Hacker News Reader

Comment by smsm42 | Hacker News Reader

"Good information" is very subjective. Some people think Fox News information is good, others think it's all lies. Some people think Wikipedia is the closest thing to Word of God that we can get, others think it's a bunch of kids with too much free time on their hands writing about things they have no idea about. Some people think the US government is trustworthy, others think it lied so many times only an idiot can trust it again. Which information is "good"? If a hundred newspapers publish the same article because they are all owned by the same company which told them to - is it "spam"? If a hundred TV anchors all read the same message while pretending it's local news - is it "spam"?

I am not saying there's nothing to be done here - but let's not pretend it's easy and obvious and there's some objective way to see what's "good information" that does not involve a lot of human judgement and a lot of bias that comes with it.

View on HN

Determining which search results are good to return is the hard part about creating a search engine, and it’s definitely not a problem limited to politically charged content.

Having worked on search engines, I know it only too well. Defining "good result" is a very complex task, that even without politics intruding involves a lot of judgement and very complex and un-obvious choices. Is good result something people would frequently click on? Then clickbait would be the best search results ever - do we really want this? Is it something a lot of people search for? Is it a popular site? Is it a site belonging to a large advertiser? And when politics comes to it, it becomes a mess. So pretending it's clear and obvious - just remove "bad results" and show "good results" - is really not understanding the problem.

You are listing a bunch of methods that might be useful indicators of "good content", useful for search engine algorithms. But those are merely (possible) symptoms. "Good" is a quality of the content itself, made up of factors such as relevancy, accuracy, completeness, readability etc. Nobody said it was necessarily easy to judge these things, especially at scale. But that doesn't mean that it doesn't exist.

> "Good" is a quality of the content itself, made up of factors such as relevancy, accuracy, completeness, readability etc.

It’s impossible to quantify a rating based on these subjective qualities you listed. I think the point is you can’t remove the human element from ranking search results without leaving the search engine open to exploitation.

"Good information", almost synonymous with "truth", is not subjective. It is objective, observable fact. The fact that is a lot of disagreement, propaganda, and lies concerning that truth does not render it subjective.

Absolutely, but then who is responsible for judging what is true and what is false? And why should I trust those people not only to be correct all the time, but also to never fall victim to their own biases when making these decisions?

The point is that there is no such thing as a perfect arbiter of truth. I may trust a friend with whom I have a close personal relationship to always tell me the truth, but it seems foolish to place that kind of trust in a corporation full of people you don't know.

In the end, we're back to just not filtering information at all, but instead leaving it up to each individual to decide what's true and which isn't. This also clearly has not worked. Maybe this problem is just not solvable.

Back during the first Iraq war (Kuwait) my sociology class did an analysis of all the news to see how they reported it differently. We look at the 6:00 news on ABC, NBC, and CBS and I asked if we could also add CNN because this was when cable was still being adopted. What we learned was back then CNN largely reported the facts/observations with qualified statements and not much commentary at all, and all the other news outlets got the bulk of their facts (except for local news) from the CNN newswires/people on the ground and then fluffed them with their own subjective language. That was when CNN was just Headline News. They've really lost credibility since then. I'm not even sure if Headline News even exists anymore. It used to just be every 30 minutes the news cycled over and over and updated as new info came up.

AP News is generally what everyone pulls from and reports mostly straight facts

I see this comment in so many forms across many channels. “Oh no, there is no truth anymore!” (Sorry if I sound a bit pedantic)

But that’s not truth either. While gaining a 20/20 truth picture is hard, it’s not impossible to get a picture that’s accurate enough.

Lots of information can be lock-stepped into a picture that gives enough clarity. A combination of experts, reports and documentation, high quality sources, context awareness, and people I know around the globe gives me a good enough picture.

Also the message that there is no such thing as truth is exactly what disinformation strategists want to amplify.

Yes it’s worrying, but not unsolvable.

I feel like good news can distill down to: if there's not enough fact checking in the right domain(s), the editor fucked up. Conspiracy theories, and propaganda obviously fail this test, but do not do so online often enough. Especially not when grifters/propagandists are trying to make money through ads or are state funded by an autocratic government.

Should DDG take into account what is and isn't fact-checked? Absolutely, and if DDG gets better at this, their users will benefit.

I agree. I also believe transparency should be there on what changes are made in some shape or form. Unlike Google, DDG could actually do this.

A disinformation search engine would be interesting..

You're saying as knowing what the "truth" is is an obvious thing, and anybody could do it, and anybody could recognize whether certain statement is "truth" or not. But it is far from being the case. Many very smart and very honest people routinely disagree on many subjects whether something is truth or not, and on top of that many people also make statements that may look like truth but turn out not to be after detailed and complicated investigation. You can not expect an owner of the search engine to be the arbiter in these disagreements and to perform those investigations. And even if they did, who would ensure they themselves are being objective and not just enacting their own biases and hiding facts that they think are inconvenient to them? How do you observe this fact? The truth may be not subjective, in a pure Platonic sense, but our knowledge about it and our trust in ways how to get to it certainly would be.

"Good information", almost synonymous with "truth", is not subjective.

Truth is "not subjective" in the sense that force of will can't change the answer to even very fraught questions like "what programming paradigm works best in situation X" or "which running back would do best on team X in superbowl Y".

But one's judgement on the truth of complex question depends on one's opinions on the meaning of various terms, reliability of various individuals, the dynamics of human psychology, etc. This overall situation results in people's opinions being hard to compare and "truth is subjective" is often shorthand for this, although it would be nicer to have a different word.

> "Good information", almost synonymous with "truth"

Truth gets you about 1% of the way towards returning "good information". What are the ideal search results for "orange"?

This one is actually easy. Most commonly, orange is a color and a fruit. I'm sure we can all pretty easily list some objective facts about both of those things.

Orange is also the name of a town in some places, and a county in others. But I think you need more search terms in order to disambiguate. The search engine can't read minds, after all.

Maybe the user was actually looking to pay their phone bill? (The first result in DDG is orange.fr, not the color or the fruit)

My point is, there's no objective way to determine what good search results are. There are thousands of different objective measurements which people combine in subjective ways to create a subjective algorithm.

Truth, because different people have different axioms or goals, might be subjective. I don't think this relates to quality of information.

A higher quality information media will give you the tools to get more information, and this is for example where wikipedia shines by requiring citations. Even if some piece of information there is wrong, you usually get the tools to obtain more information, and you get a summary (where they do attempt to stay neutral) of multiple positions when there is no consensus.

Quality of information is not precisely equals to truth, some questions to ask of any information relay: - Do they give you the sources for their information ? - Do they make a good faith attempt to stay neutral ? - If not staying neutral, do they make their position and conflicts of interest clear ? - Do they provide adequate (non-strawmen) summaries of opposing views ? - What is their process for correcting information ?

Of course wikipedia isn't perfect, but it is a higher than average quality of information on the internet. You should use wikipedia (or any source really) as an authoritative source (something isn't tree just because it is written their), but it is an excellent starting point. As most other good sources of information are.

You can't easily judge the veracity of any piece of information on any platform, but you can more easily judge a good faith attempt at providing you the tools to obtain further information.

No platform is going to be perfect, but these criteria should help anyone to filter out bad news sources, and it should not be impossible to convince most people that these are good criteria.

Reply on news.ycombinator.com