Comment by arvidkahl | Hacker News Reader

Comment by arvidkahl | Hacker News Reader

The best products happen at the intersection of an existing niche you are some level of expert in already and a technology that has not yet been adopted in that niche.

I've found one approach work very well with my mentees:

- Figure out which "special interest groups" you are part of beyond software engineering. That can be "aquarium owner", "coffee lover", "morning person", "diligent grandson" — the less technical, the better.

- Among these "niches", find the ones that could benefit from a transfer of technology, like (spitballing here) "teachers who work from home" (education niche) + "automated submission and pre-grading of homework" could work (digital document collection and rule-based checking logic), or "aquarium owner" + "nitrate cycle tracking IoT device" (hardware-enabled analytics) + "optimal light scheduling" (machine-learning-supported recommendation engine).

Do that for all the groups you're part of, and you will find lots of ideas that aren't just "scratch-your-own-itch". They are 'scratch-an-itch-you-understand-and-know-how-to-remedy'.

Is it only me here who experienced such paradox:

I love tech side-projects but often I don't want to introduce more tech in my non-tech hobbies and activites as they are remedy for too much screen time. E.g. my gardening hobby highest tech is $25 weekly schedule watering valve and I feel very good about it.

"teachers who work from home" (education niche) + "automated submission and pre-grading of homework"

I homeschooled for a lot of years and my first blog was a homeschooling blog. I was involved with The TAG Project as part of supporting my homeschooling and I really hate this idea.

"Teaching to the test" is terrible teaching. Multi-guess answers aren't good ways to test student knowledge. They are just convenient for a teacher who needs to grade 25+ students.

Please don't run around thinking of yet more ways to making it nominally more convenient and easy for individual teachers to process tests or homework or whatever from large numbers of students. This is not a way to enhance the transfer of knowledge to the future generation and is also not a good credentialing method.

There are cool things being done with tech in the education space. There are people going online and learning things they want to learn because there are rich materials available for free. But they don't look anything like "pre-grading of homework" for teachers.

Yeah, but you're making an assumption, who said it had to be multi-choice answers?

Why can't someone build a NLP model to test essays of middle schoolers or high schoolers graders on American history? You can look for learning objectives in the essay and grade upon that.

Why not build a tool with D3 or another modeling tool to teach chemistry? You could have them build compounds and then if they make a mistake, insert via a tool tip and tell them why. As they combine things, you read out what it is or test them based on the string they put in for the compound. You have to build it and name it correctly.

Why not build a tool with D3 or another modeling tool to teach chemistry? You could have them build compounds and then if they make a mistake, insert via a tool tip and tell them why. As they combine things, you read out what it is or test them based on the string they put in for the compound. You have to build it and name it correctly.

This sounds really cool and is completely unrelated to what I am saying "Please don't do."

I think there are two great arguments here: nobody needs tools that perpetuate a vanishing system and there are lots of interesting things happening in EdTech. While I am grateful for your example, I would like to offer a different perspective.

I know of a university teacher who is looking into solving the arduous task of grading without going crazy. I approve of a tool being built to facilitate this.

I am torn on this issue as well: I strongly dislike formal education, with it following a metric of "teaching to the test" and "learning to test instead of retention." I wish that there was more systemic change towards a different way of assessment.

But not for one second will I assume to know what is best for the teachers in that space. If building a pre-grading (which is merely an example here) can help them shave off an hour of tedious work a day, I think it's a valid side project idea. It may even be a valid business idea.

The project that changed my life[1] was an EdTech productivity system that generated student feedback for online teachers working with young students in China. Student feedback as text-only is a horrible way of communicating individualized instruction. Yet we help teachers speed up that process, which allowed them to do one thing: teach more. Be more present in the classroom. Foster stronger relationships.

I would argue that making anything nominally more convenient is a good thing: it removes an inconvenience so that things that matter can take place. No teacher would call teaching inconvenient. Lots of them would call admin stuff very inconvenient.

Again, it's not about "teaching to the test" per see. It's about removing barriers to allow for meaningful, accomplished, and impactful work.

[1] https://thebootstrappedfounder.com/from-founding-to-exit-in-...

I know of a university teacher who is looking into solving the arduous task of grading without going crazy. I approve of a tool being built to facilitate this.

This is a different thing to me and a better suggestion: Solve the problem of someone you personally know whose work you believe in.

That's completely different in my mind from "pull some generic education-related blurb off the internet and build that." Even if, nominally, the two projects sound like exactly the same thing.

In practice, they probably aren't remotely the same thing, which is part of why we have memes like "Ideas are worthless. Execution is everything."

Thanks for clarifying this. I agree with this, one hundred percent. I think that if you're not an education expert or enthusiast, you'd better go look for problems to solve somewhere else — which was the premise of my original comment. It pains me every time I see someone building a business without the passion to help their audience.

Aquarium owner here. Thinking of IoT hardware but turns out the sensor is expensive and error/breaking prone.

Thinking to create yet another social media for people to post their aquariums and track the maintenance schedule then I realized I won't have the bandwidth (time and money) to deal with storing massive amounts of pictures, or policing the content to make sure it doesn't go down like social media dumpster fire of porn, gore, politics, religion, and other offensive materials

Take the Google Chrome approach and just have the AI randomly ban posts at a high rate.

You don't need "AI" for that

If you call it that you can raise a million dollars first.

I guess that's what I did. Niche is fighter pilot, product is generating target imagery for training. dieselplanning.com.

Reply on news.ycombinator.com