Hacker News Reader

Comment by Alex3917 | Hacker News Reader

I've been messing around with something similar, specifically though focused on tracking strengths and weaknesses in US infrastructure:

http://www.alexkrupp.com/Citevault.html

It's actually a lot easier than you'd think, because thanks to Zipf's law pretty much every article in these evergreen topic areas is using the same set of 1,000 or so facts. And these facts mostly come from the same set of government or NGE reports, which are updated at most once a year, and often only once every ten years.

The cool thing is that you can then use a javascript snippet to track which facts are being used in which documents, automatically mark facts as outdated when they change, etc.

View on HN

Thats a really great idea! We definitely need some primary source of truth to refute all this half-wisdom thats going around, and of course to self-educate.

Not to nitpick, but I looked up a single random article, namely http://www.alexkrupp.com/Citevault.html#parenting, and Wikipedia says its not that simple:

"An attempt to replicate this study in 5934 8 year old children failed: No relationship of the common C allele to negative effects of formula feeding was apparent, and contra to the original report, the rare GG homozygote children performed worse when formula fed than other children on formula milk.[5] A study of over 700 families recently found no evidence for either main or moderating effects of the original SNP (rs174575), nor of two additional FADS2 polymorphisms (rs1535 and rs174583), nor any effect of maternal FADS2 status on offspring IQ.[6]"

Source: http://en.wikipedia.org/wiki/FADS2

Good catch, that's exactly why this needs to be open sourced!

I actually looked at that wiki article, but probably before that was added. And that's a good example a fact that had some editorializing on my part, which I ultimately want to eliminate. The original idea was to give context to facts, but I think it would make more sense to create a metadata format to do this, e.g.:

- To understand this fact, you need to understand X, Y, Z.

- This fact is needed in order to understand facts X, Y, Z.

This way facts can live and die on their own, rather than being bundled together according to the tastes of a single curator.

I actually registered findmeaning.in to be the domain for a website I plan to build. Happy to accept partners! If you can find my email at http://qbix.com/about then hit me up.

The website was originally supposed to be an annotation site like rapgenius - and I have already spoken to people at LyricFind who would give me a license to use their entire database of songs!

But then I thought it could go much further. First of all, you could release a bookmarklet and browser plugins that would let people mark up websites for others to see. Each such mark would have its own associated discussions etc.

But we could go even further. We could source facts and debunkings of claims. We could in fact build a graph of dependencies between claims and give people a way to assess their truth value. In this way we could for example have a discussion of varios religious claims once and for all, and any visitor would see what other claims they depend on. I thought, for instance, that Coca Cola invented the modern image of Santa Claus as wearing red. I was wrong.

Snopes, Politifact, etc. could all be sources. Something like facebook posts could definitely use this bookmarklet.

So do you want to findmeaning.in/The-US-Constitution? Or findmeaning.in/Pokerface or findmeaning.in/Book-Of-Jubilees?

If you want to help me build it, drop me a line.

PS: Also I find it really valuable to have a followup after a news story has faded, eg to see if that kid facing life in jail for hash brownies got off (he did).

Genius.com meets first SERP in a meta reincarnation of hypertext. It's so esoteric that seems like such a good idea but requires so much definition and scoping in execution. Good luck :)

This is amazing! I might be able to help get more media people to use it. I'm on of the founders at Beacon. Right now we're focused on helping journalism projects get crowdfunding but I'd love to chat more about what you're doing! Email me if you're game: adrian@beaconreader.com

Thanks, I'll send you a note.

Bookmarked. Editing nit in the FAQ.

  C.f. comments on Peter Thiel's graph of the year...

Should be "Cf." -- also I believe "cf." should be read as "compare", so really it shouldn't even be that (see http://en.wikipedia.org/wiki/Cf.)

aaaaaaand no one cares.

I didn't know about the English equivalent, so I did care.

Thanks, updated!

Wow, this is.. Something else. I will definitely suggest this as a possible master thesis research topic at my work place.

Thanks! Yeah I spent a couple thousand hours on it, but for whatever reason have gotten pretty lukewarm feedback from various media organizations I've pitched it to. I think the power to make conent creation 10x cheaper while vastly improving quality and total derived value is pretty self evident, but go figure. So for now it's been fairly dormant. I think the sections on antibiotic resistance and arrest statistics have both changed slightly since I last updated this, but everything else should be pretty current.

It's never easy to get someone to change their behavior. When you pitch them, all they hear is the pain of changing platforms. Your assumption that media companies actually care about better quality, is also flawed. They care about sensationalism.

the people want to be entertained, and unless you can show how this entertains them, you'll have a hard time getting past their resistance to change.

Sounds to me like it would be an awesome resource for mathematicians. It's just a different brand of information, but many people would love to have a useable proof reference and knowledge dependency system. Especially if it can be audited and contributed to easily.

Pretty sure the major news services handle this already through their CMSes. Editors can cut and paste "story blocks" that are updated over time. Unfortunately, many news agencies don't do actual research or fact-checking anymore, so your implementation probably failed because it targeted the way news did business 20 years ago, not the entertainment-driven news media we have today. You probably also didn't understand enough about their workflows and CMS systems.

I can't be bothered to look for it right now, but the NYTimes did a behind-the-scenes look at their CMS probably a year ago.

EDIT: Ok, never mind, I looked for it: http://open.blogs.nytimes.com/2014/06/17/scoop-a-glimpse-int...

I'm familiar with the scoop functionality you mention, and my last job involved making APIs to deliver content to media CMS, so I have some general familiarity with the space. Obviously infotainment will always be a large portion of what the media produces, and this isn't really relevant to that. But once you get beyond that type of content, media organizations aren't actively trying to produce low quality articles, it's just that that's what they're incentivized to produce. If you can make it less expensive and more profitable to produce high quality content, then that's what will get produced.

Also there was never really any implementation that failed per se, I actually still think this would work. I'm just currently spending my time on other stuff that I think has more potential. For the last several years this has been kind of a hobby project that I work on here and there whenever I get burnt out on whatever I'm supposed to be doing.

The main problem with it is that it just takes an enormous amount of time to add new content. Any given fact can easily take an entire day to add... By the time you go through all the competing claims, find the the primary sources, read through the methodology for each one, etc., that's easily 8 - 10 hours per fact. Which is why there still isn't any code or UI even after multiple years of work.

This really seems valuable, you are on the right track and should keep pushing with this project.

Awesome!

This is awesome! I didn't notice a license—is there one?

Good call. This is the first time I've shared a link to this publicly, so I haven't put much thought into this yet. A lot of the quotes, especially in the sections about healthcare and pharma, come from copyrighted books. So there might be some issues with allowing people to redistribute this entire thing wholesale, especially for commercial use. But for now you only need to cite the primary source for any given fact. At some point I'll dump this on citevault.org, and I'll have to put some thought into licensing issues I guess.

Reply on news.ycombinator.com