Hey.
For the past 2 years, I have been working on LearnObit (https://learnobit.com).
When I first started to learn coding, I used Anki and the old tried-and-tested Spaced Repetition algorithm, but it wasn't suitable for more complex, general topics like Mathematics or Physics.
Because of that, Spaced Repetition has always been widely overlooked despite its proven effectiveness in learning.
I wanted to unlock this effectiveness for a wider audience by building LearnObit,
a tool that combines tree-structure note-taking tools with Spaced Repetition so that you can apply it to more than just simple fact pairs.
And now it even has the feature to export data to Anki, so you can leave with your data when you decide not to use it!
(Of course I don't want anyone to use it, but I couldn't help it, because people wouldn't start using it without this in the first place...)
Anyway, I'll be here anytime if you have any questions for me. Thanks for your time.
Just a general recommendation: for math/physics/CS it's far more useful to work through problems than memorize facts. I'll use linear algebra as an example since that shows up often in your cards.
Say you want to learn about eigenvectors and eigenvalues. Instead of writing the definition, you might have a card which asks the following:
"Create a matrix whose eigenvectors or eigenvalues are always equal."
This will force you to do two things. First, you will learn from creating the card since you're forced to synthesize the information rather than copy. Second, your recall will force you to work through this newly created problem.
Another general strategy is to just pick a problem from a book that uses the concept. If you ever get the problem wrong, you oblige yourself to solving an additional related problem from the same book (i.e. find the basis of one of the following matrices).
I guarantee you will learn 10x more effectively.
Also beware of falling victim to spending more time on building tools than learning... I've found that pencil and paper go a lot farther than people give credit.
I don't think it's one or the other. I struggled a lot through differential equations because I didn't have good memorization skills. Had I memorized all of the formula it would have helped me get more out of doing the problems.
It's like learning Spanish. If you memorize a bunch of words you won't be able to speak Spanish. But if you have a bunch of Spanish words memorized than you'll get a lot more out of reading/listening/ trying to speak Spanish than if you don't.
And memorization is usually a far less taxing activity that you can do on your phone when waiting in line or using the bathroom.
there is some truth to this but some areas of math do really benefit from just raw memorization. At a certain point, remembering the definitions and the statements of theorems may not be so trivial, and having them in your head makes it much easier to solve problems.
I agree here. Actually, in most cases, it was enough to know only the definitions. I could come up with applications from definitions on the spot. The thing to watch out for is to get used to the way to solve problems and forget the definitions. I think this is a more frequent occurrence (especially when preparing for the exam.) Then the one will get trouble if they need other applications.
Yup. There are a lot of physics concepts I broadly understand, but solving a problem for them is a different beast.
edit: Also, speaking as someone who mostly taught themselves physics from the books in college and didn't go to class. (I was not a great student, I did _okay_, and probably got lucky).
The caveat to this: I would recommend putting such questions in a separate deck. Solving these problems take time.
What I do: Fact based stuff that I want to just recall quickly go in one deck. I do this deck daily - 5-10 minutes max while having breakfast.
The more complicated cards go in the second deck, which I do only occasionally (e.g. weekends when I have time).
When I initially had them all in one deck, there was no way to do a daily review in a few minutes, and so I would often do it less frequently which messes up the whole point of SRS.
It seems to be a good way to assist with main usage. But I think it's dangerous to try to compose questions with it alone. In my personal case, I found it too burdensome to create such questions. Of course, if I want to learn something really deep, the way you said would be good.
I wanted this at university and tried to write something similar before inevitable employment and distraction. I wish you every success, a couple of ideas that may be applicable:
Probably quite cheap to pay people to onboard sets of well established notes, eg cambridge undergrad maths notes or similar. This may help with dedicated users in niche communities.
Collaboration is hard, but a forking mechanism might be easy+valuable.
Tree structure could guide spatial repetition probabilities, see eg that half-life regression paper from Duolingo.
Thanks for the suggestion. The app already has export and import functions, so it would be nice to have established notes. But I also think people need a note-taking and organizing something by themselves for learning. I think that's why people take notes even though they have well-organized tons of books.
I'm not really sure what it means for the forking mechanism. I would be grateful if you could show me a link to the resource for the forking mechanism you mentioned.
As for the tree structure, the tree structure is already reflected to some extent. As with breadth-first search, the questions starts from the higher level, but when the parent node is in the wrong state, the child node does not appear in the problem. I'll search more about the half-life regression you mentioned!
People would publish flashcard sets (share part of their notes tree) and others could fork and modify those sets, like GitHub. This is in my plans if I ever have enough time to compete in this space.
I think if you pay not much money for this, you'll get bad results. It's very easy to make memorisation materials, but it's hard to make good memorisation materials! Your best bet might be to convince a few Cambridge students that their path to victory lies in making these resources for themselves, and then pay them to give the resources to you afterwards.
The idea is awesome. This is something I didn't know I needed, the biggest advantage is the hierarchical structure makes anki cards much easier to manage. I'm totally your target audience. I like Anki but find the interface for making cards cumbersome.
But a couple of pieces of advice.
1. I think there is too much stuff on the screen and it can feel overwhelming.
2. It doesn't seem obvious how to use it. I think if the first screen you went to was a hierarchical example deck it would useful. An illustrative example is worth a 1000 tool tips.
Yeah it's there, I just think it's hard to find. If I'm new to the app I won't immediately click playground then example. My first time using the app it took me a little while find it. I don't know if this was other people's experience, but here is what mine was.
I went to the website and logged in. I'm trying to imagine how I would use this application instead of Anki, how do I enter/organize information and how do I memorize it.
I'm greeted by a blank screen. Ok I've used workflowy before I see that I can generate some of type of hierarchical structure. But how would that information turn into cards? Ok let me click memorizing. It says I don't have anything to memorize.
Let me try to find an example so I can see how the cards works. Wow there is a lot on the screen, let me try clicking different things. Lots of clicks later... Ahh here's an example.
Clicked memorize. Oh that doesn't work, it doesn't apply to the example. Let me try to copy this over. Oh I can't. Ok looking at this example I know how to build out the relevant structure to generate the cards I want. Build out structure. Ok now I click memorize. SUCCESS!!
I think if I landed on deck, then I'd immediately see how to organize information. And when I clicked memorize it would work and show me how the organized information translates into cards. It greatly shorten the "arrival -> AHA" time.
Okay... Thank you for telling me about your experience. I got where the confusing point is.
Maybe you've noticed it yourself now, Playground is only for organizing parts. I think it would be nice to have a Playground for the Memorizing part, but now it doesn't have that one... To make it appear there you have to put the content in the real Organizing part. Anyway, I found out that I need to make that part a little more understandable.
I think rather than having a playground, you should just put some sample nodes / lists in the Organizing section that people can delete later if they want to. As an example Notion (https://www.notion.so/) populates your workspace with examples from the start, with mock data filled in. That way you can click around with existing data, see how it works, and delete it if you don't want it.
I'm trying to see how the memorize function works and because I have no entries in Organising it won't work. I see there's examples in Playground but I can't see what they look like when trying to Memorize. I spent a few minutes trying to work out if it's possible which was pretty frustrating, and it looks like it isn't, and if it is possible then I have no idea how to do it. Having to invest time into creating realistic mock questions is a pain when there are some right there but just out of reach. The Playground is essentially just Workflowy and doesn't let you experience the rest of the app, it's not a good experience.
That being said, I really like this idea as this reflects how I used to organise a lot of my notes. I'll definitely give it a shot and look forward to see how it progresses.
Thanks for the advice. I didn't know that people would want to see how things actually appear in the Memorizing part. What you suggested (putting data that can be deleted into the Organizing part) sounds very good.
Shameless plug: for bidirectional linking, check out Obsidian! We're in private beta: https://obsidian.md
Spaced repetition might also be supported in the future, but it's not as high priority as links. We also have a graph view for link relationship. Oh, and one thing different from Roam is that Obsidian reads your local folder, so you don't need to put stuff in the cloud, and you own your data 100% (if that's your thing).
Yes. I've heard people who see this tool as an outliner tool want that feature. But to be frank, the priority is low, so implementing that feature will be pretty late.
I love the idea of combining an outliner and an SRS. I don't love the idea of needing a new special editor for them, and a new siloed data store for them.
Now, if this were an SRS that read my existing data (like from Notes.app, org-mode, OmniOutliner, or Excel) and presented it as an SRS, that would be very cool.
I tried to use the "Tell me anything" button in anonymous mode, but it didn't work.
Organizing made sense, but I really wanted to see how the spaced repetition worked from one of the examples, and couldn't figure out how to do that. then, I tried making some tests in organization, and I got "Some error occured" with a few options to refresh. However, I couldn't see the space repetition module. The console had the error ReferenceError: "Intercom is not defined"
I like the idea conceptually, but I found it hard to translate the organization to memorization without seeing an example. I think a video or tutorial front and center is really what I want.
It is an error. Sorry for having such an experience. I was surprised to hear it. I don't know why such an error occurred. Anyway, thank you for letting me know.
I liked the sound of an anki alternative, but it'd have to still be free software and also not need webengine/electron to be a big improvement, imo. I'm not seeing a link to source here after a quick glance...
This is why I love HN. I get to see wonderful new ideas and great tools being developed and also feel a sense of community when the creators that enable the creators also get involved!
I know this is a low quality/value comment to the overall discussion, but I could not help myself!
Hey Dave, I'm currently building a sound management app in electron right now.
I used the electron react boilerplate on github. Can you tell me what advantage using todesktop would bring ?
There's a typo on the About Pricing page that says 'I will likely do this thorugh a subscription model'
Should have been 'through' a subscription model. Like others already pointed out, allowing to play around the app without having to create an account is great. Good work. Cheers!
Feels like more and more there's a ton of smart people converging on ideas. I wrote a flashcard web app with WF as the filesystem. Right now it's very powerful as a personal use system, and I've considered making it public.
It could be cool to take all the FAQs you have in separate blog posts and make them into a 'tree' format, so that I can learn about your product while learning about your product... if that makes sense
This is completely different from LearnObit. If that is an SR tool, it would be a combination of learning resources and SR functions, but LO is a note-taking tool combined with SR function, sort of. Anyway, thank you for showing me an interesting example. I planned to add a viewer for learning resources to LO.
I didn't know the meaning of obit at first. My intention was to reduce "orbit".
(and LearnObit-Rabbit, Learn oh beat, Learn a bit)
I think the meaning of obit fits well with the overall mood of LearnObit, so now I think it's good. Some weirdness is good.
It's nice to see you on Reddit and see you here again.
For now, I will focus on functions as a basic SR program for the time being. So the next big updates are:
The feature to manage contents through a tree structure, the feature to divide media such as images into multiple parts to make it questions, and of course mobile app.
And, actually, I have more ambitious goals than just the SR program. My goal for this is to build an integrated learning environment. In my opinion, the ability to hold personal knowledge can be small but right entry point for bigger things.
Specifically, the first is a ML-powered learning resource viewer.
If you can simply view multiple formats (pdf, video, etc) in one place and make it work with LO on the spot, that would be good.
But it can go further than that. Now, text readers are all bad, because I think they aren't taking full advantage of the digital interactivity. They are just putting paper books on the screen. I want to build a viewer that allows users to interact with texts using ML (like IDE for reading).
The second is a finder of learning resources. Now, the learning resources are quite scattered, and I think it would be nice to be able to manage various learning resources in one place. I think that way, I can build an app like Netflix for learning resources.
And like any other edtech software, I may be able to make it possible for users to collaborate with others in this app. But I don't have many ideas for this part yet.
I am thinking of various functions. For example, when a concept is introduced and later, when the concept appears again, one can immediately see what it was when it is first introduced. That's why I said it's like an IDE.
So you create a backlink by detecting that a concept here refers to a prior concept? Hmm thats interesting. But can't this also be done with a simple word comparison, or am I missing something?
What can ML be used for that is not otherwise in scope?
Maybe the goals you've mentioned aren't about features. As for the success of the app, actually, it couldn't get popularity and is used only by very few people. Maybe I better put all of my brilliant plans behind, pay attention to marketing above all and decide the fate of this depending on the result... (it won't go away, but how much I should focus on it)
Product tours may be enough for people with Anki and WorkFlowy experience, however, I have no experience with them and, I still have no idea what your product does. A demo video would be great.
When I first started to learn coding, I used Anki and the old tried-and-tested Spaced Repetition algorithm, but it wasn't suitable for more complex, general topics like Mathematics or Physics. Because of that, Spaced Repetition has always been widely overlooked despite its proven effectiveness in learning.
I wanted to unlock this effectiveness for a wider audience by building LearnObit, a tool that combines tree-structure note-taking tools with Spaced Repetition so that you can apply it to more than just simple fact pairs.
And now it even has the feature to export data to Anki, so you can leave with your data when you decide not to use it! (Of course I don't want anyone to use it, but I couldn't help it, because people wouldn't start using it without this in the first place...)
Anyway, I'll be here anytime if you have any questions for me. Thanks for your time.