This tool is inspired by Kagi Small Web (recent thread: https://news.ycombinator.com/item?id=47410542). A common concern raised here is that Kagi Small Web currently accepts only blogs, comics and YouTube channels. It does not accept arbitrary small websites. That limitation motivated me to build Wander.
Wander is fully decentralised. Anyone can host it on their own website. It consists of just two files: an `index.html` for the Wander console and a `wander.js` where you link to other Wander consoles. It is a bit like a webring, but more flexible. Each console can link to any number of sites and other consoles.
There is no server-side code, no database, nothing to install. If you have a website, you can set it up by uploading just two files. In fact, you can host it on GitHub Pages or Codeberg Pages too.
If you like the idea, please join the network. I would love to see it grow.
To me the way Kagi curates its small web directory always feel contradictory to the spirit of independent web publishing because it seems like you have to submit entries through GitHub — a highly centralized platform owned by one of the largest tech companies.
kudos to you for figuring out a decentralized solution.
This is so delightful! I'll be deploying this and sharing a link on the issue.
One issue I see: If I get you to include a link to my console but I don't link to any others, I can trap wanderers within my recommendations until they refresh.
If that's not desirable, it could be avoided by having the client keep a running list of all the consoles it has discovered this session and choosing from that list at random.
> If that's not desirable, it could be avoided by having the client keep a running list of all the consoles it has discovered this session and choosing from that list at random.
It would be nice if it wasn’t so framed in. How about a simpler “wander” overlay button that stays fixed (a bit like google recaptcha in the bottom corner) and pops up the console with links?
Glad you like it. Yes, you are right. This is something I realised too initially as a natural consequence of being the only participant in the console network at the beginning. Keeping a list of discovered consoles is exactly what I was thinking too. I built this tool rather quickly as a proof-of-concept while taking a break from another activity, so I couldn't quite find the time to implement this solution. But I might implement it in the next update. Thank you for taking a close look at this project!
I hope platforms like these find a way to attract people outside tech circles. I looked at around a dozen recommended sites and only two of them isn't the personal website of someone who works in tech and writes mostly about tech, which gets boring rather quickly.
There is a world of non-tech bloggers writing stuffs about history, culture and nature who would likely never learn about this project simply because they are not in the right social spaces. I hope there is a way to have them in the ecosystem too.
It doesn't seem very hard to implement in your own site, so it might gain some traction? It's not super-complex, I understand it is some sort of interconnected spin on webrings, which are still somewhat popular among small websites.
If anybody wants to find truly random small websites, I recommend using Wiby (search engine). It has some neat stuff.
> It doesn't seem very hard to implement in your own site
Depends on how technically sophisticated the author is. Many of the blogs I was thinking about were not written by people familiar with web stuffs. They are hosted on managed hosting services like wordpress.com and blogspot, or on hosting providers with streamlined services that require no technical skills to use. Setting up this tool may very well be beyond what the authors are comfortable with or capable of.
Very cool. Reminds me of stumbleupon, which I lost many hours to back in the day.
Curated discovery is one of biggest gripes with modern platforms like youtube - discovering something truly new and outside of your normal interests is really difficult, and the same goes for the web. If you have a topic you want to explore it's fine, but finding random things you'd never have thought of yourself is much harder.
I've thought of a service that scans new websites and GitHub repos and looks for things that don't look like anything else (using something like hdbscan for outlier detection), and creates a feed for people to follow.
I love this as a concept. The wander button is great, but it still needs some curating to decide what pages you like, and getting to the actual content. I guess I'd like to know the workflow moving forward? Just re-download the repo every couple weeks, and diff to see what new sites are on the list?
Thank you for taking a look at this project. I'm glad you like the concept. I am not sure I have understood your question accurately, but let me attempt a response anyway. If I get it wrong, please feel free to correct me or ask me again.
There is no need to re-download https://codeberg.org/susam/wander every few weeks. The setup is a one-time activity. From that repository, you copy exactly two files (index.html and wander.js) and place them on your web server, preferably within a /wander/ directory. After that, you only maintain the wander.js file.
You curate your own links and choose which other Wander consoles to link to as neighbours. The contents of wander.js are entirely yours to define. There is no need to diff or compare it with the version in the repository.
In fact, if you do not care about updating or curating links often, you can leave both files untouched indefinitely. The only downside is that some links may eventually succumb to link rot, which could affect the wandering experience. So it may help to review your links occasionally and remove dead ones, but beyond that no ongoing maintenance is required.
I get that, but right now, if you traverse each "console", you end up with a list of 28 trusted "small web" links. The project grows in value if that list gets bigger over time either by you personally adding nodes or the community adding nodes. I don't really have a way of knowing if you are intending to add more links to your console (thus growing the project) or this is a one and done type of system.
> I get that, but right now, if you traverse each "console", you end up with a list of 28 trusted "small web" links. The project grows in value if that list gets bigger over time either by you personally adding nodes or the community adding nodes.
Yes, all of this makes sense.
> I don't really have a way of knowing if you are intending to add more links to your console (thus growing the project) or this is a one and done type of system.
I personally do not plan to add too many page links to my console. However, I will add more console links, which has the effect of expanding my console neighbourhood and thereby increasing the pool of recommendations.
That said, I am not sure why it matters whether I add more links to my console specifically. In my opinion, any single Wander instance should not matter much on its own. What matters more is whether the network as a whole grows, that is, more consoles being set up and more of them linking to each other.
One of my design goals has been to avoid giving any particular console a special status. All consoles are equal participants in the network from a technical perspective. You should be able to pick any console from the network, perhaps one belonging to your favourite blogger, perhaps even your own and explore the neighbourhood from there. Yes, the neighbourhood would look different from each console but that's pretty much the point of this project. As long as the overall graph of consoles is connected, you could in theory reach any community recommendation from any starting point. Even if the graph is not fully connected, I do not see that as a significant issue. It just reflects how connections tend to form in a decentralised system. Please let me know if you think I have missed your point again.
Before webrings and the very first directories and search engines, the tools for exploring the web were memory, bookmarks and the links sections of web sites.
What you are describing sounds a bit like a blogroll, which many of us do indeed maintain. Mine is here, by the way: https://susam.net/roll.html
However, Wander is meant to be a bit like StumbleUpon, but without requiring a centralised service that everyone must go through. One limitation of a blogroll is that it does not provide a consistent way to discover recommendations recursively. For example, I might visit your website A, which recommends website B. I might then visit B, but B may not have any recommendations at all.
Every Wander instance, on the other hand, has a defined list of recommendations. It also links to the /wander pages of its neighbouring sites. If you visit the /wander page of website A, the tool can discover its neighbours (B, C, etc.), then the neighbours of those neighbours and so on. It can fetch recommended links from them and present the links within the same console.
Additionally, the tool provides a way to leave the current console and move to a neighbour's console if the visitor wants to continue browsing from there.
OK I think I get it. It's fundamentally more about the streamlined experience. I played a bit with it and it _felt_ special. So my initial reaction might have been reductive.
Because the discovery is transitive. When you wander, it fetches another console's wander.js and picks from their pages, so you're not just exploring one person's list, you're hopping across a graph of curated lists. A static link page can't necessarily do that.
Cloudhiker.net has been doing this for a while too. Great to see more grassroot-ish attempt at expanding the web (or i guess more accurately returning it to its purpose)
Love this, but I need it to allow me to break the frame. I found some neat sites, but I need a button to open them in a new tab, frame-free, if I want to.
Thank you for taking a look at this project. I'm glad you liked it.
You make a fair point. I put this together in about 1.5 hours in the early hours of the morning while taking a break from other academic work, so it is not as polished as it could be. Adding an 'Open' button to launch a link in a new tab sounds like a useful feature. I will consider this for the next update. Thank you for the feedback!
I love the idea. One small thing - Ran into a problem almost immediately
--
Nightly Can’t Open This Page
To protect your security, drkhsh.at will not allow Nightly to display the page if another site has embedded it. To see this page, you need to open it in a new window.
If you are on a computer, you might be able to see the previous links that were loaded in your web browser's Developer Tools > Console. The Wander console tool writes logs there to describe what it is doing internally.
The number of URLs grow as more people add this tool to their website and add each other as neighbours. The tool is capable of discovering neighbouring consoles and showing URLs from there.
I kind of want webrings but with federated OIDC. I.e. you can create an account at any of the sites and re-use the login on the others to leave comments/add content. This has to exist in some neat package right?
There is indeed a lot of similarity with webrings and StumbleUpon but there are some differences as well. I discuss this in more detail in the project README here: https://codeberg.org/susam/wander#faq
It would be cool if the javascript were re-written a bit to only use standard long existing javascript features so one didn't have to run a modern corporate browser to use the "wander" functionality. It did not work in my browser even with JS enabled.
Sorry to know it did not work in your browser and thank you for reporting the problem here. If someone is able to reproduce it and share details either here or at <https://codeberg.org/susam/wander/issues>, especially which JavaScript features are not working, I would be happy to update the code.
I am aware that I have used a number of relatively modern features. For example, I suspect `for ... of` might be one, and perhaps `Element.append()` (instead of `Node.appendChild()`) could also be an issue. Unfortunately, I do not have old browsers to test this reliably, so any help in identifying the exact constructs or functions causing problems would be very appreciated.
I don't know how useful this is, but I am getting
tired of Google and co ruining the world wide web
how it once was. Something has to be done. I have
no idea whether this here can be of help or not
but the more people think about this, the better.
Otherwise the quality will continue to degrade.
That was me sorry! I was quickly fixing it up while a meeting was starting, and I hurriedly re-typed it rather than copying+pasting like I should always do with project names.
This tool is inspired by Kagi Small Web (recent thread: https://news.ycombinator.com/item?id=47410542). A common concern raised here is that Kagi Small Web currently accepts only blogs, comics and YouTube channels. It does not accept arbitrary small websites. That limitation motivated me to build Wander.
Wander is fully decentralised. Anyone can host it on their own website. It consists of just two files: an `index.html` for the Wander console and a `wander.js` where you link to other Wander consoles. It is a bit like a webring, but more flexible. Each console can link to any number of sites and other consoles.
There is no server-side code, no database, nothing to install. If you have a website, you can set it up by uploading just two files. In fact, you can host it on GitHub Pages or Codeberg Pages too.
If you like the idea, please join the network. I would love to see it grow.
More details about how it works and how to set it up here: https://codeberg.org/susam/wander#readme
To me the way Kagi curates its small web directory always feel contradictory to the spirit of independent web publishing because it seems like you have to submit entries through GitHub — a highly centralized platform owned by one of the largest tech companies.
kudos to you for figuring out a decentralized solution.
One issue I see: If I get you to include a link to my console but I don't link to any others, I can trap wanderers within my recommendations until they refresh.
If that's not desirable, it could be avoided by having the client keep a running list of all the consoles it has discovered this session and choosing from that list at random.
Implemented in <https://codeberg.org/susam/wander/commit/f4d95fa>. Thanks again for the discussion!
There is a world of non-tech bloggers writing stuffs about history, culture and nature who would likely never learn about this project simply because they are not in the right social spaces. I hope there is a way to have them in the ecosystem too.
If anybody wants to find truly random small websites, I recommend using Wiby (search engine). It has some neat stuff.
It has indexed lots of different websites categories
Depends on how technically sophisticated the author is. Many of the blogs I was thinking about were not written by people familiar with web stuffs. They are hosted on managed hosting services like wordpress.com and blogspot, or on hosting providers with streamlined services that require no technical skills to use. Setting up this tool may very well be beyond what the authors are comfortable with or capable of.
Curated discovery is one of biggest gripes with modern platforms like youtube - discovering something truly new and outside of your normal interests is really difficult, and the same goes for the web. If you have a topic you want to explore it's fine, but finding random things you'd never have thought of yourself is much harder.
There is no need to re-download https://codeberg.org/susam/wander every few weeks. The setup is a one-time activity. From that repository, you copy exactly two files (index.html and wander.js) and place them on your web server, preferably within a /wander/ directory. After that, you only maintain the wander.js file.
You curate your own links and choose which other Wander consoles to link to as neighbours. The contents of wander.js are entirely yours to define. There is no need to diff or compare it with the version in the repository.
In fact, if you do not care about updating or curating links often, you can leave both files untouched indefinitely. The only downside is that some links may eventually succumb to link rot, which could affect the wandering experience. So it may help to review your links occasionally and remove dead ones, but beyond that no ongoing maintenance is required.
Yes, all of this makes sense.
> I don't really have a way of knowing if you are intending to add more links to your console (thus growing the project) or this is a one and done type of system.
I personally do not plan to add too many page links to my console. However, I will add more console links, which has the effect of expanding my console neighbourhood and thereby increasing the pool of recommendations.
That said, I am not sure why it matters whether I add more links to my console specifically. In my opinion, any single Wander instance should not matter much on its own. What matters more is whether the network as a whole grows, that is, more consoles being set up and more of them linking to each other.
One of my design goals has been to avoid giving any particular console a special status. All consoles are equal participants in the network from a technical perspective. You should be able to pick any console from the network, perhaps one belonging to your favourite blogger, perhaps even your own and explore the neighbourhood from there. Yes, the neighbourhood would look different from each console but that's pretty much the point of this project. As long as the overall graph of consoles is connected, you could in theory reach any community recommendation from any starting point. Even if the graph is not fully connected, I do not see that as a significant issue. It just reflects how connections tend to form in a decentralised system. Please let me know if you think I have missed your point again.
However, Wander is meant to be a bit like StumbleUpon, but without requiring a centralised service that everyone must go through. One limitation of a blogroll is that it does not provide a consistent way to discover recommendations recursively. For example, I might visit your website A, which recommends website B. I might then visit B, but B may not have any recommendations at all.
Every Wander instance, on the other hand, has a defined list of recommendations. It also links to the /wander pages of its neighbouring sites. If you visit the /wander page of website A, the tool can discover its neighbours (B, C, etc.), then the neighbours of those neighbours and so on. It can fetch recommended links from them and present the links within the same console.
Additionally, the tool provides a way to leave the current console and move to a neighbour's console if the visitor wants to continue browsing from there.
You make a fair point. I put this together in about 1.5 hours in the early hours of the morning while taking a break from other academic work, so it is not as polished as it could be. Adding an 'Open' button to launch a link in a new tab sounds like a useful feature. I will consider this for the next update. Thank you for the feedback!
Throwback to the StumbleUpon days.
--
Nightly Can’t Open This Page
To protect your security, drkhsh.at will not allow Nightly to display the page if another site has embedded it. To see this page, you need to open it in a new window.
See also a similar discussion here: https://codeberg.org/susam/wander/issues/3
There is also a note in the project README requesting console owners to be mindful of this while adding links: https://codeberg.org/susam/wander#caution
The number of URLs grow as more people add this tool to their website and add each other as neighbours. The tool is capable of discovering neighbouring consoles and showing URLs from there.
(Not a diss. Webrings were cool.)
Would it be possible to create a list of sites with a Wander?
I used to spend hoooooooours wandering there.
It is a project I keep postponing lol
I am aware that I have used a number of relatively modern features. For example, I suspect `for ... of` might be one, and perhaps `Element.append()` (instead of `Node.appendChild()`) could also be an issue. Unfortunately, I do not have old browsers to test this reliably, so any help in identifying the exact constructs or functions causing problems would be very appreciated.
Never used it personally, but might get some mileage out of the free plan before their time-based usage expires.
https://www.browserstack.com/docs/automate-self-hosted/getti...
I think we need things like this. To be reintroduced over and over.