with multiple channels can have a doorway for each channel, sort of like a building with doors that lead to more than one street. Too fancy? Maybe.

@kickscondor

This comment has been minimized.

kickscondor commented Oct 17, 2018

Hi, folks - just jumping in because this is my wheel house a bit. I have been having an extended discussion with Brad Enslen (so, on our blogs: ramblinggit.com and kickscondor.com) about discovery. We talk a lot about how this is more of a human problem than a technology problem - and that technology has played a negative role in this, perhaps.

(My part in this is: I have been spending time every day for the past six months searching for blogs - to see what the Web looks like outside of social networks. So I have a good perspective on where one can search nowadays - you can't just type 'blogs' into Google. And I'm starting to get a good feel for where I would want to go to find blogs.)

blogrolls didn't scale. ring blogs just sucked. SEO is just survival-of-the-fittest money pit.

Yes, so - for sure. (See Brad's comment on Google here: https://www.kickscondor.com/when-the-social-silos-fall/#comments.)

In addition, self-promotion has become a dirtier word these days - you can't just post your blog to Reddit and Instagram - it's seen as being overly assertive. So there is almost nowhere for blogs to go.

The thing is: no, blogrolls didn't scale - but I think they are pretty essential. We've traded a human-curated list of links for a 'friends' list that is really just a number on an individual's feed. And the best blogrolls had nice descriptions of who was who (see: https://boffosocko.com/about/following/ as a good example by Chris Aldrich) which is a generous way of turning your readers on to other good work.

I guess I just think of it practically: how would we treat our friends and the other 'writers'/'artists' we admire - by making them a number in our list? Or by spelling it out: "Annie writes about her processes as a sci-fi writer and how to improve online relationships. Basically - it's uplifting to read her."

Blog clusters are emergent. Fake or not, blogs with posts on similar topics will be mapped to same cluster which can be seen as a place in which a blog belongs to. Once we have that, a blog reader should be able to 'pop out' of that blog and see some visual representation of that cluster with neighboring blogs, not unlike a shopper leaving a store will see a street lined with other shops. That's how discovery is done IRL and I envision that may be possible online.

Sweet - feels practical. One question I have here is: ok, so blogs have also become more topic-based. The most common blogs are recipe blogs, movie blogs, etc. But a great 'lost' element of blogs was just the original web journal or meta blog, where a person is just writing about whatever - I think of stuff like the old J-Walk blog or Bifurcated Rivets. Even Boing Boing used to be more this way. (So like an online 'zine'.)

I think the orderliness of the Internet and the systems for discovery - these blogs were not found through Google, but only because there was more of an ethic of linking to each other among early blogs. A lot of discovery was just being done by bloggers back then - people simply passed links around.

Again, 'likes' have drained linking of a lot of its bite. We don't write so much about why we like something - we like it and move on. And it's so easy to 'like', it is done so vigorously that even we can't keep up with our own likes - whereas we used to be limited by how much energy we would spend dressing up our links.

I'm with Don on this -- whatever is going to have a chance to work has to be emergent, meaning it can't require any investment on the part of writers.

I think 'emergent' can require work - in fact, it might demand work. Yes, too much work will dissuade anyone. But if it's too easy, then it's virtually worthless. I think the value of human curation is in its additional care.

An algorithm cannot simulate the care. Chris' blogroll linked above is done with care - a human can plainly see that another human has taken the time to write about others. And the more time he spends designing it and improving it, the more it shows that care. People can visit my blog and see that it is built with care. (To me 'care' can be represented by thoughtful writing and splendid artistry or shaping of the information.)

Ok - sorry to go on so long, I hope you see this as my effort to generously engage in your discussion.

The effort Brad and I are now engaged in is an effort to bring back the link directory and to attempt to innovate it based on what we've learned. (Link directories have already evolved several times into: blogrolls, wikis, link blogs, even the App Store's new 'magazine' approach, etc.) The idea is to jump right into discovery and link up with anyone else who wants to get in on it. Thus, my reply today!

Good to meet you all - take care.

@scripting

This comment has been minimized.

Owner

scripting commented Oct 17, 2018

No problem with the long reply and clearly you've thought about this a lot.

The "emergent" thing I was agreeing with Don about was something that would take no work on the part of the blogger. The idea would be to read the text, figure out when the author is pointing to a blog, and from the text around it, develop an idea of what the blog is about, if possible.

Not saying I know how to do it, or would want to figure it out, or know it's possible. However if such a thing existed, I could learn its language (the same way I learned how to influence Google in my writing), and by writing what I normally write program it to make it smarter about blogs.

Writing and curating a blogroll like you describe is something I appreciate when others do it, but have never had the focus to do it myself. Key point (mentioned previously) I don't blog as my main thing, I use it as a place to record ideas that I have that I want to share and want to come back to later. So having a tool (concordance generator) that organized my blogging bits after I write them would be incredibly useful. If such a tool existed, it could be used to organize the writing of groups of writers.

Many years ago I was trying to add categorizing to my blog software so it would be absolutely as easy as possible. I factored and tried ideas out, and convinced myself it couldn't possibly require less effort. Yet I stopped doing it. It's the kind of thing where if you skip doing it a few times, you give up on the whole thing. The reward comes when you do it for everything for enough time. I never seem to get there.

I have thought that having an open Like system would actually be a good idea. A way to like a thing that isn't on a social network, that's just on the web. It's probably worth trying out. Like requires no effort, but it is a useful gesture. No reason it has to be exclusive to silos. It could be done by an open service.

Anyway I try not to count anything out. There are all kinds of people in the world with lots of different ways of looking at things. It could be there's something we haven't tried, or need to try again.

@donpark

This comment has been minimized.

donpark commented Oct 17, 2018

An algorithm cannot simulate the care.

This jumped out at me. I agree but the care can be signaled in text, tags, and through UI which algorithms can use. Likes are signals. So are links.

Re blogrolls, I think following changes can make it more scalable and bring it back as a useful widget without increasing maintenance cost:

  1. Show recent post(s) - this would make blogroll content more interesting and inviting to readers.
  2. Sort by update time - this would take care of orphaned blogs. It's heartbreaking to find some interesting blog and, wanting more on the subject, walk down the blogroll only to find a ghost town.

I think we need to fire all cylinders to rejuvenate blogging, meaning enhanced blogrolls is just part of the remedy

@kickscondor

This comment has been minimized.

kickscondor commented Oct 17, 2018

Cool, cool - I appreciate the interesting replies!

The "emergent" thing I was agreeing with Don about was something that would take no work on the part of the blogger. The idea would be to read the text, figure out when the author is pointing to a blog, and from the text around it, develop an idea of what the blog is about, if possible.

Ok - I see. Yes, I had misconstrued this. Now that I reread this thread, it seems you are talking about various shapes of an algorithm - from Don's idea to map 'a given post/blog to a cluster of posts/blogs' to your concept of automatically creating a book or 'a tool (concordance generator) that organized my blogging bits after I write them'.

I'm not anti-algorithm, but I think they've panned out pretty poorly overall - if you think about the rash of algorithms to produce 'related' blog content (you often see these at the bottom of Wordpress posts, pictures of the royals or of some Trump scandal), it's dominated by clickbait. So it's like we've developed this inverted spam filter that has honed its filtering so much that the only thing that can pass through the filter are the bits that can game the algorithm the best.

This seems to be exactly what Netflix is describing in this article on their thumbnail composition: https://medium.com/netflix-techblog/artwork-personalization-c589f074ad76. "How do we hack the human to click the image?" rather than "How do we simulate a friend recommending a film?" I mean the best way (imho) to meet a blogger is literally to have a friend say, "Hey I want to introduce you to this really interesting writer I know."

Writing and curating a blogroll like you describe is something I appreciate when others do it, but have never had the focus to do it myself. [And then on categorization:] It's the kind of thing where if you skip doing it a few times, you give up on the whole thing.

Yeah, this is tough - I don't know of a good answer. (But yeah - let's still try to figure it out.) I feel like this is the 'human problem' we have in front of us: it's work to spend time on discovery but we don't want to do it, so give it to a computer. And now we're learning that maybe discovery is the most important thing. And we need the computers to be us in order to do it properly.

After all: a scientist makes a discovery - we all value that. But now others need to 'discover' the discovery. And that's a lot of work to write about it, get it through the filters, say, "Hey, this is important!"

(Oh and on the topic of 'likes': I agree with you. I think shallow reactions are cool - more here: https://www.kickscondor.com/microsponses/)

Continuing these thoughts, though - Don's response:

Re blogrolls, I think following changes can make it more scalable and bring it back as a useful widget without increasing maintenance cost:

Love this idea - I've been meaning to do this with my own blogroll!

Here's another one: what if we moved away from blogs being only about recency? This has already taken a modicum of shape in the form of Twitter 'pinned' posts. Here we have a single post that acts like a introduction - and not generated by an algorithm, but simpler than that - just a regular tweet that stays fixed.

What if you could make a post that would go to a 'blogroll' type of area - or a post that would go to your recommendations - and you could build a more permanent collection. In fact, what if blogging became just that - the day-to-day activity of a larger 'wiki' of content. (There is a website called philosopher.life where a guy is doing just that kind of work and I think what he's doing is innovative.) Then just the permanent collection could be mined by an algorithm to help you find other nodes across the network. And it wouldn't need to be a complicated new protocol, just plain HTML pages with links.

@scripting

This comment has been minimized.

Owner

scripting commented Oct 17, 2018

Here's another one: what if we moved away from blogs being only about recency?

Yes. I do that actually, on a separate site, http://this.how/.

For things that are going to be long-term campaigns or docs.

It's not a wiki, it's all my writing. But it's not temporal. The pieces are intended to be updated over an indefinite period of time, and this, it turns out, I actually do.

@kickscondor

This comment has been minimized.

kickscondor commented Oct 17, 2018

this, it turns out, I actually do.

Mmmn - this is cool. Yeah, well, I would guess that you do it because it is useful to you - it is a permanent record of what is most important to you. That would be a treasure to me.

@scripting

This comment has been minimized.

Owner

scripting commented Oct 18, 2018

Followup to this conversation. Terry Heaton sent me a note yesterday reminding that this month is the 15th anniversary of BloggerCon I. This morning I did a bit of reconnaissance, looking for interesting stuff to archive from the event.

  1. I did a this.how page for it, because I want to keep coming back to it. There were four BloggerCons between 2003 and 2006. So it should be revisited at least three more times. Knock wood, praise Murphy.

  2. We did a blogroll for everyone at the conference. Since it was a very early blogging conference just before blogging boomed, everyone pretty much had a blog. So the blogroll is pretty good. It was one of the things I restored, esp considering this thread.

Dave

@donpark

This comment has been minimized.

donpark commented Oct 18, 2018

Here's another one: what if we moved away from blogs being only about recency?

Blog as a context has a strong timeline element to it but posts displayed in other context don't have to be tied to recency. Say a post tagged with #how_to #mk #fix_stabs could be crawled and collected into a single mechanical keyboard maintenance page. All that really calls for is emergent keywords from communities and tagging posts which bloggers can do and automations can assists with. This can become an important component in the new discovery system like how awesome-blahblah github repos are playing a key role in open source discovery.

@donpark

This comment has been minimized.

donpark commented Oct 18, 2018

Another idea on enhancing blogrolls related to the clustering:

Blogrolls don't have to limited to other bloggers. We can have a section top or top area of each subsection that links to a page dedicated to a curated blog topic, like awesome-blah-blah pages.
This addresses the concept of 'popping out' or 'stepping out' which is likely to be strange to most people, not unlike Edwin Abott got a taste of 3D in the Flatland. In contrast, blogrolls are familiar and fairly well understood as a gateway to other destinations.

@jgmac1106

This comment has been minimized.

jgmac1106 commented Oct 18, 2018

I agree with @kickscondor this is a human issue.

I alo agree with @davewiner this needs to be dead simple. I think we are close based on the chatter I am seeing in the #indieweb -dev channels to being able to have ability to push button subrscribe to people and add to our blog rolls.

WordPress.com pretty much already figured it out.

Until now @kickscondor I think I need to revise all my follow post and have them syndicate to indieweb.xyz using blogroll-topic. We could just start building community lists with one line of html.

(Originally published at: http://jgregorymcverry.com/7956-2/)

@scripting

This comment has been minimized.

Owner

scripting commented Oct 19, 2018

@kickscondor

This comment has been minimized.

kickscondor commented Oct 19, 2018

Say a post tagged with #how_to #mk #fix_stabs could be crawled and collected into a single mechanical keyboard maintenance page. All that really calls for is emergent keywords from communities and tagging posts which bloggers can do and automations can assists with.

This does sound a lot like Indieweb.xyz, as @jgmac1106 mentioned. The concept is simple:

  • Blogger 'tags' their post with a URL: https://indieweb.xyz/en/mk.
  • Their Webmention (pingback) software notifies that URL: "Hey, a post has been made on this tag."
  • Indieweb.xyz checks the page for a valid link - sure enough.
  • The blog post is added to that URL on Indieweb.xyz.

So the emergence should come from blogs clustering around a given URL.

I've been wondering if they could do a similar thing with http://www.adfreeblog.org/ - a 'general' blog community could be established around a simple ideal like that.

Might look like this:

  • Blog links to adfreeblog.org on their home page.
  • Adfreeblog.org notices visitors coming from that page and checks that page for the link and the image.
  • If found, it adds the blog to a directory, using the meta description and keyword tags.

The adfreeblog.org home page then becomes a directory of the community. So, kind of like a webring, but actually organized. With Twitter cards and such floating in the metadata, it is probably much easier to extrapolate a good directory entry.

Spam is an issue with this approach - but it's a start toward discovery. There aren't a whole lot of ways for a blog to jump out from the aether and say, "I'm over here - blogging about keyboards too!" And, in a way, the efforts to squash abuse and harassment are making it more difficult.

This can become an important component in the new discovery system like how awesome-blahblah github repos are playing a key role in open source discovery.

I think it's important to point out, though, that 'awesome' directories are intended to be human-curated, not generative. They feel like a modern incarnation of the old 'expert' pages.

@kickscondor

This comment has been minimized.

kickscondor commented Oct 19, 2018

Ok ok, one other thing that has dawned on me: it's not just the emergent connections between writers that is salient when clustering. It's the connections between readers as well! (This is one thing that Google cannot possibly capture.)

To akaKenSmith's point:

Having found each other, kindred parties need a work space where they can develop shared understandings.

The old Delicious was this kind of workspace for readers - a similar effort can be found in Pinboard.

One interesting thing I like to do with Pinboard is to look up a link - say 'The Zymoglyphic Musem' (results here and then look at the other links by those who found the link. For example, the user PistachioRoux.

All of those links are now related to 'The Zymoglyphic Museum' by virtue of being in the realm of interest of PistachioRoux. YouTube uses these sorts of algorithms to find related videos by matching your realms of interest with someone else's. However, in the process, that person is removed. (Or 'those people', more appropriately.) PistachioRoux is removed.

But perhaps PistachioRoux is the most interesting part of the discovery.

Particularly in a world which is becoming dominated by writers rather than readers - maybe the discovery of valuable readers is part of this.

@donpark

This comment has been minimized.

donpark commented Oct 20, 2018

it's not just the emergent connections between writers that is salient when clustering. It's the connections between readers as well!

Agreed. Reader activity creating connections between posts and blogs. Collecting that data may be tricky but very useful data source.

@davmillar

This comment has been minimized.

davmillar commented Oct 20, 2018

IndieWeb.xyz sounds like an inverse of the paradigm I’m experiencing on Mastodon and the fediverse. A bunch of instances have bots that opt into my instance and then gain the capability to display my hashtagged content in their hashtag streams, rather than the explicit act of sending webmentions to a (really cool and well-intentioned) silo like xyz. Granted at some point there’s an overflow of spam as with anything nice and automated, but it sounds reasonably close to what @donpark mentioned above with mechanical keyboards.

@kickscondor

This comment has been minimized.

kickscondor commented Oct 21, 2018

@davmillar Ok, that's interesting to hear - so it simulates Twitter's hashtag search in a way? Have you find these hashtag streams useful for finding others? (To some degree this works on social networks - like with very specific tags that people use to find each other on Twitter.)

@donpark

This comment has been minimized.

donpark commented Oct 21, 2018

IMO the question of optimal mechanism for collection and distribution of metadata needed to drive discovery is secondary to the question of optimal scale.

I like small towns. They have clear identities for everyone, including travelers and residents, to latch onto which helps with discovery. Communal identities are important to discovery. I dislike sprawling cities without clear divisions between regions. We can build city equivalent online for discovery but that introduces question of costs and incentives. At the town level, a cluster of 100 or less blogs, those issues can be more manageable.

We need tools to help blogs, readers, and curators organize their community so equivalent of towns for blogs can emerge naturally. As to what those towns will look like, they can be a new type of blogs, not unlike the way portfolio blogs are distinct from original blogs, with its own distinct structure, features, and style.

@kickscondor

This comment has been minimized.

kickscondor commented Oct 22, 2018

IMO the question of optimal mechanism for collection and distribution of metadata needed to drive discovery is secondary to the question of optimal scale.

Absolutely agree with this! Although - I think Reddit has proven that communities can become very large. There are ~300k member communities on there that seem to still function well - perhaps I should interview the moderators of some of them, to see if they really are functioning. At any rate, there are many cases of subreddits splintering into new subreddits - so as long as the namespace is big enough to fit everyone, it's fine.

(It's also interesting to think of 'moderation' as a type of human curation - people clearly seem drawn to moderation positions more than they are drawn to curation. They are not that different - both are about shaping the 'filter' of a community.)

I think your comment also applies to Reddit in its conclusion - subreddits are just another type of blog.

@scripting

This comment has been minimized.

Owner

scripting commented Oct 22, 2018

Since it came up in this thread, I wrote up how one-click subscribe worked in Radio UserLand in 2002.

http://scripting.com/2018/10/22/141111.html

Anyone is free to use this idea. No need to re-invent the wheel. ;-)

Dave

@scripting

This comment has been minimized.

Owner

scripting commented Oct 22, 2018

Also, I posted some thoughts about this thread on my blog, which made sense to me because this is about blogging. ;-)

http://scripting.com/2018/10/22.html#a134227

Dave

@davmillar

This comment has been minimized.

davmillar commented Oct 22, 2018

@kickscondor yes, like Twitter hashtag search. The only caveat being that an instance can only display posts it knows about, so bots are sent around to crawl as many profiles as possible and add that knowledge to their home server.

@scripting

This comment has been minimized.

Owner

scripting commented Oct 22, 2018

More thoughts about where we're at.

https://twitter.com/davewiner/status/1054384967508262912

Dave

@jgmac1106

This comment has been minimized.

jgmac1106 commented Oct 22, 2018

@kickscondor

This comment has been minimized.

kickscondor commented Oct 26, 2018

https://seblog.nl/2018/10/22/13/three-things-about-readers
A post about using a query language to suss out emergent connections. The data is drawn from an RSS-type reader. So: who in my feeds is liking what? Perhaps in an effort to wrangle a large number of feeds.

I wonder if there could be value in finding a middleweight query language that can be used by the 'reader' or 'archivist' or whatever to bake new algorithms. So, basically, search terms were the first query language. Then, hashtags have achieved some prominence as search terms that are negotiated between writer and reader - or writers and other writers. (Perhaps there are others between these, of course.)

I guess I'm wondering if we might establish a kind of language for self-building the feed algo:

  • Rank higher if it's nearby.
  • Rank higher if these tags are included.
  • Rank lower if it comes from these domains.
  • Rank higher if it links to these domains.
  • Rank higher if it is generally liked.

It seems like you could then have drop-in algos like ad block software has filter lists that can be subscribed to.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment