Hacker News Reader

Comment by Spivak | Hacker News Reader

This is a weird stance to take about a search engine which is "tell me what to read as a service." Like you're literally using them to surface the wheat from the chaff. The algorithm that ranks results is already biased as hell and sometimes it's going to get it wrong. Google surfacing lots of stackoverflow spam sites isn't "unbiased" it's junk and manual correction is totally fine.

View on HN

This word keeps coming in this comments section: "spam." Spam was never mentioned in the Tweet, the source of this discussion. In fact no one in the comments section is talking about spam either, unless its a pathetic debunking attempt. They are talking about genuine results being down-ranked based on their ideological, even morally unscrupulous content, and why that is harmful to intelligent people who don't need their information retrieval service helping them form thoughts.

I just used spam sites as an example of "site that's highly ranked by the algorithm but are uncontroversially recognized as an error."

Define a "genuine result" for a search engine that isn't the result of a process that arbitrarily ranks sites by certain metrics? Like it's humans all the way down, you can't escape that search engines are a large scale system representing "what a group of humans thinks the best results are." There's no unbiased algorithms in search, it's all curation.

As I said - nobody here is confused what spam entails. I'll remind you again the Tweet did not say it it targeting spam - and your conflation of spam with misinformation is a useful bit of ignorance for those who want control narratives. Slippery slope, friend.

Is propaganda a "genuine result"? Yeah yeah, one person's truth is another person's propaganda and sometimes the lines are blurry. But cases like RT are pretty clear cut. Sometimes people actually do lie, and those lies are verifiable. Verifiable lies, and the sources who create them predictably and consistently, should not be considered a "genuine result" for a query regarding things that are happening right now.

Pretending that both sides are equal is not always helpful. Bad actors do exist, and pretending that everything is equally indeterminate is a fatally nihilistic and dangerous view. Sometimes we actually can tell when something is a lie, and we should treat it appropriately. For a tool that retrieves genuine results matching a query, that means down-ranking them.

For the purposes of current event queries, disinformation is certainly spam. The purpose of spam is to fool you into doing something you probably wouldn't otherwise do, usually based on false pretenses. Disinformation is the same. The action it promotes is more indirect - it seeks to influence beliefs, and therefore voting patterns and soft power - but it is still spread with a purpose that is disingenuous. It disguises it's true intent and is dangerous to the user. It is effectively spam applied to the domain of politics and current events, and should be treated as spam is. It should not be a valid result in an information retrieval service.

I agree that this ranking could be the start of a slippery slope. But DDG is being transparent, and you can still get to RT if you disagree. This is a good measure for people unfamiliar with Russia news sources and propaganda networks, who are the people most likely to be susceptible to Russian propaganda.

Cases like RT are clear cut to you. Others have other opinions. I think that the Russian claims that Ukraine is harboring neo-nazis are significantly bunk and are poor excuse for the invasion, but I also think you can't deny that the Azov Brigade exists, and I want to hear the Russian take on that even if I think it's half lies.

Then you can still go over to rt.com and check that out for yourself. With your knowledge that it is half lies, you can be actively engaged and discerning. But those lies should not be easily accessible for the uninitiated who are just casually curious about current events.

Intelligent people are better off not patronized - and fighting disinformation through censorship rather than rebuttal is a pathetic reflection on the state of society. This type of censorship makes it harder for intelligent people to make clear rebuttals and reinforces a culture of tarring people taking the first step of looking into the other side for this purpose.

This is always the line but it ignores that it's more vastly more effort to correct disinformation and propaganda than it is to spew it. Don't patronize intelligent people by making them have to spend literally their entire day having to fight this garbage rather than doing useful work.

Or you could do the obvious: give people capable of contextualizing search results unnerfed tools.

I think there is a conflation of literal fake news and propaganda. Propaganda can be subtle and not involve outright lies. A lot of propaganda consists of unsubstantiated rumors or speculation.

Sometimes it is helpful to read unsubstantiated rumors or speculation because it is notable. For example, the United States recently claimed that a foreign power was targeting its diplomats with a high powered microwave weapon. Substantial evidence was never produced, but American news media reported the story uncritically. I think that much of the reporting on this topic was disinformation or propaganda. At the same time, if I wanted to read about it, I would expect search engines not to make the editorial decision to censor this reporting. Even if it is disinformation, informed people should be aware of these accusations. In this case, I don’t expect the search engine to return the “truth” but rather “what is being said on this topic.”

> This is a weird stance to take about a search engine which is "tell me what to read as a service."

If users don't like the change in the search engine algorithm (for whatever reason), they'll stop using the search engine. It's not a weird stance to take.

At the very least this HN submission is surfacing search engine competitors like Startpage, Brave, and Kagi.

People can want a search engine that does stuff without running it through a Politburo filter. For a very long time that was the social norm in search engines. DDG marketed itself as "we follow the old norm, pick us and not the other bad guys." Now they are adopting the political censorship norm, contrary to their earlier marketing.

There's a difference between a low level filtering algorithm and manually curated results based on politics.

Of course, the right way to do it is to survey the common elements of sites whose politics you don't like and then add an "unbiased" content-agnostic rule punishing those elements.

Reply on news.ycombinator.com