Stephen Downes

Knowledge, Learning, Community
Design and Development of a Personal Learning Environment


Edited from Google Recorder transcription.

This is 'Design and Development of a Personal Learning Environment'. It actually slots in as presentation two of four that I'm doing in a couple weeks span. The next two are at Otessa, and they will be on Wednesday if you're still kicking around town. And they cover the idea of the alternative approach to online learning I and other colleagues have been developing over the years.

This is not a presentation on AI, you can thank me later, although it does have AI elements in it. And it does, as the other presentations show, speak to how we might want to think about approaching AI in learning.

But, as they say: Begin with the demo. So, here's the demo.

So, this is a thing that I built. It's a personal learning environment. I'm actually pretty pleased to have done this because there hasn't actually been, despite an awful lot of talk in our field, a real live personal learning environment built that people can use, but now there is one.

It's not perfect. I wrote it, so it will have problems and mistakes. It's a proof of concept, nothing more, and the idea is to use this as a basis for thinking about how we can think differently about how we do learning online.

I'm going to log in. I don't have to log in. But I need the place to access my accounts, which we'll see in a bit. I could just type them all in by hand, but that would get tiring very quickly. I could also save them on a key that I could just load into my computer, but I haven't written that software yet. So, the main thing is, I have to have access to my accounts.

So I'm logging in, not because I'm signing into a platform or something like that. This belongs to me. But because I need access to my data, and I protect my data from the rest of the world, hence the need for a sign-in.

So you can see on the left hand side, a bunch of buttons. Each button corresponds to a different account. I'll select 'Downes'. That connects to my Mastodon account at mastodon.social. And I can read the posts that I'm following. It's all the people that I'm following on Mastodon, all their posts.

Now, what you don't see because it's behind the scenes is that the people who I follow that write in Slovakian (and I actually do follow people who write Slovakian) their content has been automatically translated for me. It just happens behind the scenes. I don't even worry about.

Actually, let me be honest. I do worry about it because I used my Google Cloud accounts to do a translation. And if you don't keep an eye on it, those costs can add up. I accidentally spent about three hundred dollars in November, testing my system. I really need an on/off button. I've also been thinking about ways of optimizing the translation, so it doesn't cost me an arm and a leg. Part of the future of AI is we'll have to pay for it.

Okay, well, you're thinking, well, a big deal. I've got my home feed. I could use Mastodon for that. That's true. I've got the usual Mastodon functions. I can reply, I can boost. I can star it. I can bookmark it. One of the nice things about the bookmarks is. I can see my bookmarks here as well. There they are.

But as well I have this: I can feed it over. There's a little arrow button here. It's kind of hard to see, and now it's in my editing window. So, it's not beautiful. The editing window this is just a plain text window, but as you can see, it's taking the content of the posts that I'm commenting on and it puts it into the editing window, and now I can comment on it.

So here's a comment. You also note at the bottom. It's keeping track of the reference information for that now.

Okay, that's great. Let's go over to Blue Sky now. I also have an account on Blue Sky. It's going to take a second, maybe? Oh, I think it's here already. Uh, timeline. These are posts on Blue Sky. Um, again, reading them in the same environment. Same format. The icons are a bit different under the post because blue sky has different functions than Mastodon, but I can still do the same thing here.

I can now add the blue sky Post in here. And so now I'm able to take the two posts, one from Mastodon, one from Blue Sky, put them into the same environment, and now work in this environment, and then once I'm done working in this, I can post it and I can post it either back to my original Mastodon post, or I can publish it as a blog post on half an hour, which is my blog, my blogger website.

I can select one or more, I click 'publish'. It's going to do its thing. (Blogger refuses to publish it). Ah, you stupid thing. Well, that that's Google for you. That's publishing on my other blog. And that should have published. Let's take a look. Told you there would be bugs and issues there. It is. There's the the content I actually published.

I've taken my content, published it, or edited from multiple sources, edited it together and published it on to my blog. Now, I can also publish back to social media. In theory, although this isn't working yet, if I select both Mastodon to the short form and blogger, which is a long form, then it'll put the long version in Blogger and put a link to the long version in Mastodon so that people can see it, mark it up, and then go to where I spent 500 words or a thousand words or whatever.

More features to talk about. Let's go back to Blue Sky. So, let's find here we have a thread. So, here's a thread. Here's a thread of things that people actually said something in. And I click this button. This one doesn't cost me money happily. But it takes a little bit of time, but what's going to pop up there there it is? The AI generated summary of the thread.

Okay, and so we see the the automated summary. Again, I'm not just taking some AI stuff. I'm publishing it, right? I'm taking some AI stuff and using it to get stuff into my editing window that I can then work with in some way and then publish it out there on the social network.

A few more things. I can load the editor, various types of editor. For example, I can load from the file system so I can load any document locally. I can also save any document locally, so I've got read/write on my local computer. I can also load the template.

Or more. Let's generate a new template. Hey, it's not going to generate a new template. Okay, one not working. What that does is it gives me a selection of templates to use. I select the template, and it uses the AI to build a template of whatever it is I wanted. So, I have a selection like business letter, case, study report, etc., and I get that template obviously. That's a very simple example. I could use any number of templates that I want.

I'm also connected to an Etherpad server. Etherpad, you may know, is a collaborative editing tool where multiple people can edit together in the same environment. So, I've got some existing documents already. The idea here is now I can share with other people the editing of this document.  I can still use these (read) functions. I'll use my Mastodon Co-social accountant this time and load content into the Etherpad.

So now, I'm taking content that I loaded putting it in the ether pad. And now, I can work with it. There's also a chat. Now, I'm the only person using the system right now. If there were other people logged on and using this, we could have a chat. If we were chatting together, or if we were both in the chat, and I wanted to share my Etherpad with them, I would click on the share button. It would send a button to them to click on so that they can open the Etherpad in their personal learning environment and then edit.

I've taken 12 of my 25 minutes for the demo. But I want you to think about the sorts of possibilities here. For one thing, all of those accounts are my accounts. They're my personal accounts. I can get them from anywhere. I haven't listed all the different types of accounts that I can access, but you know, I can use search, I can use Etherpad, Mastodon, etc., pretty much any service that provides content, and I'm including an LMS in here, although I haven't actually hooked it up to one yet (it's in in the works).

Get the content from that, feed it into my reading area, and then perhaps even select some content. And feed the selected content into the writing area. In the writing area I've got a simple text editor,  I've got a collaborative text editor. I have automatically generated templates that aren't working for no particular reason.

This will work in etherpad? Generate new template. Oh, here we go. Business letter case study. I'll choose the output format there we go. That's what happened. So, I accidentally chose text, so it's going to generate my case study in a text editing form.

We were talking about h5p. H5p would work brilliantly in that editing window. Just load your HP h5p template into it and then feed in from the stuff that you're reading again. It just makes it easier to work with. The key thing, and, and what's important here, is the AI isn't doing the work you're doing. The work on choosing that, or too well, choose a text editor. Here's my case study. This is generated by AI. Any AI prompt that you can think of can create the basis for something that I'm working on.

I made some promises in the abstract which I'll literally wave my hand at.

All right, what even is a personal learning environment? Because maybe you weren't around 20 years ago when people first started talking about them or ten years ago, when people were actually doing projects on them and failing. I include myself in that group. The idea of a personal learning environment is that it's an environment for you to learn in, not an environment hosted by some educational institution, but an environment owned completely and hosted by you or by a host that you trust.

You can put it wherever you want. The way I've designed this, it's in pure JavaScript, which means that it can go anywhere. In theory, you can run it off your own computer. You can certainly run it off of any web service provider that you wanted. I could have run it off of her computer. No problem. Just go to it and it would work. So you're not locked into a specific platform.

The idea of a PLE is you choose the resources, maybe with some help. You're not on your own. You organize content for yourself. Pick what you're interested in. And you can work with other people collaboratively in any through any system that allows you to collaborate, to work together, to create any type of content. I showed you text based content, but clearly we're not limited to that right. There are collaborative visual editor screens, for example.

Here's some of the famous diagrams (I have no problem with PowerPoint. I have problem with Microsoft's constant prompts for me to try AI. Uh, yeah, it's just trying to help. That's the difference between a PLE and an LMS. An LMS is always just trying to help, but with a PLE all pick when I want help, thank you).

So, here's the famous Scott Wilson diagram of a personal learning environment. The personal learning environment is at the center. That's the thing that I built, and then there are various services that it connects to services that read services that process data for you. Services that allow you to get together with other people and do stuff services that allow you to publish and write.

Why a PLE? Because it changes the dynamics of learning from a focus on the content, from a focus on the educational provider, to a focus on the learner and what it is that the learner is trying to do. Typically any given user of a PLE will have our own projects, their own interests, in mind. And the PLE allows them to draw from those resources and those services that exactly match the needs that they have to accomplish those purposes. So it's purpose driven, individual driven, and individually owned.

So here (referencing disgram) is the LMS, and the institution is at the center. Everything's around the the institution, around the course, the students are all supplicants. I mean, they're they're all connected to the LMS. What's really significant? We've seen this over and over - I don't know if this is a word, 'silofication'. Because of the LMS, there's one LMS here or one institution, there's another one here for another institution, never the twain shall meet.

If you're on the Western University LMS, you don't get to go and have a conversation with the McGill students. On their own personal learning environment, though, it doesn't matter, you're here, you can connect to McGill, Western, Face, Google Docs, iBook, whatever (services were different back then).

What's new? What's changed between now and say 2010? The social media landscape has finally begun to kick the habit of large centralized platforms on Twitter and Facebook and Linkedin. You need to evolve toward federated services where there is not one centralized platform, but multiple interconnected platforms (if you're not clear on that concept, I have a talk coming up at OTESSA,  something like 'Federated learning in the blogosphere', where I explain all of that. I have to bracket that because I have three minutes left. 

So? I've been working on this concept for 20 years. What a thing to spend a life on, huh? There's all kinds of business issues with the ple that really makes it hard to get off the ground. The same issues face the fediverse generally. Institutions aren't interested in licensing personal software that you can use anywhere, and if they're not interested in licensing it, they're not interested in paying for it.

There's no incentive for institutions to let people hop to other institutions (I remember trying to launch a project in the early days of RSS, getting all of the engineering schools in Alberta together to provide resources by RSS for the Association of Petroleum Engineers and Geologists of Alberta; were they interested? No, they didn't want to share their content. Same logic still exists today, which is so silly, if you ask me).

A lot of the time, access controls and restrictions are enforced by content providers. It's not as bad now as it used to be. It used to be horrible, now it's just bad.

Institutions love to control the learning environment. I would (I don't have the diagram there; pretend the diagram is there) I would add xAPI to this so that you can at least track learning activities and then feed that back to the institution. You don't have to be in the institution's space.

There's also a strong culture of traditional pedagogical models. This learning environment, the PLE,  isn't a show and tell kind of system. It's a "learner goes in and does stuff" kind of system. Now, somebody talking about MOOCs earlier this morning talked about the average person who succeeds at Mooc is a well-educated white professional.

That's quite true, because they're the ones who managed to get the skills to be, you know, independent, self-motivating, and thus the people who learn best in MOOCs. But that's not an indictment of moocs. That's an indictment of society. Why is only that demographic able to master those skills? Everybody should have the opportunity to master those skills!

Different attempts to create personal learning environments are difficult because the concept is difficult if it's not something that you're used to, if it's not something that you've had practice with. In the past, I talk about email sometimes in this context, and people adapted to that. It's not in simple thing, email. It's actually pretty complex, especially behind the scenes. People learn. Yeah, you choose your email provider. Yeah, you run your own email. You, you are the one responsible for sending emails. And if you want to receive emails, they don't come to you by magic, except spam. You have to in some way, initiate that.

Development challenges were difficult as well. A single user website. There's no real model for that beyond blogs, and blogs all shifted to platforms like LMSs and like social media. The cloud 10 or 20 years ago really wasn't useful. This (the PLE) makes use of the cloud that's a fair bit (The Norvig and Thrun MOOC used the cloud quite a bit; they also used automated grading).

There's platform lockdown.. You may have noticed I haven't connected Twitter, Facebook, and Linkedin in my personal learning environment. You wonder, why? Well, it's because they won't let me. They won't let me get content from their services and link it to my own personal learning environment. You must go to them, view their news feed and read their advertising if you want to use their service.

And then everything involved is complex, put together. Overall, this is the flow - aggregate, remix, repurpose, feed forward. Those of you who are familiar with my history know that this is the basic model that we use to develop MOOCs. This was the learning process. Now, this is the tool that people can use to support them in that learning process.

Sorry, I'm two minutes late, but I think I did okay, yeah.


Stephen Downes Stephen Downes, Casselman, Canada
stephen@downes.ca

Copyright 2025
Last Updated: Aug 28, 2025 9:15 p.m.

Canadian Flag Creative Commons License.