- Joined
- Jun 14, 2014
- Messages
- 17
- Reaction score
- 7
Hi,
I would like to bounce some ideas with you guys here to see if I am going in the right direction.
I would like to invite you to an experiment on collaborative online learning on the http://buboflash.eu website that I created. It uses statistical model of your memory to make sure you have very high knowledge retention (like 98%) and you learn about 20 times faster than traditionally from books alone.
All learning material is free of charge and under Creative Commons licence or comes from publicily available webpages.
This is how I did it:
In BuboFlash knowledge is represented as flashcards (cards with question and anwser). When you learn a new flashcard, your memory strength - defined as probability of recall - starts decaying in time. The longer you wait, the lower probability of recall is. However, if you manage to repeat it while you still remember it, the memory strength is reset back to original 100% (certain recall, you just recalled it after all) and it starts decaying slower, so you remember the flashcard for longer. Each successful repetition makes subsequent memory decay slower, so BuboFlash can increase intervals between repetitions - 1 day, 3 days, a week, 10 days, a month etc. You can easily learn 1,000 new flashcards per month or new 10,000 flashcards per year, working with BuboFlash about 40 minutes a day.
No time wasted on repeating things you still know well. No frustration of forgetting too much.
You will repeat only what you need and only when you need it.
I am well aware that what I described is known as Spaced Repetition Software - I have been an active user of many systems (mostly Supermemo) for nearly 20 years myself. There are even programs targetting specifically medical students: Firecracker, formerly Gunner Training, or students selling their Anki decks, a new startup called Enki Learning, Osmosis, and the New England Journal of Medicine's "Knowledge+".
BUT
I found all these programs inappropriate for lifelong learning, which is utterly annoying. You can't collaborate on creating knowledge there effectively or at all. For me it was always like learn something, then drop it. Never all-encompassing, primary study tool. I tried to solve it on BuboFlash by letting everybody edit everything and storing all versions. Then I let users bookmark which version they like most. Wikipedia is also crowdsourced but in Wikipedia it is always the last version that wins, which is unacceptable for learning, people would overwrite each other's versions. On one hand, you would be surprised during repetitions to see something different than what you learned, and on the other hand, you may want to get latest changes from somebody else.
Next thing was the source of flashcards themselves. Of course you can buy them, but come on, it is inflexible, and often tied to a specific program. And the Web is full of extremly valuable learning material. My bet is on incremental reading as a source for flashcards, which works as follows:
Incremental reading is both a source for flashcards and a study tool on its own.
Let me emphasise it, whatever you guys create - I don't claim it, it is yours. You can share it by default on BuboFlash, and for tech-minded people, I can export it as JSON, so you can do whatever you want with it, no problem here.
It is almost summer holiday time, so if I get enough feedback, I will have enough time to make it a killer app before next semester 🙂
Cheers,
Piotr (aka puchacz)
PS. I am a real person who wrote buboflash, not an anonymous face of a company, you can mail me directly: piotr.wasik [at] gmail.com or discuss your ideas here.
The screenshots:
1. Annotate an article
2. Make flashcards (on the right) by hiding some parts in annotation itself:
3. Repeat the flashcard
I would like to bounce some ideas with you guys here to see if I am going in the right direction.
I would like to invite you to an experiment on collaborative online learning on the http://buboflash.eu website that I created. It uses statistical model of your memory to make sure you have very high knowledge retention (like 98%) and you learn about 20 times faster than traditionally from books alone.
All learning material is free of charge and under Creative Commons licence or comes from publicily available webpages.
This is how I did it:
In BuboFlash knowledge is represented as flashcards (cards with question and anwser). When you learn a new flashcard, your memory strength - defined as probability of recall - starts decaying in time. The longer you wait, the lower probability of recall is. However, if you manage to repeat it while you still remember it, the memory strength is reset back to original 100% (certain recall, you just recalled it after all) and it starts decaying slower, so you remember the flashcard for longer. Each successful repetition makes subsequent memory decay slower, so BuboFlash can increase intervals between repetitions - 1 day, 3 days, a week, 10 days, a month etc. You can easily learn 1,000 new flashcards per month or new 10,000 flashcards per year, working with BuboFlash about 40 minutes a day.
No time wasted on repeating things you still know well. No frustration of forgetting too much.
You will repeat only what you need and only when you need it.

I am well aware that what I described is known as Spaced Repetition Software - I have been an active user of many systems (mostly Supermemo) for nearly 20 years myself. There are even programs targetting specifically medical students: Firecracker, formerly Gunner Training, or students selling their Anki decks, a new startup called Enki Learning, Osmosis, and the New England Journal of Medicine's "Knowledge+".
BUT
I found all these programs inappropriate for lifelong learning, which is utterly annoying. You can't collaborate on creating knowledge there effectively or at all. For me it was always like learn something, then drop it. Never all-encompassing, primary study tool. I tried to solve it on BuboFlash by letting everybody edit everything and storing all versions. Then I let users bookmark which version they like most. Wikipedia is also crowdsourced but in Wikipedia it is always the last version that wins, which is unacceptable for learning, people would overwrite each other's versions. On one hand, you would be surprised during repetitions to see something different than what you learned, and on the other hand, you may want to get latest changes from somebody else.
Next thing was the source of flashcards themselves. Of course you can buy them, but come on, it is inflexible, and often tied to a specific program. And the Web is full of extremly valuable learning material. My bet is on incremental reading as a source for flashcards, which works as follows:
- you read the web and highlight interesting passages
- you review previously highlighted passages - it is a priority list, not a date based scheduler like with flashcards,
- when reviewing passively you will notice that you start losing context and your understanding starts to fade, so:
- if you really want to remember them, convert them into flashcards by hiding some parts of the text and start repeating them seriously as described above.
Incremental reading is both a source for flashcards and a study tool on its own.
Let me emphasise it, whatever you guys create - I don't claim it, it is yours. You can share it by default on BuboFlash, and for tech-minded people, I can export it as JSON, so you can do whatever you want with it, no problem here.
It is almost summer holiday time, so if I get enough feedback, I will have enough time to make it a killer app before next semester 🙂
Cheers,
Piotr (aka puchacz)
PS. I am a real person who wrote buboflash, not an anonymous face of a company, you can mail me directly: piotr.wasik [at] gmail.com or discuss your ideas here.
The screenshots:
1. Annotate an article

2. Make flashcards (on the right) by hiding some parts in annotation itself:

3. Repeat the flashcard

Last edited: