Basically, I wanted something that would let you search for your high school friends in all the jails & prisons, state and federal. Optionally I thought about hooking it into facebook, for automatic updates.

"Ben Jerry has just been imprisoned for 5 years for aggravated assault!"

Hope someone doesn't steal my idea; then again, if they do that means I'd get to reap the fun benefit without the hard work!

Except that these stats won't change much...

Though, I've recently been contacted by a high school student who intends to implement a subset of the idea for some sort of school project. I'll be interested to see what he does with it.

If someone like Google were to implement this as a response to every email that had positively been identified as spam then the scammers would spend so much time trying to sort the wheat from the chaff that they would never get anything else done!

At the moment the signal-to-noise ratio is low for us, as spam receivers, but if we made the signal-to-noise ratio lower for the spammers and scammers then they would have a much harder time making money!

I consider this crazy because abandoning the development and having it just sit there seemed crazy to a lot of people. Pikluk was a damn great idea with a few paying customers who are absolutely in love with a tiny percentage of what actually got implemented.

But in the end, after exhausting some personal savings, we just could not find the resources to keep working on it and we lacked PR to sell it on eBay for a six-figure sum.

If its any consolation you did a very, very nice job on the interface.

I salute you guys! Without you the world as we know it would not turn. Now, I'm off to code in Ruby...

Another idea I had, which I thought was good, was building a Twitter like site to track what you were eating, and marry the concept with Reddit style voting and a collaborative filtering + Bayes classifier to find new things to eat. I had a working demo, but never tried to promote it. The idea of the Bayes classifier was to eliminate ingredients you strongly disliked. Not sure if what I did worked entirely correctly since I didn't amass enough users to get feedback.

EDIT: The actual link is working again. which is actually much more up to date.

It's sort of a pixel art discussion board. You would have to check it out to really understand it.

It was fun to build, but it would be hard to commercialize. Plus nurturing a community that is more interested in quality art than anonymous penis drawings would be more work than I have motivation for.

http://www.imageboard.net/921/slideshow/

I would like to see this kind of thing in the USA someday, but there are some pretty significant practical hurdles in the way. I think it's possible to design around the problems, but I just haven't had the time + energy to move forward.

1. Most people aren't equipped to think about the kind of issues that come up with legislation. There are subtle interactions between laws, small economic provisions can have huge market effects, and many sentences end up meaning the exact opposite of what they seem to say. Ideally (for me, though you may disagree), the person getting voted in is equipped by experience, education, and general smarts. I want someone voting who is studying law, proposed and current, and making informed decisions. I don't always get that, but I'm more likely to get that with one person than with many.

2. Many Americans are terribly prejudiced. Pick any poll you have handy showing that more than 50% of Americans believe in something ridiculous and imagine them voting on a law related to that belief. Historically, perhaps, this vision is dimmer. It's a cultural issue, not an absolute one, so if we get better as a people, this argument against will become less important.

I do like when these kinds of ideas are brought up, however. Lots of interesting talk happens.

I think that because of the way things run now we are also predisposed to think that other people will take care of the minutiae. The only problem is that in a direct democracy there is no 'other people'...

I think that a direct democracy is possible, but I think that the transition to it and the period afterward would be very rocky, possibly rocky enough to bring back the old system.

I like the idea, but I don't think you would ever get someone elected as congressional seats are supposed to represent the constituents of the state/district. Why would anyone in elect someone that is going to listen to all citizens vs those who they are stated as representing?

I was talking with a friend of mine who is a political analyst a while ago about a similar scheme... While sketching out a basic plan I realized how much work it would be and let it die...

- A JavaScript framework: Like jQuery wasn't good enough?

- A client-side search engine: Used AJAX calls to crawl a website and return results... Can you say slooow? Yuck.

- Hoppy: A simple little program that creates several new versions of itself, with names generated by a markov chain, then runs those new versions after a time delay. I never inflicted this upon anyone else, but trying to shut it down without re-booting was hell!

One time something interesting happened though. I noticed the latest copies had commented out the mutation part of the code. They had "evolved" to not mutate. I'm guessing that was the best strategy to make viable offspring.

My take away lesson was that mutation has to come from the environment for early life forms to advance.

But really, the source code of that would be really cool to see.

At least your script started out with a vaguely practical purpose in mind!

Predictably, this was a disaster.

In retrospect, I do not recommend Viaweb as an introductory Hello World web app. (I also do not recommend Java web programming in general.)

For me this was a potential solution to one of the fundamental problems with education: it is full of unchallenging, unrelated-to-life, contrived tasks. When any process (creative writing, math, coding) is harnessed to a fascinating, meaningful-to-students goal I believe you get motivation. And creativity.

So this service would just help mentors and apprentices in all sorts of fields find each other.

This idea was originally inspired by Dr. Tae's presentation:

"Building A New Culture Of Teaching And Learning: Why is skateboarding a better model for education than school?"

I wasn't able to find anything at mentorapp.org; has the hosting expired?

I'd love to talk about possibly collaborating on this (although I think it would be tough to monetize, so I see it as more of a "pro-bono side project"). My contact info is in my profile.

I will say that the time spent made me a better poker player and taught me some nifty things about AI and game theory to which I wouldn't have otherwise been exposed.

I attract a line of ants to eat the bait (in my case, cat food), then periodically suck (or blow) ants into an un-escapable garbage bag, but only a few at a time to keep the line of ants stable. (or until the entire colony is exhausted)

Compared to the traditional poison method, it is much more satisfying for me to dispose of a garbage bags full of ants which I might make good use of. The mass of corpses could be good fertilizer, or bird feed.

I simply hate the idea of chemical warfare to combat ants.

In the right mood, sometimes I think my current project is crazy, but I'm sticking with it!

I've been working on a CMS called Picl for the last 6 months or so, it's already in production with a few client websites. But every so often I look at it and go 'why? wasn't there enough CMS's already?'. I just have to remember to look at the original design goals and remember that it is still a good idea!

Allow people to report when they are victims of sexual assault. I started discussions with the local sexual assault center, but after hearing feedback, it seemed like people did not want stuff like "3 of your friends were inappropriately touched last night" popping up on Facebook. Although, I do still think part of the problem with sexual assault on college campuses is that nobody talks about it.

Has a possible business model too: rev share with legal teams.

Feel free to run with the idea!

- Filter than mounts to the front of your tv and polarizes in alternating directions based on a sync signal hidden in the channel that lets you do 3D tv at home for cheap.

http://svn.wonderwarp.com/tuplet/

Decided to work on something profitable instead.

Biggest one was a social networking site (which technically is still in my mind, but pretty dead) to compete with the big boys. In the end I was offered enough money for the work I had done on caching and some other bits that it was silly not to take it. :)

I'd love to have another go some day.

There was Qoppa which was intended to be an extensible Browser game server written in Python. That fell apart when there was no one much to work on the artistic side. http://hg.errant.me.uk/errant/qoppa (dont smirk :D)

What else. A quite-advanced attempt to write a distributed file store for really large write-once/read many files (2GB+) in Python. I got as far as implementing a REST interface and then got bored of the tedious work on the replication side.

A CMS/Blog tool specifically for magazines. Gave up when it turned into a massive amount of work and the magazine who tentatively were my first client decided not to "go online".

I could list a few more. My theory is just to keep plugging away at a few projects till one makes it. I was nearly there with the social network.

Currently Im planning to try and assault the blogging empire with a Wordpress competitor.

The idea was to make a reusable FOSS engine then build some commercial games on it :) the market is quite competitive, good graphics is becoming quite important.

I had already built the storage subsystem, indexing and single-column search ... when I found out that the size of data types differs across platforms and I had hardcoded the size throughout the database.

A "Post Secret" type of a blog that would allow people to anonymously own up to those times they broke the law and got away with it. Bad idea all around.

- a search engine (noseo)

- an operating system called younite (unix + dynamite ;) )

I'm not proud of any of these but they were fun to build at the time, even if the projects went nowhere I learned a lot from them. The search engine was done using 'leo', a literate programming editor. That alone was worth the experience.

In the course of doing all of that work I really learned the meaning of tree-spanning algorithms, big O notation, and soforth - but I never created a usable SYSTEM. The map data was just too huge; I only encoded a small part of it. Also, the amount of computation required grew so fast that I realized if I ever did succeed in encoding the entire map the amount of computational time to produce usable routes would be impossibly huge.

My grades suffered horribly and I found actual paying work writing software for an insurance company. There were other issues, too. Finally I had to just quit the whole thing.

In hindsight, had I had access to Tiger/Line (I think it existed then but cost far more than I could afford) and been more advanced as a programmer and had more (any) support the problem could have been solved. Tom-Tom, Google Maps', et al route-finding fascinates me. It's obvious that they don't do a complete graph search - I think they use clever tricks like identify arterial roads limiting the graph search to those roads - things like that.

The code looks like:

  var api = new AjaxAPI(
  "page = /testapi.php \n" + 
  "mydate= (^\\d\\d-\\d\\d-\\d\\d\\d\\d$) \n" +
  "int=(^\\d?+$) \n" +
  "text = (^.+$) \n" +
  
  "get_messages : GET /messages/project={int}/date={mydate}/ \n" +
  "get_comments: GET /comments/project={int} \n" +
  "save_comment:POST /save_comment?project={int}&comment={text} \n" +
  "delete_comment :POST /delete_comment/project={int}?comment_id={int} \n" +
  "echo: GET {page}?_method=echo&[foo={int}]&name={text}&[baz=(text)] \n");

  api.call.get_messages({project: 12, comment_id: 21}, function (request, response) {...});

It was fun to build and gave me a good look at Twitter oAuth and the APIs but I doubt it'll take it any further than it is already. I do use it on a fairly regular basis, as I'm kind of forgetful, and it does work pretty well so who knows.

I am optimistic that Google Chrome OS constitutes an absolutely enormous step towards #2 by building the actual Web OS -- which is more like the Palm Pre's OS than any other existing operating system. But I won't know if it actually works until it actually comes out, probably a few days to 2 weeks before YC apps are due, because Google usually launches about a week before Microsoft and Windows 7 comes out then.

Yakrazor.com: time management app. I can't remember exactly where I was going to go with it, but the basic idea was to bring some order to the process of Yak Shaving.

ToShove.com: provide a nice REST interface on top of Apple's push notification service.

CarZero.net: TSD rally site; mostly an online scorekeeping app. Still considering building this out with an iPhone-based front end.

SportsCarClub.net: basically WordPress for local car clubs, plus postal-code search for nearby events and clubs.

And then I have a half dozen more that I'm actually doing something with or seriously plan to do something with.

Started to gain some traction, but beyond a (very) few amazon affiliate sales, I couldn't really see how to effectively monetise it.

* welcome to the "anti-social" web

* http://www.suicidespam.com

* website caters to those who really hate the world, this life, and want to get out ...

* ... and to make the biggest splash they possibly can in the process

* website offers high-quality, expensive video/audio equipment, great high-power wireless/satellite IP services connectivity, and great battery life

* website customers purchase the equipement (they're about to kill themselves, they'll charge $15,000 to their credit card, no problem)

* they record their own suicide with the equipment

* beforehand they register 100, 200, 500 e-mail address of their dear friends, relatives, immediate family (intro, premier, and gold levels)

* the equipment transmits their suicide to central servers ...

* which then uploads HD video of their suicide to youtube.com ...

* ... and makes sure their 100, 200, or 500 contacts gets direct links to watch their suicide

* ... plus 500,000 additional random e-mail addresses

revenue potential is excellent ... suiciders don't care about debt, they'll pay anything!! this will be a viral product!! spread by word-of-death!!

i'd like to pitch this seriously at some VC conference. let me into y-combinator or tech-crunch or some other startup contest. i'm sure someone will take me up on the proposal!

I actually applied to YC with this some time ago. After I took this screencast, I even re-worked the UI to make it look all slick and web2.0-y. I abandoned it because it was too ambitious for a part-time project.

On the other side it's a win about cache locality, and even mmeory-wise when the object happens to be linked to a linked list most of the times (you win a pointer, 8 bytes in a 64 bit arch).

I wrote it in C++ and when launched with very downsized dimensions, it ate up 6 Gb of RAM and crashed with out of memory. I may try again when I can afford several hundred Gb of RAM.