What if you started with a textarea, in which you added an item per line. You could be specific ('500g Barilla fettuccine') or vague ('pasta'). On the next page, 6-10 search results would each be presented in a vertical DIV (like a shelf of books). Instead of each click holding up the whole process, each DIV/IFRAME could be dealt with independently until resolved/ordered. The more specific items could be resolved and ordered quickly. The vague ones might need a bit of drilling down. Any resolved (ordered or cancelled) item would be replaced by the next thing on your list.
Partner that with shortcuts to things you order regularly and so on and I think it could speed up online shopping somewhat.
The closest I've seen to this is Aptana Studio ( http://www.aptana.com/ ) , which is far from what you suggest, but is at least is on the same track.
> Flashy feature is showing structure of html pages (block-level elements) in 3d, where each block is separated along Z from the previous one, so you can rotate the page and understand its structure better.
Far from flashy, this would be AWESOME.
The URL is http://www.ctwug.za.net/ .
The news recommendation one is one that I actually discussed with someone only yesterday, with a very concrete application in mind (finance business).
Calculator sharing: maybe a layer on top of google docs?
minipapers is cool, but lots of science is really hard to get without sufficient background. The jargon alone can make a 1 page summary in to a 10 page explanation.
Food shopping suggestion: a simple checklist of the stuff that you normally would buy, then try to anticipate the frequency with which you order things to preset the checked items.
This is a problem -- but an important, widespread, interesting problem, where an innovative service with carefully-tuned UI could make a major difference. e.g. create interfaces that can expand/contract their level of detail according to the user's background. A term can be expandable into a one-sentence definition, and/or linked to external explanations. A sentence can expand into the corresponding paragraph from the article. many more things along similar lines -- none earth-shattering, but good execution could get you a nice market.
This is exactly where I am going with job4dev.com. My progress has been slow this summer, but I'm planning to get the resume sorting and interview tracking done by the end of the year.
The triage is still in very early form. While most of the Brazilian listings are posted by actual users, we are still trying to break out of the chicken-and-egg problem everywhere else around the world.
To do that, we are pulling job listings from different sites. However, there is some "post-processing" to be done:
- we need to get information about the company (we only publish job listings from companies we know about).
- We need to add tags to the company page and the listing page.
- We try to break the job description into two parts: one describing what is expected of the candidate, the other describing what the position is about.
So, "triage" is some sort of post-processing queue. We are doing to get these job listings from other sites and to put in our format. We are automating it as much as possible, but there are always some detail that needs to be done by hand.
If you create an account, you'll see that we barely put any copy to explain how it works. I'm just dog-fooding for now. As I find more ways to automate or semi-automate the process, the better this triage system should work.
I don't want to put too much effort on it, though. Even if this system gets to be perfect, we would be no better than any job aggregator like indeed or simply-hired. I'd rather have companies coming directly to our site because we are offering some different service (the hiring process management) than by being a "better aggregator".
distributed ISP using long-range wifi and mesh networking -- technology should be just about there, and there are plenty of papers on "incentive compatible routing" so you can automatically pay people fairly for setting up efficient routers. Compete with Comcast etc.
frugality helper, fun financial planning software that helps you reduce spending, maybe through social / achievement-style means (give points for making dinner at home)
news recommendation (worked on this for a while; lots of ideas for making it work, but I needed a really good realtime news recommendation algorithm)
internet food shopping redux (I hate food shopping, and would love to do it over the internet, if it wasn't even more painful than in real life)
minipapers: for a given field (e.g., computer science systems), every day the site posts a new academic-level paper that can be read in 15 minutes by someone moderately knowledgeable in the field. So not a long paper, but one which presents a new idea in accessible form. The site is for "casual academics" -- people who like following research, but don't necessarily work in academia and don't have time to read full-length papers.
statistics for the non-statistician: most people suck at stats, but they'd like to know whether a particular conclusion can be made from some data. Give them the tools to do so (plug data in, ask question) without requiring them to know what a chi-square is or a t-test (but help them learn it if they are interested). Explain confidence intervals etc. using plain language.
calculator sharing: allow users to create simple forms which evaluate a simple formula (temperature conversion etc) and share them on a website with others. In other words, a one-dimensional spreadsheet with sharing functions built in. iPhone app would be important here.
ratebike.com: there are no good bicycle rating sites. I registered this domain intending to do something with it, and haven't yet.
webapp IDE: a full-featured html/javascript/css IDE for building webapps. Has a model of what content is static and what is generated; fills generated content with sample data; allows you to break one page into many template files; understands enough Django/Rails markup to recognize foreach loops and so on. Flashy feature is showing structure of html pages (block-level elements) in 3d, where each block is separated along Z from the previous one, so you can rotate the page and understand its structure better. CSS & JS edits take effect as you type.
Fixing technical recruiting: job postings suck; people seem to get good jobs through referrals. Recruiters make a lot of money. It seems like there is room for an app which either helped recruiters or did the job for them. Perhaps this can manage hiring flow on the company side too: tracking, sorting and archiving resumes; managing the interview/hiring process; allowing interviewers to keep and share notes on candidates.
"Greasemonkey deployed": for freelance web consultants, if you have a good prospective customer, you might want to show them a demo of your work on their website. Right now if you want to do that, you have to download their site, rehost it, and make your changes. This site would instead allow you to rehost a target site (via proxy) and insert arbitrary javascript on the rehosted version. That way the consultant can say "Here's how your website would look with my work", and it can be tested live.
writing helper: smart grammar fixing, catches & fixes most errors that people often make. has an api, so you can feed comments on your site through it. (Don't know if this is hard, but it seems like it wouldn't be too bad, especially if it were conservative.)