Ask HN: What do you wish someone would build?

12 points by prmph ↗ HN
Looking to build something really productive and useful as I invest time into learning additional technologies.

My skills are available to HN members. I'm able and willing to build pretty much anything along the lines of B2C or B2B tools (therefore excluding compilers, database engines, or operating system, and the like).

What do wish someone would build for you? The emphasis is on something that solves a practical problem for you, not just interesting.

I will select a wish that appeals to me, and give free invites to HN members when it is done.

27 comments

[ 2.6 ms ] story [ 52.5 ms ] thread
A video chat app for desktop and mobile that is simple, minimal, just works, and works well even under low bandwidth conditions.

All the existing solutions I can find are feature-encrusted junk where the core video chat functionality is just awful. If anything, video conferencing apps have declined noticeably in quality in recent years.

I recommend Appear.in http://appear.in
I second appear.in

Several times now I've been in a group chat where someone's audio or video isn't working so we go to appear.in and everything works perfectly

I'm writing what could potentially be a competitor to appear.in and I also recommend them :P .

They're definitely the cream of the WebRTC crop right now.

I wonder if TeamTalk (http://bearware.dk/) would meet your needs. The desktop front-end is relatively light as these things go, and isn't browser-based.
I'm not suggesting it's great, but just to spur discussion: Does Google Hangouts solve this problem for you? If not, what would you change about it?

I don't love it, but I work on a distributed team and we use it every day for standups.

This is really interesting. I'm particularly struck by this item: "order in a way that takes urgency, importance and deadlines into account"

Do you envisage being able to configure this ordering (for example by weighting the various factors)?

Personally, my mental to-do model consists to a pretty large part of recurring tasks with more or less vague "deadlines."

For example, at home, taking out trash, watering plants, stocking up on groceries, paying the rent, etc.

However, almost all to-do systems seem to be aimed at scheduling one-off tasks.

So I would like the to-do system to more accurately model my real to-do model, which would probably mean having a special item type for these circular, Sisyphean tasks.

I use Toodledo [1] for this, especially for the recurring tasks. There are even different ways to define how it recurs, eg taking out the trash might be every Wednesday, but watering plants might be "2 days after I last completed it". When you complete the task, it automatically enters the next recurring task for you (and hides it from you until the appropriate day).

[1] http://www.toodledo.com

Potentially, at some point (i.e. not the MVP) as a "power user" option.

For the MVP, I would simply order by the Due Date, with "Important" (starred) items jumping to the top of the list (and if there are multiple "Important" items, they get ranked by Due Date. "Overdue" items (i.e. where the Due Date is in the past) would get jump over the "Important" items.

So, in this mock-up - https://disruptdecentralisedisintermediate.files.wordpress.c... - you see a bunch of items ranked by Due Date (i.e. the ones towards the bottom of the list have a Due Date further in the future), with an "Important" item ("Draft product roadmap presentation") above the others (even though it is Due _after_ the "Get milk" and "Call Cynthia.." items) and the "Call Joe Bloggs" item appears right at the top (and is highlighted) because it is Overdue.

dude, you can do almost all of that with a simple google spreadsheet. You can just invite others to collaborate on the doc and add a column for who the task is assigned to etc.
See if you can solve this problem with software. The article and comments pretty well articulate the problem.

Weeding the Worst Library Books https://news.ycombinator.com/item?id=11586061

A combination of data from Amazon and Google Books along with library checkout and interlibrary loan requests might be able to lead you to figure out what books to store at what library location in a city, what books to dispose, and what type of books patrons of a specific library prefer.

Interesting; I see this as a special case of a more general problem that I might take on. I have ideas about an organizational decision making tool that aggregates input from stakeholders and web APIs in an intelligent and transparent way to rank options
Beets [0] for arbitrary files with available tags/metadata defined through config on a per-filetype basis.

Being able to organize and query all of my files on the same level I can my music would be amazing.

[0] http://beets.readthedocs.io/en/latest/index.html

I've forgotten what it's like to having file-level access to my music at this point...

I suppose it could be made to work with an API too.

Can you elaborate on this a bit more?
Over the past few years, I have mostly used streaming services (Spotify, Apple Music, Beats) which have DRM on their local caches to prevent accessing the files directly.
I was actually referring to the API bit of your comment.

I suspect providing this functionality just on Linux-y systems might not be very useful. If there is a way to make it web-based but still work with local files on each platform, that might be more useful.

Yes! The Spotlight Search on OSX is nice, but to be able to write full SQL-like queries against file data/metadata would be unbelievable.
Motivation to spend more free time doing two things - going to the gym and solving programming challenges.
more like motivation stuff based on what you want to do
A robust, fast (think BLAS / LAPACK) matrix library a la Numpy or Eigen, except for Scheme. Something like an R7RS library would be great, however if it only worked on one of Gambit / Gauche / CHICKEN / Guile that would be fine too.

One of the biggest problems seems to be coming up with an efficient / sane way of slicing or passing matrices without requiring explicit copies. It would also need to eventually support sparse matrix operations, as well as complex numbers.

To be honest, recreating something similar to Numpy / Eigen in any language would be a pretty big feat. I've toyed with the idea of wrapping Eigen into CHICKEN Scheme before, however incorporating C++ classes into Scheme can be confusing and cause some serious headaches.

As for why I want this -> I want to be able to do really fast matrix operations with the elegance of Scheme as a language. Outside of fast and easy linear algebra, I've wanted for nothing whenever I've used Scheme.

An ID/Medical/Etc card scanning/viewing mobile app. Something that will nicely scan my wallet cards and allow me to easily bring the scans up and (if needed) email/print/photocopy the images of them on demand.

Bonus points for:

- Search capabilities (requires OCR, or at least manual tagging)

- Geolocation (if I'm at a doctor's office, bring my medical card to the front).

- Security - Encrypt the images in memory and require a PIN to unlock.

- Bright mode for barcode scanning/photocopying (which overlaps with other apps like Passbook, Wallet, etc, but they don't support non-shopping cards).

you can use google drive to do this. Click the new document button in the app and select "scan".
I would want a better cli or curses debugger for javascript. One that does not involve connecting to chrome over a port.

Also, if you're interested remote-pair-programming on whatever you're building, please let me know.

I will likely take you on your offer. Stay tuned.