Show HN: Find members of Congress with similar voting records
congressbuddies.comIt's not symmetric! None of Henry Waxman's friends are friending him back, but they're mostly in a clique with each other.
Lonely little CA-33, nobody understands you...
I'm not sure I get how it couldn't be symmetric? Is there an explanation or I guess I don't get the definition of like.
It's based on how similar someone is to someone else and some people have closer matches than others. Fictitious Example where your listing top 5:
Thus Adam links to Bob as his #1, but Bob does not link to Adam in his top 5.Adam and Bob are a 80% match. Bob and Ted are a 90% match. Adam has 1 match better than 75%, but Bob has 5 matches better than 85%.yep! for a great example, try out Rush Holt (http://www.congressbuddies.com/?name=Rush%20Holt)
None of his top-5 "buddies" have him anywhere in their own top-5 lists, which is pretty unusual.
Cool! I would love to be able to "vote" on a couple (dozen) bills and see who I vote like.
This is fantastic. We need more sites like this. There are so many simple / enlightening / controversial patterns that are just waiting to have some light shed.
I've always thought we need a Congressperson "report card" that just summarizes their voting history. I mean it's really the only thing that matters about them.
Say there are some important arching issue "dimensions":
- Collective bargaining
- Obamacare
- Education
- Gun Control
- Civil Rights
- Same-Sex Marraige
- Finance Industry Protection
- Tort Reform
- Raise Taxes on Rich
- Raise Taxes on Middle Class
- Lower Taxes on Rich
- ...
If each bill in Congress is tagged with relevant dimensions and placed somewhere for each, then a Congressperson's record can be grouped on issues... That's what I want. I think the only thing stopping anyone from doing this is having this schema and maintaining the record of where each bill stands on all the important issues. Figuring out how to collate that information is a challenge, but it's definitely doable. In fact I think there are a lot of trustworthy people and orgs willing to pitch in on that. Personally I wouldn't even care about bills very far into the past if we were only capturing this sort of schema for forward-looking data so it can be useful in the future.
I wonder what Taubere / Govtrack.us thinks about that sort of information and how it can be organized. I'm sure he has ideas about it and could give insight. Obviously any analysis beyond yae/nay records has an inherent partisanship, if even slight, which I assume is part of the reason isn't on Govtrack, but he has to have thought about how those kind of data fit to the govtrack schema.
Awesome idea: Let's make it pleasant and fun to understand how your congress reps really vote!
Historically, these kinds of things don't get much interest because of the lack of money that can be generated.
That's why crowd funding is exciting. I'd love to see this project as a kickstarter.
A quick search doesn't show any sorts of projects like these already existing. I wonder the reason.
Edit: while kickstarter doesn't have these projects (against the rules?), indiegogo seems to.
For instance, an app that allows you to vote alongside your congressman.
http://www.indiegogo.com/projects/capitol-bells-mobile-congr...
Thank you for your well written feedback. You clearly captured our initial motivations for starting congressbuddies with the line "...patterns that are just waiting to have some light shed." Exploring "dimensions" while making it accessible and informative from a UI and data crunching standpoint is clearly our next step. Furthermore even though we are open source, we want to make it transparent from an enduser's perspective regarding how this data is being computed all the way from the raw json to how a scale like the one you've mentioned is established. Our ultimate goal throughout this process is to raise awareness about our representatives and make it informative through a highly visual and graphic manner.
I'm British, but I thought I had seen something along these lines for the US. Here's the UK version: http://www.theyworkforyou.com/mps/
I took a Coursera course this summer on linear algebra which showed how to compute this in one of the labs. It gives you a data file with each of the votes of each of the senators in the 109th congress, and you use vectors and dot product for determining most/least similar voting, mean dem/repub voting, bitter rivals, and whether John McCain is really a maverick.
This was probably one of the most fun labs in any course I've ever taken. The course materials are still online: https://spark-public.s3.amazonaws.com/matrix/politics_lab.zi...
EDIT: Forgot to mention the course material is all in python. It's worth checking out.
I get a 500 Internal Server Error on Congressman Raúl R. Labrador (R-ID).
Looks like it's a unicode problem, because, if I manually fix the URL to eliminate the accented character, it works:
should be fixed now, thanks for letting us know!
Cool idea, it would be nice if you could provide a little bit more information about why the results are the way they are, or the percentage that they match, maybe recent bills that they agreed on. As someone from Canada I have no idea who these people are or why the relationship would be expected/unusual.
Thanks for the feedback. We plan to add a relative index to represent how closely a congressperson votes with respect to one another. The main idea of the map is to show geographic voting trends among congress members. In the future we want to expand on the recent bills idea you gave and provide an in-depth analysis of the different bills people voted on.
I'm additionally interested in who never vote together. Do some Congressmen just never agree? Are there aggregate patterns here that change over time [e.g. partisanship]? I believe there's a ton of interesting data in this area, and this is a good starting point.
Likewise, how does the voting structure and people's votes differ from their norm when voting result is near break-even. E.g. 7/6 split. I feel there's a lot of media and public perception management going on, agreed upon by parties or those in who have majority playing games OR worse those who are playing double-agent, mostly voting for "their" side, but when it matters, switching over to their real team / beliefs / friends / money-buddies (lobbyists wants). Would too be interesting to match up the data with who all received money, and how votes look based on industry, and specific companies, etc..
Thanks for the comment. I agree it's a good starting point but theres a lot that can be done. One plan we wanted to implement was a filter to match congressman based on different topics they voted on. Ex funding, etc. For aggregate patterns we would have to look at prior voting records as well, but it seems like a great idea. All our code is on github, and we'll slowly list all the features we want to add on the wiki there.
Implementing what I wrote in comment https://news.ycombinator.com/item?id=6575348 would give great insight too. I'm sure a lot is blatantly obvious, but visualizing data is always helpful to reenforce it in someone's mind.
Leaderboards, all of this needs to be in sortable leaderboards, front and center...
Great work so far!
Cool tool, but the search interface could use a little work. Some sort of autocomplete would make it easier to find congressmen by name (I typed "Bernie Sanders" when I had to put in "Bernard Sanders").
I'd like to see a principal components projection. Are there clusters? The buddies are interesting, but it seems like there's more here! :-)
my feeble mind does not understand "See who votes like who in congress". there is no explanation of the site other than those few words.
You guys finished it! Come a long way since the FB hackathon.
P.S. Love the title
thanks! Yea it's gotten a lot better looking (and hopefully more stable) since then.