Live data from Hacker News

Show HN: Govbase – Follow a bill from source text to news bias to social posts

govbase.com

81–90 of 94 posts

Re: Show HN: Govbase – Follow a bill from source text to news bias to social posts

#81
post #34

For context, this is a solo project I've been building over the past year while working full-time. I've been responding as "we" in the comments since I got used to doing it other places lol Looking for feedback and advice. I'm an engineer, not a journalist or policy researcher, so a lot of this domain is still new to me despite working on it for a year.

Hello,i don't mean to be dismissive but i think lawmaking definitely needs less AI and less pretendly-neutral summaries.

It's already a problem when researching a law proposal that dozens of news outlets will just copy-paste a bland summary of arguments from both sides, neither being explored fully.

I would recommend giving direct links to actually partisan information so people can situate the bill's intent and consequences in a broader context.

Where it helps:

- sometimes a bill's proponent is just an industry puppet whose talking points will be repeated in the media, but aren't solid enough to warrant a proper article… unlike opponents criticizing specific (for example deregulation) points

- sometimes, there are strong feelings and arguments on both sides of a bill and it makes sense to view them in their entirety; seeing one's side unhinged logic sometimes reveals more about the bill than the bill's text itself

- "same-side" opposition: sometimes a bill is perceived as "left-wing" or "right-wing" but receives opposition from the same side; for example, the democrat party is very divided on helping the rich vs taxing the rich, while the republican party is (less than 20 years ago) divided between hardcore authoritarian trumpists and libertarians defending civil liberties

All in all, i believe AI is a plight for society. We are only starting to understand the ecological and psychological costs. There are areas where machine learning can be useful (translation), but i strongly believe politics is not one of them. Please don't try to apply it to anything serious. Don't take it from me, take it from James Mickens in a talk where he explains both how ML works and how it related to the field of computer security:

https://www.usenix.org/conference/usenixsecurity18/presentat...

Re: Show HN: Govbase – Follow a bill from source text to news bias to social posts

#82

Earlier quoted context omitted.

The thing is that there is a voting process eventually so it can at least in principle be known which version is proposed by which legislator, or which committee.

The people behind changes aren't actually as attributable as it sounds though because amendment text gets collaborated on, so showerst might propose an amendment with the key parts of LPisGood's wishlist in it, and then the bill itself will die and then various parts will get cherry picked into an omnibus bill in 6 months anyway.

I've see similar stuff in open source a few times. X post a bug report, Y replies with minimized case, X makes a PR, Z makes a few suggestions, Y add some comments, X updates the PR and W merges it.

Re: Show HN: Govbase – Follow a bill from source text to news bias to social posts

#83
"An AI pipeline breaks each one down into plain-language summaries and shows who it impacts by demographic group."

What controls are in place for bias in this step? Most existing plaintext law coverage has bias in it, so how was it trained to convert the law to plaintext without that biased material?

Edit: Also, where it says it hurts or helps a specific issue, it would be helpful to understand why it was rated that way, especially if there was nothing in it's summary specifically stated about that issue.

Re: Show HN: Govbase – Follow a bill from source text to news bias to social posts

#84
post #75

The current summary on the home page contains bias / one-sided reporting. > While the administration describes the strikes as a necessary move to stop nuclear weapons, the conflict has already seen accidental friendly fire and threats of a ground invasion. The balance to the assertion "this was necessary" isn't "but there's been some consequences" -- it is an exploration of the truth of the assertion.

Given there is an "AI pipeline" in play, I suspect this is just the typical compulsive equivocation from an LLM. Never assert strong opinions. Find something to say while actually saying nothing. Always give "both sides" equal treatment and consideration no matter what the sides actually are.

"Always give "both sides" equal treatment and consideration no matter what the sides actually are."

It can't even do that correctly. Looking in the list of rights, it has some things called rights and others called policies - "Abortion Rights" vs "Gun Policy". Either call them both policies or call them both rights.

Re: Show HN: Govbase – Follow a bill from source text to news bias to social posts

#85
post #75

Earlier quoted context omitted.

Given there is an "AI pipeline" in play, I suspect this is just the typical compulsive equivocation from an LLM. Never assert strong opinions. Find something to say while actually saying nothing. Always give "both sides" equal treatment and consideration no matter what the sides actually are.

That's not a fair assessment. Context: I hate Trump as much as Khomeini. A "both sides" treatment would be: US & Israel illegally assassinate Iranian leader in bombing campaign, calling agression "necessity". Now, if you'd like to lean to one side or the other, you can either: - remove information about legality and the fact that they are the authors of the agression, add something about Iran being a threat to its ne…

You are so clearly biased your “both sides” is seething hatred of one side.

Re: Show HN: Govbase – Follow a bill from source text to news bias to social posts

#87

>shows who it impacts by demographic group. Why do we need to bake this zero-sum mentality into everything?

to be clear impact groups are things like government programs (e.g. snap benefits), disabilities, immigration status, etc. They're not "democrat vs republican". The whole point of Govbase is for users to be able to follow things that matter to them and see how the government is impacting their lives.

You can take a look at https://govbase.com/methodology for more info.

Re: Show HN: Govbase – Follow a bill from source text to news bias to social posts

#88

Hey nice site. The Ledge ( https://the-ledge.ai ) is a site I developed over the past year with more of a focus on bill content. Ultimately I found it useful to use all three LLM (Gemini, Claude, OpenAi) to ingest and dissect bill text en route to producing useful reports. Especially challenging for bills that run 1000+ pages.

super cool to see a similar project! I'll take a look. Yes I use a mix of Gemini and Claude for certain simple categorizing to full impact analysis.

Re: Show HN: Govbase – Follow a bill from source text to news bias to social posts

#89
Interesting. I've been just now exploring something related or tangential, but for Finnish governance (why not all countries eventually).

Idea: Pull in all related docs and statements from the law and its process. And analyze its mechanisms across all timeframes, and across all capital stocks including social/human/moral etc capital. (finding: no one understands game theory nor causal models apparently)

Re: Show HN: Govbase – Follow a bill from source text to news bias to social posts

#90

Earlier quoted context omitted.

That's not a fair assessment. Context: I hate Trump as much as Khomeini. A "both sides" treatment would be: US & Israel illegally assassinate Iranian leader in bombing campaign, calling agression "necessity". Now, if you'd like to lean to one side or the other, you can either: - remove information about legality and the fact that they are the authors of the agression, add something about Iran being a threat to its ne…

You are so clearly biased your “both sides” is seething hatred of one side.

Of course i'm biased (though probably not like you mean), but that "both sides" depiction was fair and rather neutral. I'm personally very happy Khomenei is dead, and so are my iranian friends. But we are all very concerned that he is dead for the wrong reason, under a wrong pretext, and with very grim perspectives (see also what the US did in all the countries it bombed in the past 20 years).

I think Khomenei and Trump are two sides of the same coin: bloody authoritarianism and religious zealotry. They're both pretty bad, but one side in this conflict was clearly the aggressor, and denying that is in itself picking sides. One can both sympathize with a victim of unjust aggression, and at the same time thinking they're a profound piece of shit.

One could even point out that just a few years ago, Trump was very insistent about "no more wars", and that he regularly mockingly predicted that Obama would attack Iran to avoid talking about domestic policy. Turns out the hypocrisy level is high and he really is beyond a doubt the bad guy in this story, even if that does not make the iranian ayatollahs good guys by any measure.

Post reply on HN