Memorang (https://www.memorangapp.com) | Los Angeles, CA | Full-Time | Remote-OK | Software Engineering Lead |
# Overview
We're building the world's first content-agnostic study platform and exam marketplace from grade school through graduate school. Our mission is to automate and innovate upon cognitive techniques and strategies to level the playing field in education, especially as it pertains to knowledge acquisition, retention, and mastery.
The core of the platform is our data model that supports complex content, from art history to engineering certifications (think Google Knowledge Graph). By unifying the crowdsourced community and premium content marketplace (for 3rd-party businesses) we aim to turn Memorang into a lifelong learning platform for hundreds of career paths. Think of the product as an education-focused chimera of Anki, Kaplan, Dropbox, Amazon, and Squarespace.
# Traction
200K monthly learners (20X growth in 2 years), 30% of US med students, profitability (good unit economics), just raised $500K funding.
# Opportunity
This is right up your alley if you:
- Want to own the tech stack and roadmap
- Want to lead/grow a team (i.e. tech lead/CTO)
- Geek out on spaced repetition, cognitive science, and machine learning
- Like making immediate, positive impacts on users that empowers their future
I can completely relate to that frustration. I came from an engineering background before I went to medical school and I struggled a lot with the different styles of learning (applied principles at MIT vs. knowledge acquisition in med school). The biggest mistake you can make is think that your ability to remember is just an innate ability instead of a skill that can be cultivated.
The problem with spaced repetition software is that "SR" is just one of many known cognitive principles to aid in learning and memory. The encoding process is just as, if not more important than, the review process. I highly encourage looking into interleaving vs blocking, overlearning, learning vs. performance, retrieval-induced forgetting, contextual variability, and the (pre)testing effect. Learning how to incorporate these into your existing study process could really help you out.
This is where great existing software (e.g. Anki) falls flat for many people. If you're interested in seeing an alternative approach you can check out my company's platform Memorang (https://www.memorangapp.com). We recently raised a $500K seed round and are working on some really interesting problems around learning and memory, including a new type of spaced repetition that deviates significantly from the deterministic algorithms like supermemo3/Anki. Probably even more interesting, in addition to building apps for students we're also implementing a platform for developers and education entrepreneurs to monetize their efforts.
Although we're still fairly new we finished second to duolingo at the international reimagine education awards last week. Check out Memorang (web/Android) and let us know what you think (or come join us as a dev --> founders@memorangapp.com).
Wish it could import Supermemo knowledge (with history). Simply can't afford switching to a new tool at this point (7 year Supermemo user with thousands of knowledge items).
If i may ask: how do you feel supermemo aided you? How regularly/much do you use it? What items do you store and what card models have you come to prefer (cloze, q-a, ...?) How has this impacted your life?
I have seen some kind of supermemo import for anki, not sure if it's maintained. Possibly it was just a script someone created - sorry can't find it right now.
I use it to memorize Latin (and some Greek) vocabulary. What I use mostly is Q-A. Typically it looks like this:
Q. lat. adj. black (shiny)
A. niger, -a, -um
And the other way:
Q. niger, -a, -um
A. black (shiny)
So when I enter the knowledge, I write in a .txt file the first form, and the script I run generates the second. (I read on Supermemo's website it's good to test the reverse case too, so by default I do this for all knowledge items). Over time I found it was useful to prefix vocabulary with the language (lat.) because you'll typically have the same words in another language (grk.) and if it's an adjective or adverb to append (adj./adv.) after the language prefix (I wanted a single database). For verbs I found it's sufficient to write the English part as the infinite to .. (e.g. Q. lat. to kill, A. neco (1). I will occasionally write the word with macrons in brackets after the word if it's weird, e.g.
Q. lat. adv. daily
A. cotidie (cotIdiE)
I wanted to separate the two because real Latin literature does not contain the long/short macros and I wanted to condition myself to that from the start.
I use cloze mainly for grammar.
I use Supermemo daily and it takes about 20-30 minutes to memorize what is currently ~6500 items. My daily workload is between 70-100 items and my recall is about 85%. I will admit that I'm extremely hard on myself and will only score a 5 if it's absolutely perfect.
Whatever I've looked at before of Supermemo -> Anki import did not get it right 100%. I don't want my repetition history or workload screwed around with. I also want the ability to go back to Supermemo if it doesn't work out. Unless I feel comfortable that I can switch between both systems seamlessly I'll probably never switch. I run my Supermemo on a Windows VirtualBox in Linux so it can stay that way forever.
I would probably never use Supermemo for a "live" language. It would be a waste of time, imho. For languages which are not really spoken like Attic Greek or Latin, it is indispensable.
My only real problem with Supermemo is that there is no way to access it over the Internet. The database lives in that Virtualbox Windows. It would be amazing if it had the ability to spawn a little webserver that you could do your drills from.
I'm not sure what you are angry about, but please let me try and clarify a few things:
>A better app that doesn't crash as much
Working on it. Worth noting that it's free to download.
>Thousands of free decks
Memorang has about 5 million free flashcards currently as part of the crowdsourced ecosystem. You can get content in 4 ways: creating, importing, searching, and purchasing. You can import from Quizlet, Excel, Anki, StudyBlue, and Cram, or collaborate in groups on common topics.
>An open ecosystem
While I think we could do a better job with content discovery, the collaboration tools on Memorang do not exist on Anki. If you wanted to work together on an Anki deck, it would have to be with some combination of dropbox or git. My opinion is that Anki is pretty antisocial.
>Source code
I think HN readers know more than anyone that there's nothing wrong with not open sourcing your company's code. Developers need to eat too. That being said, we are open to making our API publicly available to build learning applications on top of.
>Doesn't spam your email
I welcome your suggestions regarding email and onboarding. Currently we send a welcome email and account verification. Over the course of 1-2 weeks we give short snippets of how to use certain learning modes on the platform. You can unsubscribe from emails at any time via the unsubscribe link or via your account settings.
>Isn't an obvious monetisation of anki's general premise
It's worth pointing out that the creator of Anki has made tens of millions of dollars off of charging for downloads. I'm not sure what "premise" you are referring to - but ours is that the core learning application should be free and not a paid download.
>How anyone thinks that paywalling a bunch of flashcards is a good idea (when anki offers them free) is beyond me
Crowdsourced content is free on both Anki and Memorang, so I think you may have missed that the "paywalled" flashcards are actually an optional upgrade. As a bit of background, we've spent thousands of dollars on content creation and are committed to continuously upgrading and maintaining that content. For example, our USMLE Step 1 and MCAT content are one-time purchases for lifetime access and we maintain updates to stay current with the latest medical information at no extra charge. Users are happy to pay for that level of trust and security.
I hope the above information helps you better understand what we are doing.
You can also click the search bar in the top left and go through the user-generated database. I believe there are about 5 million flashcards there currently.
We have this at the admin level and plan on making this a user-accessible feature as soon as we can.
>Does your service offer benefits to people who aren't and don't have the resulting new card review burden or need for cramming?
It depends on your goal. In its current iteration, Memorang functions very well for mastering new concepts on a shorter-term scale (e.g. hours, days, or a semester). However, long-term retention on the order of years is something that we are working on a different implementation of. That will be a huge focus in the next semester after we finish shipping this 2.0 launch in January.
We're starting with the assumption that learners will not be able to religiously follow a schedule. That means that whether you're cramming, checking in sporadically, or completing a daily queue - the algorithms should ideally adapt accordingly without making your "to do" list untenable. One component of this is determining whether your long-term retention is bounded to a specific timeframe (e.g. learn content X by future date Y) or unbounded (lifelong learning).
What we're borrowing is the concept of memory decay, but I don't think there are many similarities beyond that. Imagine learning spanish in college: you could set a goal for your final exam but also have a lifelong learning goal. Whether you're following your schedule or cramming, your answer events will communicate with the intersecting algorithms so that you're being optimized for several different use cases simultaneously. When we launch the "goal setting" feature, gjcourt may write a blog post about it and post it to HN...
Don't have strong opinions one way or the other on Knewton. That being said, I generally feel that enterprise software where you sell to institutions and publishers is fundamentally disconnected from software that's targeted at the users directly.
I would greatly prefer to hear students raving about a learning product rather than their instructors. In your case you seem to be a bit underwhelmed by what they offer.
Very good work. I am extremely happy for you! I have no clue why some people expect this all to be either open source or free. I have personally invested a lot of money on building an adaptive learning tool (and two years of salary losses). We are not yet live, but when we are, I hope to have some revenue model as well.
Thank you, it really means a lot since we've been working on this for over 3 years now (finally hiring employees this past month!). I wish the best of success to you too.
In reference to people wanting freebies - that's just human behavior. If they see sufficient value in a product (e.g. iPhones), then they will happily pay for it.
What you're describing is exactly in line with what I've experienced anecdotally and what we've seen in our studies and retrospective analysis of tens of millions of questions answered.
I think the problem with Anki can be distilled down to root assumptions it makes about how people should ideally learn and how to model what is forgotten over time. In studying for a course there are basically three strategies: (1) cram at the end (2) study reasonably throughout the whole course (3) try to keep re-learning the course indefinitely.
We've thought long and hard about how we can cater to these different styles and think we've done a good job with a hybrid of #1 and 2. What would make more sense is something coming early next year as an option feature, which is inputting the date of the final exam of your course and then periodically adding in flashcards to a growing playlist that optimizes retention for the day of your exam. After you finish, you don't necessarily have to continue, but you could set a longer-retention date. An example of this would be that you have weekly quizzes and a final exam, and the algorithm could adjust the date for each quiz while balancing the overall arc towards the final.
That's interesting... my original comment wasn't actually meant as a criticism against Anki itself, more a criticism against college schedules. And that's because I was taking as gospel that the Anki/SuperMemo style of algorithm is basically what you need for long-term retention.
It sounds like you're not so much challenging that as you are offering more flexibility and control over what and when you'd like to retain?
(I wonder what college degree programs would look like if they actually were structured in a way to better reinforce long-term retention.)
# Overview We're building the world's first content-agnostic study platform and exam marketplace from grade school through graduate school. Our mission is to automate and innovate upon cognitive techniques and strategies to level the playing field in education, especially as it pertains to knowledge acquisition, retention, and mastery.
The core of the platform is our data model that supports complex content, from art history to engineering certifications (think Google Knowledge Graph). By unifying the crowdsourced community and premium content marketplace (for 3rd-party businesses) we aim to turn Memorang into a lifelong learning platform for hundreds of career paths. Think of the product as an education-focused chimera of Anki, Kaplan, Dropbox, Amazon, and Squarespace.
# Traction 200K monthly learners (20X growth in 2 years), 30% of US med students, profitability (good unit economics), just raised $500K funding.
# Opportunity This is right up your alley if you: - Want to own the tech stack and roadmap - Want to lead/grow a team (i.e. tech lead/CTO) - Geek out on spaced repetition, cognitive science, and machine learning - Like making immediate, positive impacts on users that empowers their future
# Stack Python / Django, Node, Redis, PostgreSQL, React(JS/Native)
I'm the founder and can't wait to hear from you: yermie@memorangapp.com