Ask HN: What features do you want in a modern spreadsheet app?
Hi HN, I want to build a new spreadsheet app which works on all devices. Excel is great for what it does but it doesn't understand the internet and collaboration. Google docs is great for collaboration, but is still a bit clunky and lacks a lot of useful functionality. The Google drive mobile spreadsheet apps have terrible UX.
So I want to build a new spreadsheet app that works great on desktop browsers and also on phones and tablets.
These are some features I have in mind:
1. Hook into all the rich web APIs around us and pull in data from places like Evernote, Twitter, Fitbit, Basecamp, etc straight into a spreadsheet to play with. I can't believe we don't already have this!
2. Reliable format conversion between CSV, JSON, TSV, Excel, Gdocs, etc (as suggested by couple of people in the comments).
3. Shortcuts for common data manipulation functions like basic arithmetic, concatenating strings, converting delimited data to columns etc. I'm imagining simple big buttons for these to make them easy and fast to do on a touch device.
Now I want to find out what things HN would find useful to help me prioritise some features for v1.
Any ideas and comments welcome. Thanks! tl;dr: Use spreadsheet for simple small business accounting, directly, with ability to export to accounting package (and to gen timesheets and invoices), out of the box, without me having to grow my own macros, etc. This may be a very limited use-case, but.... I'm a one-man shop. Every month, using my (relatively quick and lightweight) spreadsheet software, I enter time worked for various clients, prepare timesheets and invoices for them... and then re-enter said data into my (slower, less friendly) accounting package. The regular monthly stuff is almost always the same, very cookie cutter. Having an accounting package is almost overkill, but it supports "accountant interoperability"[1] - and using the latter to do the former is definitely overkill. And I've never been able to figure out the automation/recurring capabilities, they seemed geared for more complex enterprises. I hate the fact I enter data multiple times, I hate that I have to use the accounting package to manage expenses, etc., etc. - but integration and automation are complex, definitely overkill for my needs. What I'd really like is the ability to use a spreadsheet for basic double entry bookkeeping (there are only a handful of accounts I use regularly; I can use the accounting package for the infrequent ones) and for entering my time, and have dirt-simple spreadsheet functions for generating client timesheets (sometimes based on the client's format, since my clients are often middlemen), for generating invoices (based on my format), and for generating a file that can be imported into my accounting package periodically. +1 if the spreadsheet knows about regular monthly things and reminds/walks me through them. Lightweight, fast, cross-platform. All the things my accounting package (Windows-only, so runs under a VM) is not. [1] Accountants are like lawyers, IMHO: If you're unsure whether you need one, you probably do; my annual taxes cost me more than doing them myself, but I have greater assurance of minimizing my tax bill while staying off Revenue's radar - audits are expensive. What is a spreadsheet, really? It's an interactive programming environment with an interface optimized for data entry and some other things. The difference between Excel and, say, Matlab or R, is excel's strength at data entry, visual/intuitive nature, and total lack of accessible higher-level programming features. So the ideal solution in this problem space, to me, looks not much like excel at all, but more like a high-powered computation environment a-la mathematica, except with roughly the execution model and visual/accellerated/intuitive interface strength (and then some, as you suggest) of spreadsheets. For example, the whole grid thing might have to go, instead making tables first class objects in the language, and then have a strong visual representation with an accelerated (not plain text) interface for manipulating them. But maybe you want to take it in a different direction than I do? In any case, excel can trace a pretty direct line to VisiCalc, with very little fundamental reevaluation applied since 1978 or so. So you might want to consider not being bound too closely to the "this is how spreadsheets have always been done" thing. Best of luck to you! I suspect you don't know how large this particular iceberg is under the surface[1], but I wish you the best. :) 1: There's a reason that POI-HSSF, the Apache Java-to-Excel library, stands for Poor Obfuscation Implementation-Horrible SpreadSheet Format! What I want is for people to stop using spreadsheets for anything but numbers. There's a terrible habit just about everywhere computers are used to use Excel as a catch-all table generator for whatever grid-like data someone wants to capture (task lists, for example). It's a terrible tool for that - the cell editor is actively hostile to entering text. And frankly, I think the majority of Excel use is for exactly this sort of clunky, awkward crud, not actual spreadsheeting. Rather than trying to build a better spreadsheet, build a better grid-oriented text capturing device with some spreadsheet functionality. You have struck the nail on the head. I have wished for this exact thing - a better grid-oriented text capturing device - so many times. By inverting the focus, you have helped me look at it in a way that might just be the thing I needed. Thank you so much. Drop me a line at the email in my profile... I'm really interested in how you go about this. I've thought about exactly this product, but stuck on the "If you have a hammer, every problem looks like a nail" effect, which is why businesses use horrid Excel for this duty in the first place. How do you get them to buy in? Your multi-platform approach might be the answer. ps: Have you used any of 37signals' products? Look to them for UI simplicity. I think there's a real danger in adding too much functionality here. Make number 1 into an API and that's your business. In other words: all these services can easily provide a button: see this data in excel, and your service sends the data to Google docs (or your own excel version if you must). I'd like to be able to tie blocks in block diagram to the cells in a spreadsheet. The use case is to view which specifications are required per functional block in a large system. Just today I had a couple issues with Excel if I wanted to save something as JSON or if I wanted to save something as an UTF-8 encoded CSV (Excel converts it to ASCII). I've come across several data format conversion issues, in my personal use cases and of others. Would a reliable format conversion app be of value on its own? My company uses the French version of MS Office and after upgrading, I can't quite get it to generate CSV files in the right encoding needed to work with the company's proprietary software. So I need to go through the hassle of manually generating the CSV files in the correct format. An easy way to work with different encodings and formats will be especially useful for collaboration. One thing that would be simple and useful is the ability to generate or convert between different CSV file formats. For example, French CSV files use semicolons instead of commas to separate values, and use commas as decimal points. Thanks for your comment, Jongseong. I'd like to learn a bit more about your file conversion issues. Can I email you somewhere? Yeah, probably. A closely related problem is cleaning up data. Something that can detect unclosed quotes and/or unclosed HTML tags. Something that can figure out when a word should be capitalized or lowercase. Here's another problem: how do you make a field in a CSV that when you convert the document into JSON, the field becomes an array? seems like you need to define who you are tagerting first. If you want to replace excel it's bankers, consultants etc. Anything that does not integrate backward compatibility with excel and vba macros has 0 chance in the marketplace - too many complex models built on this s* by banks Wolfram Alpha and MatLab integration. Thanks for the suggestion. Can you tell me how you would use such an integration? At least for me, what would be really valuable is the ability to easily add data from WA to a cell. For example you could type "population of Russia" into a cell, and it would grab the integer from WA for the math. Really a low-latency, tabular form of Wolfram Alpha would be the spreadsheet of tomorrow alone.