Ask HN: How to Make News Aggregation Work in 2024?
6 comments
The word/sentence bias should work if you train an LLM to it I would say. You can only feign enlightened centerist and allow others to get away with things for so long, at some point you reveal your cards.
But the breaking story contrast method I find people using like newscord[1] is doomed to fail, they all omit in the same way and arrange facts in ways that mislead by altering a sense of timeframe or magnitude or in countless other ways that can barely be pinned down to one comma. Reputation is still a good enough measure but you have to be a reader with standards willing to walk away when they drop or go negative. Most consumers are habitual FOMO types though. When journalists collective response to events cause them all to tarnish their reputation in the same way we need to remind ourselves that they are not special people, just people we gladly relinquish the effort to collate happenings for us whatever their own unusual desires.
[1] https://newscord.org
But the breaking story contrast method I find people using like newscord[1] is doomed to fail, they all omit in the same way and arrange facts in ways that mislead by altering a sense of timeframe or magnitude or in countless other ways that can barely be pinned down to one comma. Reputation is still a good enough measure but you have to be a reader with standards willing to walk away when they drop or go negative. Most consumers are habitual FOMO types though. When journalists collective response to events cause them all to tarnish their reputation in the same way we need to remind ourselves that they are not special people, just people we gladly relinquish the effort to collate happenings for us whatever their own unusual desires.
[1] https://newscord.org
>The word/sentence bias should work if you train an LLM to it I would say. You can only feign enlightened centerist and allow others to get away with things for so long, at some point you reveal your cards.
When I checked out that one, it worked quite well for the USA but it failed horribly for Canada. This was years ago, llms have advanced a great deal and no doubt that product.
I dont particularly blame the project, Canadian journalists are pretty fake. I mentioned the freedom protest that the courts have officially declared was unconstitutionally crushed. The judge himself literally said that on the day it happened he sided with the government and thought it was necessary.
>When journalists collective response to events cause them all to tarnish their reputation in the same way we need to remind ourselves that they are not special people, just people we gladly relinquish the effort to collate happenings for us whatever their own unusual desires.
The journalists will so often write as if what they are writing is the undoubtable truth and they ruin themselves.
When I checked out that one, it worked quite well for the USA but it failed horribly for Canada. This was years ago, llms have advanced a great deal and no doubt that product.
I dont particularly blame the project, Canadian journalists are pretty fake. I mentioned the freedom protest that the courts have officially declared was unconstitutionally crushed. The judge himself literally said that on the day it happened he sided with the government and thought it was necessary.
>When journalists collective response to events cause them all to tarnish their reputation in the same way we need to remind ourselves that they are not special people, just people we gladly relinquish the effort to collate happenings for us whatever their own unusual desires.
The journalists will so often write as if what they are writing is the undoubtable truth and they ruin themselves.
No clue, because the general quality and incentives of the news business is so terrible.
I enjoy Techmeme[1] and sumi.news[2], though.
1. https://techmeme.com/river
2. https://sumi.news (the creator, Alex, is on HN somewhere)
I enjoy Techmeme[1] and sumi.news[2], though.
1. https://techmeme.com/river
2. https://sumi.news (the creator, Alex, is on HN somewhere)
For personal use:
What works is a good plain old rss.
Delivered to my client of choice, via gui or cli. Skipping ads and clickbait articles to save me time.
With a wave of generated content flooding some automated systems, the best curation will be done by yourself by finding reliable sources to subscribe to.
What works is a good plain old rss.
Delivered to my client of choice, via gui or cli. Skipping ads and clickbait articles to save me time.
With a wave of generated content flooding some automated systems, the best curation will be done by yourself by finding reliable sources to subscribe to.
When Google News did their major redesign (back in, what, 2014?) it removed pretty much everything about the service that was useful to me.
I switched to curating my own news feeds from RSS feeds of various sources. It turned out to be the best thing I could have done and now is more useful to me than any third party news aggregator that I've seen.
I recommend this very much.
I switched to curating my own news feeds from RSS feeds of various sources. It turned out to be the best thing I could have done and now is more useful to me than any third party news aggregator that I've seen.
I recommend this very much.
I knew the new industry was going to collapse. What more do you need from a new aggregator? Journalism was doomed since though.
I had tried https://ground.news/ but never stuck with it. I like the idea but it fails at dimension. There is no left vs center vs right in Canada. Politics in Canada and Europe is 3 dimensional at least with wormholes in our politics.
Worse yet, you can have a paper like globe and mail, seen as right of center. But does that mean the journalists are all right of center? No, not at all.
I tried a machine-ai news startup which attempted to judge authors/journalists on their perceived bias using keywords and such. With presumably the goal to see how biased the papers are in reality. Ideal newspapers would have a strong spectrum of diverse thought. However, it failed to properly assess bias. Overall, machine ai wasnt able to assess through sarcasm, lies, or counter speech?
I checked out the 'machine AI rewrites articles with maximum neutrality' but the bias remains or possibly just got worse.
I have thought on what the next big attempt to improve aggregators. I've seen several attempts from others, but I think what is important to do is some algorithmic approach that isn't on bias. It has more to do with post-analysis. Follow how stories really unfold in truth.
So for example, I would take the analysis where the government found that journalists outright lied about the freedom protest in Canada. Misleading the government into unconstitutionally crushing a legitimate protest. CBC's david cochrane or global's david akin and rachel gilmore should be dramatically punished for this.