Ask HN: I've gotten some feedback on my MVP, what do I do next?
1 comments
There are positives to the data-entry pain: once somebody's entered their data, they have a real investment in your system. Question banks can be reused and grown, so the pain does decrease over time. Depending on how easy or difficult you make it to export their data, it may help user retention.
For users zero to ten, you might want to consider taking this problem off their hands: let them upload a photo of a worksheet, and let the system "magically" transform this into the proper format. (ie, OCR the data, then edit it yourself or pass it on to mechanical turk). If they're not completely convinced of your value, even with this level of assistance, then you've got no market. If they do buy in, then you can figure out how much pain you can leak back into the system before they balk.
One potential value-add: math problems from teachers often have hard-coded values. Current test-bank software lets teachers convert these to formulaic question generators, but this is even more time-intensive of a process than straight data-entry. Applying some "magic" to this process is a ripe opportunity.
For users zero to ten, you might want to consider taking this problem off their hands: let them upload a photo of a worksheet, and let the system "magically" transform this into the proper format. (ie, OCR the data, then edit it yourself or pass it on to mechanical turk). If they're not completely convinced of your value, even with this level of assistance, then you've got no market. If they do buy in, then you can figure out how much pain you can leak back into the system before they balk.
One potential value-add: math problems from teachers often have hard-coded values. Current test-bank software lets teachers convert these to formulaic question generators, but this is even more time-intensive of a process than straight data-entry. Applying some "magic" to this process is a ripe opportunity.
I like coding, but I don't like coding for the sake of coding. I also feel like I have the most domain expertise of any particular area in K-12 education, because I'm a high school student. Last year I started working on a project, something that makes it easy for students to do common[1] homework online and for teachers to grade it faster. The value that's supposed to be gained from this is mostly for the teachers; The majority of their time for class-related activities outside of school is spent grading[2]. I've gone through three different major iterations of this, where the first I really had no idea what I was doing. With the second iteration I had more of an idea, but was approaching it wrong.
A couple weeks ago I showed two of my teachers, and some of the other students in my school (outside of my friends) the current iteration. My teachers were very impressed, but it made me realize that I'm not building something useful. Or, it might be but the barriers to entry are too high.
Before I can explain why, I need to explain the implementation[3]. Each teacher gets a subdomain, which is only for their classes. The teacher has to create the assignment, and the current iteration only allows for Q/A types of assignments (i.e a worksheet). Teachers create or copy the questions from existing worksheets, give them a maximum score, and then a due date. When the student opens it, they see the homework and can get started. The student just answers each question, and clicks the turn-in button when they're done. After the due date is over, the teacher can grade it. When they're done grading it, teachers can view an alphabetical list of who did the homework and their score.
There's a glaring issue with that though - it doesn't do anything other than save teachers the time for making 150 photocopies of each assignment, passing it out, (the students) turning it in, grading it, and then passing it back (if at all). The time saved is roughly the same amount of time spent adding in the assignment to the webapp. That equates to nothing other than changing where the homework is done and graded, which isn't solving the intended problem.
There are more issues. There's a range of roughly 7-10 different types of homework. I can't support everyone, so that means I need to focus on one kind of assignment, but finding an implementation that doesn't involve a lot of effort into getting that assignment online is a huge challenge. In addition, there are several usability issues with the current implementation.
But, there is a time limit to this. I'm taking the CHSPE in March, going to community college for a short period of time, and then going to a university. That means I need to get something useful in my teachers[4] hands before January.
The Question:
Where should I go from here? Should I continue work on the existing iteration, fix it up a little, and then add support for different types of assignments? My teachers were willing to give it a try in their classes, should I let them in it's current state? Should I redo it from scratch and focus on a different type of assignment? Should I try a completely different approach to making grading quicker? Should I solve a different problem that I personally deal with?
What does HN recommend?
[1]: By "common", I mean the kind of assignments students get very frequently. For English, reading a chapter a night and answering 3+ questions. For Math, doing 30+ problems out of the textbook. For History and Science classes, outlining a chapter in the textbook, or worksheets.
[2]: I've noticed that there are a few different common ways for teachers to deal with large quantities of homework to grade. Many just check off the homework (and don't grade it). Some have the students grade it, and don't give full credit unless they graded it (using a teacher or textbook provided answer key). A few have their T/A grade most of their students work. Some stamp the assignment to show it was done on-time, and then collect it in large groups for grading later.
[3]: I can gave a link to the demo, but I don't want to on HN. If you want to see it send me an email.
[4]: At one point in the far future I want to sell it as a SaaS solution directly to teachers, cheap enough that they can pay for it out of their own pockets.