brain dump
I have a huge number of projects going on in the background. Here's an attempt to write one sentence about each of them. In no particular order:
- I've been trying to find more data sources for spf-all.com, but haven't had much success getting zone files out of registrars.
- I've been working on a Mandelbrot set viewer that navigates like Google maps, but currently still has bugs.
- I want to port xearth to OS X, as a way to try Cocoa programming.
- I'd like to implement SRP in Javascript for my OpenID provider.
- I have many projects I'd like to investigate in the accessibility world, eg. NVDA (free screen reader), WebAnywhere (another browser-based screen reader).
- Implement a web service to convert New Zealand Grid coordinates to WGS84 (for locating DOC huts easily).
- Implement more web games along the lines of my Othello implementation, but with server side computer play.
- I was working on a utility to sort directories by filename on my MP3 player, but it ate the filesystem one day and I had to reformat the device, so it needs fixing.
- Investigate interesting programming languages: Fortress, Scala, Converge, Erlang, Groovy.
- Freebase is like Wikipedia but with structured data, with easy programmatic query APIs.
- Try to import the whole Wikipedia history into a git repository, if it's even feasible, and then try to do text attribution like a 'blame' function.
- This solid state wind sensor looks like fun.
- I've been working on making some reports on Wikipedia revert history, such as how long it takes on average to revert vandalism.
- vim-scmdiff is a Vim script that highlights source file lines that have been changed from a base version in your SCM.
- I think it's possible to break S/Key using a concerted distributed effort.
- Build a p2p version of my Othello game using GWT.
- Create a standard milter interface for pydkim.
- Build a Google Apps version of lnk.nu just to try it.
- My MacBook has a camera on the front; it should be able to tell what I'm looking at on the screen.
I don't think that is everything. Who wants to help?
Some cool games I enjoy and would be quite interested to see you make AI players for, as I think they are feasible software projects (i.e. I won't suggest Go here!):
TZAAR
Travel Blokus
Hey! That's My Fish!
Hive
DVONN
Rumis (if you want the user interface challenge of a 3-dimensional game)
Gipf
Trax (game designed by a fellow kiwi!)
Kogworks
Quinamid
OK, I'll stop there. I was going from my most highly rated reasonable-sounding games downward from my BGG collection
"reports on Wikipedia revert history, such as how long it takes on average to revert vandalism." - that sounds cool and interesting and potentially useful!
TZAAR
Travel Blokus
Hey! That's My Fish!
Hive
DVONN
Rumis (if you want the user interface challenge of a 3-dimensional game)
Gipf
Trax (game designed by a fellow kiwi!)
Kogworks
Quinamid
OK, I'll stop there. I was going from my most highly rated reasonable-sounding games downward from my BGG collection
"reports on Wikipedia revert history, such as how long it takes on average to revert vandalism." - that sounds cool and interesting and potentially useful!
![[info]](http://stat.livejournal.com/img/userinfo.gif){kind=small}
ghewgill
2008-08-28T08:02:14Z
Thanks for the list. On that note, I saw this clever little game yesterday and thought of you: http://www.chucksmith.de/?p=14
I'm quite a ways along with the revert history project, but haven't properly collected or analysed my results. One challenging part is handing the incredible volume of Wikipedia history - as of last January the whole thing including all past revisions is 2.8 TB (18 GB compressed)! Would you believe that the median time to revert changes is 64 seconds?
I'm quite a ways along with the revert history project, but haven't properly collected or analysed my results. One challenging part is handing the incredible volume of Wikipedia history - as of last January the whole thing including all past revisions is 2.8 TB (18 GB compressed)! Would you believe that the median time to revert changes is 64 seconds?
goulo
2008-08-28T12:17:59Z
Heh, when a2na and I play Quarto, we generalize it to win by getting the points of any square, even at tilted angles like that!
myelin
: Weather2008-08-28T03:28:28Z
That wind sensor is cool. I wonder how good a weather prediction system it would be possible to make out of all the publicly available data. weatherunderground.com lists a bunch of local personal weather stations:
http://www.wunderground.com/cgi-bin/findweather/getForecast?query=christchurch,%20new%20zealand&wuSelect=WEATHER#PWS
http://www.wunderground.com/cgi-bin/findweather/getForecast?query=christchurch,%20new%20zealand&wuSelect=WEATHER#PWS
ghewgill
: Re: Weather2008-08-29T12:14:46Z
Although there's lots of local data, the real prediction data for NZ would have to come from ocean based sensors, I suspect. There's so much stuff that local sensors wouldn't even see until it's here.
With that amount of data though, you could make some interesting plots of how the temperature varies around the city. I saw a fairly technical book the other day called something like Weather Patterns of Australia and New Zealand, which had a whole chapter on very local weather patterns in Christchurch. They had plots of temperature, pressure, humidity, etc all overlaid on a city map with the CBD and built-up areas indicated.
The only bit of weather sensor data I have so far is at http://tempmon.hewgill.com powered by a QK145 and some Python scripts.
With that amount of data though, you could make some interesting plots of how the temperature varies around the city. I saw a fairly technical book the other day called something like Weather Patterns of Australia and New Zealand, which had a whole chapter on very local weather patterns in Christchurch. They had plots of temperature, pressure, humidity, etc all overlaid on a city map with the CBD and built-up areas indicated.
The only bit of weather sensor data I have so far is at http://tempmon.hewgill.com powered by a QK145 and some Python scripts.
paradox0220
2008-08-28T16:07:35Z
"Try to import the whole Wikipedia history into a git repository, if it's even feasible, and then try to do text attribution like a 'blame' function."
I've been thinking seriously about this kind of thing myself. Still trying to organize my thoughts on what exactly I'm trying to accomplish though (and how to take small steps rather than trying to implement the world). This won't do it justice but here is a short stream of random thoughts:
News is biased. People argue past each other all the time. There needs to be a tool that places information in an format that allows for marking which are facts (with references) and who put the facts (or lack of facts) there to gain/lose reputation. Then trends can be formed as facts change which may even enable to nice prediction forecasting. People can view the high level outcomes of these discussions in a very brief format or dig down to review the facts for themselves (and possibly contribute).
Imagine reading a CNN online news report and phrases are color coded from green (high verified through references) to red (wrong, misleading or out of context).
I actually thought of some interesting ways to get people interested in contributing and funding etc.
Whatcha think?
I've been thinking seriously about this kind of thing myself. Still trying to organize my thoughts on what exactly I'm trying to accomplish though (and how to take small steps rather than trying to implement the world). This won't do it justice but here is a short stream of random thoughts:
News is biased. People argue past each other all the time. There needs to be a tool that places information in an format that allows for marking which are facts (with references) and who put the facts (or lack of facts) there to gain/lose reputation. Then trends can be formed as facts change which may even enable to nice prediction forecasting. People can view the high level outcomes of these discussions in a very brief format or dig down to review the facts for themselves (and possibly contribute).
Imagine reading a CNN online news report and phrases are color coded from green (high verified through references) to red (wrong, misleading or out of context).
I actually thought of some interesting ways to get people interested in contributing and funding etc.
Whatcha think?
ghewgill
2008-08-29T12:23:45Z
I actually conflated two ideas into one. I think my first idea was to take the raw Wikipedia data and build an attribution browsing interface, where you could see who wrote what text when. Then I thought of using git to help, which it might or might not.
I hadn't extended my thought to cover news, but that's a cool application. The advantage of wikipedia data is you'd have the entire article lineage immediately available, where the challenge for news is you don't always know what the sources really are. With news you could compare articles across outlets, identifying common bits and doing something with them. I imagine this is the kind of thing that Google News already does to consolidate stories. Actually it might be more interesting to do something with the unique bits that only show up in one or two outlets!
I hadn't extended my thought to cover news, but that's a cool application. The advantage of wikipedia data is you'd have the entire article lineage immediately available, where the challenge for news is you don't always know what the sources really are. With news you could compare articles across outlets, identifying common bits and doing something with them. I imagine this is the kind of thing that Google News already does to consolidate stories. Actually it might be more interesting to do something with the unique bits that only show up in one or two outlets!
paradox0220
2008-08-29T14:55:54Z
News was just an example (and probably not the most feasible first implementation). I agree that having all the data local for the first phase is the way to go like Wikipedia does.
I think something like this needs a very specific focus to get interest and critical mass and then it can expand out. For example, it could track the funding trail from companies to lobbyists to politicians to sudden law changes. Or maybe it tracks the world economy (imports, exports, statistics etc... CIA web site has some good stats to start with but no good trends or historical information; only a snapshot that I've found so far).
At any rate, the heart of it is reputation building and reference based facts database building to make informed predictions or show past trends. I think that is the part wikipedia missed out on. I'm afraid there is a patent mind field looming.
Still thinking this through but if you are interested I have a voice over IP server set up that we could chat on for free (Ventrilo). Been pondering getting a nice colocated server if I can get a bit more firmness in some of my ideas to start working on them.
I think something like this needs a very specific focus to get interest and critical mass and then it can expand out. For example, it could track the funding trail from companies to lobbyists to politicians to sudden law changes. Or maybe it tracks the world economy (imports, exports, statistics etc... CIA web site has some good stats to start with but no good trends or historical information; only a snapshot that I've found so far).
At any rate, the heart of it is reputation building and reference based facts database building to make informed predictions or show past trends. I think that is the part wikipedia missed out on. I'm afraid there is a patent mind field looming.
Still thinking this through but if you are interested I have a voice over IP server set up that we could chat on for free (Ventrilo). Been pondering getting a nice colocated server if I can get a bit more firmness in some of my ideas to start working on them.
ghewgill
2008-08-31T01:54:11Z
You mentioned a "facts database" and how that's not Wikipedia - you might be interested in http://freebase.com which is a community editable highly linked database. I haven't played much with it yet, but I used it the other day to see whether there has been any women vice presidents of the US (answer: no, only candidates). It was surprisingly easy to find what I was looking for.
For voice chat, I'm using skype these days, it works great. I just had a quick look at Ventrilo and the idea of having spacial representation of participants in a group chat is interesting!
For voice chat, I'm using skype these days, it works great. I just had a quick look at Ventrilo and the idea of having spacial representation of participants in a group chat is interesting!
2008-08-27T11:52:19Z