This month sees the first generation of Hubs (other than mine) go live, so it’s time to imagine what comes next: AI integration? Filter-bubble Piercers? HubBots? Factcheck-driven credibility scores?
I thought I’d throw together some wireframes to illustrate some ideas for future MyHub.ai development (details follow below):
- Integrating AI systems, particularly an auto-tagger which learns from Hub Editors and Visitors, driving MyHub.ai’s ad-free business model
- Best of the Web: browse the best Web content, curated and tagged by all Hub Editors, from a single place with (faceted) search.
- Integrated Factchecking to spot disinformation and fake news.
- Integrated Machine Translation to help great content find new audiences and support multilingual communities of interest.
- Integrated text summary tools to help Users get more from content.
- Filter Bubble Analyses to help Users break out of their echo chambers.
- Text network visualisations to help users explore links between the content they discover on MyHub.ai
- Hub Discovery: find Hubs of interest and Follow them, so you can reBlog content from other Hubs onto yours, and browse content shared by them in your Private Reading Queue.
- feed all the content shared by newsletters, Twitter accounts and Lists into your Private Reading Queue.
- adding a newsletter and chatbot to each Hub.
Everyone can vote for ideas, while Editors can also suggest their own.
AI integration to support an ad-free business model
These first Hubs are, like mine, without AI support, so their Editors will tag their content manually. The plan has always been to add:
- an AI engine to auto-tag content, to be checked/corrected by Editors
- a training module to learn from the Editors’ corrections — and their visitors’ behaviour — to train that AI in a virtuous cycle.
an advertising-free haven from surveillance capitalism
While ideas for integrating other AI tools can be found below, the auto-tagger comes first as it will provide the revenue MyHub.ai needs to create an advertising-free haven from surveillance capitalism. Assuming there’s some surplus, how should we use that revenue?
Discover better content
Browse the Best of the Web at MyHub.ai/@all
Well of course you can Google the entire web, if you like ads masquerading as analysis, and don’t mind ignoring everything published in other languages.
But if we put all the content curated by all Hubs together in one place, users could browse and search this tiny fraction of high quality content which Hub Editors have chosen to curate, for themselves and you, on their Hubs. Whatever the language.
The interface, above, is pretty similar to today’s individual Hub design, except:
- visitors can browse and search all content created & curated by all Hub Editors, filtering by tag, popularity, language and credibility (next point)
- cards feature content credibility scores aggregated from factchecking services, something we hope to roll out to all Hubs — see Integrate factchecking to spot disinformation, below
- cards show how many times the content has been curated (clicking “Shared N times” takes you to the full list of Hubs, aiding Hub Discovery)
- logged-in Editors can ReBlog any card they like onto their own Hub — see Follow and reBlog other Hubs, below
- Logged-in Editors can also access (top right) other features set out later: auto-translation and -summaries of the content they’re looking at, their Private Reading Queue, Filter Bubble Checkup, etc.
Follow & reBlog other Hubs (& vice versa)
Follow a Hub to get a feed of high-quality content piped into your Private Reading Queue, and ReBlog individual resources straight to your Hub.
You should be able to:
- Follow an entire Hub’s output, or just some of its Categories and/or tags. All matching Resources will be auto-added to your Private Reading Queue (below) for you to read at your leisure
- reBlog a Resource from someone else’s Hub to yours with a click, or send it to your Queue to re-curate it yourself (wireframe, left)
- be notified when someone follows or reBlogs you.
Together, these features could create communities of interest: networks of Hubs creating, curating and exchanging content of shared interest.
create communities of Hubs creating, curating and exchanging content of shared interest
One downside could be to reinforce users’ echo chambers — hence Pop Your Filter Bubble!, below.
Search for Hubs matching your interests
You cannot Follow a Hub if you don’t know it exists.
While you could use MyHub.ai/@all (above) to discover Hubs of interest, I’d like to make it easier.
We could build a search interface, but I’m currently leaning towards a Recommendation engine, as this would recommend two Hubs to each other if they use similar tags, hopefully linking people around shared interests. Such an Engine would appear alongside your Reading Queue (next).
Feed more content into your Reading Queue
Your Queue contains resources you haven’t read yet but don’t want to miss from Hubs you Follow, enewsletters, Twitter accounts and Lists.
The full wireframe of an Editor’s Reading Queue, left, looks very similar to the Editor’s Public Hub, except that the filters include Sources: Hubs, Twitter accounts, newsletters, etc.
As the Editor hasn’t read these resources yet, the card’s Summaries were auto-filled by the Hub, and their tags auto-suggested by the AI engine.
All the other services are there, too: automatic content credibility, machine translation, auto-summary, etc.
This can boost your content productivity in many ways:
- If you spot an interesting resource online but don’t have time to curate it, one-click add it to your Queue and come back to it later.
- Follow a Hub: some or all of its content will appear in your Queue
- Turn on Twitter-Likes: every time you Like a Tweet with a link, that link ends up in your Queue, ready to curate
- Follow someone on Twitter? Have every link they Tweet end up in your Queue. Same for Twitter Lists.
- And don’t miss resources send via your favourite enewsletters — pipe them straight to your Queue.
Get more out of better content
Several services could help both Editors and their visitors extract more value out of the content organised on MyHub.ai.
Integrate factchecking to spot disinformation
The above wireframes show cards with credibility scores, with visitors to MyHub.ai/@all (above) able to filter all Hubbed content by credibility.
Clicking a score ‘flips’ the card over (a design trick beyond my wireframing software, left), showing more details on the card’s reverse side.
I’m currently no further than toying with a couple of designs, as this whole idea is speculative.
It relies on Factchecking services (the above examples use PolitiFact, Washington Post and NewsGuard) providing APIs, allowing MyHub to check if a resource has been factchecked or if a domain has a credibility score. Scores from multiple fact-checkers would be processed into the composite score displayed on the front of the card, and listed briefly on its reverse.
Clicking ‘Explore & Discuss’ on the card’s reverse, finally, could open a dedicated page where more detail can be provided, plus a conversation space where scores can be contrasted, their veracity discussed, etc.
take factchecking debates off social platforms, where they amplify the spread of disinformation
This echoes an idea I explored a couple of years ago, and could allow us to take factchecking debates off the social platforms, where such arguments amplify the spread of disinformation:
Building Credibility Indexes from Fact-Checking to #TackleFakeNews
Fake News is suddenly on the boil in the Brussels Bubble. Here are some ideas I brainstormed earlier this week…
Integrated machine translation
Browsing many Hubs means browsing content in many languages.
Whenever you find yourself looking at content in languages you don’t understand, login and click AutoTranslate (top right in the wireframe, left).
It will only translate the cards’ Titles and Summaries, but that should be enough to tell you whether it’s worth visiting the original resource and getting a full machine translation.
Pop your Filter Bubble!
The jury’s still out on whether social algorithms create filter bubbles, or help users discover content from outside their echo chamber.
Either way, it could be incredibly interesting to access a data visualisation analysing the content you curate and create on your Hub for various forms of bias.
Hence the “My Filter Bubble” option for all logged-in users.
Even better: your Hub could suggest high-quality content from outside your echo chamber, helping you widen your perspective and challenge yourself without falling foul of toxic content.
suggest high-quality content from outside your echo chamber, helping you widen your perspective
Generate automatic summaries of selected resources
Once you’ve narrowed your search to a small set of highly relevant, high-quality content, login and click AutoSummarise. Because who can read everything out there, right?
Although this could be cool, I have misgivings: it could mislead users, either through simple inaccuracy or by giving them the sense that they understand more than they do (cf How search engines make us feel smarter than we really are). Nothing beats reading properly. But if the tech works, and if Users want it, and if resources allow it, it’s in.
Discover connections across your content
Take a look at this video from @noduslabs. Now imagine how integrating something like Infranodus into MyHub.ai would help Editors and their visitors find patterns and interesting relationships between any collection of Hubbed content, either within one Hub, or across all of them.
A newsletter and chatbot for every Hub
A Hub should provide its Editor and visitors with an enewsletter, fed with recently Hubbed content.
The obvious approach is to allow Editors to connect their Hubs to their existing enewsletter platform. That way, your Hub could auto-generate a draft enewsletter edition for you based on your recently Hubbed content.
The downside is the cost of creating and maintaining multiple integrations to multiple systems, so perhaps one day we’ll build our own enewsletter system. I currently doubt it, however, as it implies entering a highly competitive market against extremely sophisticated enewsletter platforms, including some (eg Substack) which are also exploring new business models for content.
Remember when chatbots were going to take over the internet? ;)
So do I — in fact, I piloted a Hub-driven chatbot in 2016, and wrote three articles for ChatBot Magazine as a result. The ‘curatorbot’ I imagined would allow users to interrogate your Hub using chat platforms such as Messenger — i.e., people enter keywords; the bot returns relevant stuff you Think, Like and Do; users refine those results using additional tags to find what they want.
My Chatfuel-driven pilot bot was extremely limited: users could only enter one keyword, and so could not refine the results further. But a full MyHub.ai ChatBot engine could be much more powerful, using AI to understand the user’s question, and providing them with faceted search, something I wrote about in my second post:
“such a bot would walk users through refining their search, quickly drilling through mountains of content to find the diamond they need using no more than their thumb as they wait for a bus” — Boosting CuratorBots with Faceted Search.
While this may be a major technical effort, it’d be pretty cool. However, I think we’ll only go for it if chatbots start getting real traction.