I have come across some entrepreneurs who have created around 90% of their product/service but then decided not to go ahead with it for any reason.
Do you have any such plan/service/product which you probably created but have no time to pursue it?
Mention it here so that others might adopt it and take it ahead.
That said, a couple of things I toyed with, in case this sparks a useful thought for somebody else:
1. A decentralized, federated social-networking platform that would allow individual sites to include social-networking features, but where all the "social stuff" could be federated through API's, so that a user had access to his/her entire social graph, no matter which site he/she was logged into. This was a semi-novel idea 3 years ago, but now everybody and his brother is writing a decentralized social-network. <shrug />
2. This has maybe be done, not sure, but I was intrigued by the idea of a robotic avatar that could be used for remote conferencing. That is, a small robotic device with a camera, microphone, and speakers, where the head could pivot to scan the room and look up and down, etc... the idea being, if I'm hosting a teleconference in my conference room, and 3 people are "virtual" attendees, I plug 3 of these things in, and the remote conferencing software attaches a session to 1 per remote user, and lets them interact with the other people. Didn't really put much thought into what it would look like or what other functions it might have... but it wouldn't be fair to let the virtual attendees shoot at the others with frickin' lasers, if the others can't shoot back somehow. Which leads to the idea of a series of satellites, in geostationary orbits... wait, just kidding about that last bit.
3. I think there was another one, but I forget... I'll edit it in later if I remember. :-)
Create a "Taste Genome" comparable to Pandora's "Music Genome" and let people select which foods they like, then recommend foods with similar taste profiles. (foodgenome.com appears to be an abandoned attempt at this.)
Alternately, let people enter the foods they like. Using an algorithm similar to Facebook's suggested friends one, treat the food as friends. If a lot of people who share your taste also like something else, offer it as a suggestion.
This would help picky eaters (e.g., me) venture out and try new foods.
Along the similar lines, I was thinking of making a site that would take uploads of your grocery receipts (via iPhone app, or SMS or something) and use OCR to scan and determine which ingredients you've purchased. Then it could make a database of items you have in stock and recommend recipes based on the ingredients you have (and recipe ratings).
Also, it'd be able to determine which ingredients were likely to spoil soon, and thus suggest meals which make use of those ingredients.
My conclusion after a long while surfing recipe websites is that there is really no shortage of good recipes online. Also, receipts can be long and require more than one photo which would be a pain in the ass.
I do think the receipt concept has something to it, though. It'd be amazing to keep tabs on everything I own.
Automatic Toilet Handle Jiggler
There is a total of 1 prototype in existence in my parents basement. I installed it when I was ~14. It consists of a flat piece of spring steel about 5 inches long with a weight at the end and a clamp for attaching to the handle of the toilet. No more leaking toilet. No more getting back up to go jiggle the handle again. No need to spend the 15 minutes and $20 to actually fix the leaky toilet-- you know, because that's "hard." Slip on a springy weight and disaster averted!
I had an idea of building of an API data service-programming-language-cloud-management-tool-of-awesomeness that maps amazon cloud services to language primitives. Imagine using S3, SimpleDB, and SimpleQueueService as language primitives (the SimpleQueueService really works well when sending a closure to the queue).
I had most of version 2 working too (and a really good start on version 3), but I had to kill it. :(
I am in process of learning Django now (which is awesome) and I might do something on those lines in near future.