Live data from Hacker News

Any claim without a URI should be treated as suspicious

shkspr.mobi

171–180 of 282 posts

Re: Any claim without a URI should be treated as suspicious

#171
post #59

I hate URL fights. I give little extra value to people providing links with their clams. I largely given up providing links to information when called out as well. It's not worth the time. Discussion on the internet is broken in most places (HN is less broken than most). I find little value providing URLs. Much of the time I provide them, few folks seem to read it, fewer actually want to read it in an intellectually…

One of the chief benefits I find to providing links when in a discussion is - I get to realise I'm wrong before I've posted. I'll have some half-remembered quote or statistic in my head and I want to provide a link to evidence it. But after finding a decent source I find that I've misremembered something - usually it's a minor detail but sometimes not and sometimes it undermines the point I was going to make.

That kind of self-fact checking is enormously useful since i) I look less of a fool and much more importantly ii) Less bullshit is propagated overall.

I'm not sure how you can know that "nobody is really reading [your links]" - on most forums the number of people reading a thread will heavily outnumber the people participating. Who knows how many people are reading your links, perhaps months later?

Even when not in a discussion/debate I think providing a link is just good manners. Searching for something might only take 2-3 minutes but why should every single one of your readers have to do that? Isn't it more efficient that I do it once and provide the link? This is especially true when there are multiple versions/editions of (for example) videos with very similar titles but different quality or subtly different content. A URL helps to say "This is the specific thing I was talking about".

Re: Any claim without a URI should be treated as suspicious

#172
post #137

Earlier quoted context omitted.

About five or six years ago I spent a long time making outrageous claims on Reddit, and I'd wrap various parts of it in link tags. The links would go to random nytime articles, random papers, or simply nowhere at all. Then I'd watch as the outrageous claim propagated (this in niche communities). It was absolutely terrifying. I never, ever got called out. EDIT: I'm remembering now I'd pretend to be all sorts of people…

There is a huge asymmetry between the effort to make an outrageous claim and that to refute one. > I never, ever got called out. In Reddit parlance: ain't nobody got time for that

> There is a huge asymmetry between the effort to make an outrageous claim and that to refute one.

Asking "Got evidence for that?" is sufficient refutation for baseless claims.

Re: Any claim without a URI should be treated as suspicious

#173
post #52
post #18

Earlier quoted context omitted.

The difference between URI and URL is practically meaningless except to a very small circle of specialists. Noone uses the name "URI", even when they are talking about one. Except if they explicitly want to signal their membership in that circle of experts.

Am I wrong to suppose that L is locator and I is indicator, thus I being a more broadly term and a somewhat nerdy term for Source?

uri encompasses/superset of both url and urn (there was also urc that went no where). urls have a "scheme" (http/ftp/gopher/etc) to access the resource, while urn is a persistent non-location based type of name identifier e.g. urn:ietf:rfc:2648

Re: Any claim without a URI should be treated as suspicious

#174

As many of those who have read academic papers know, even claims with citations are often suspicious. Time and time again I've found cited sources that don't say what was claimed. Sometimes, citations are not specific enough (e.g., citing a huge book rather than the specific page with the evidence). So, better than giving a URL would be to give a URL with a "fragment" if possible, i.e., #id. https://en.wikipedia.org/…

I have a lawyer friend who complains about this in the legal field. A brief may cite a case for a specific proposition. They have to be pretty thorough because there is a nonzero chance that the cited case doesn't actually stand for the proposition, and in some cases says exactly the opposite!

Re: Any claim without a URI should be treated as suspicious

#175

As many of those who have read academic papers know, even claims with citations are often suspicious. Time and time again I've found cited sources that don't say what was claimed. Sometimes, citations are not specific enough (e.g., citing a huge book rather than the specific page with the evidence). So, better than giving a URL would be to give a URL with a "fragment" if possible, i.e., #id. https://en.wikipedia.org/…

> Time and time again I've found cited sources that don't say what was claimed. I've seen that here on HN, too. Person 1 claims X. Person 2 asks for evidence. Person 1 posts a URL, as if it's evidence for X. But when you read the contents carefully, you realize that it doesn't actually provide any evidence for X. It provides evidence for something that's somewhere in the same general topic as X, but that's not what P…

This is a good point although I would say that citations are still better than no citations as they at least give the basis for a reasonable discussion. Now that "evidence" for X has been cited, person 2 can come back and explain why the URL does not actually support X.

Re: Any claim without a URI should be treated as suspicious

#176
post #137

Earlier quoted context omitted.

There is a huge asymmetry between the effort to make an outrageous claim and that to refute one. > I never, ever got called out. In Reddit parlance: ain't nobody got time for that

> There is a huge asymmetry between the effort to make an outrageous claim and that to refute one. Asking "Got evidence for that?" is sufficient refutation for baseless claims.

Make the opposite of your claim with an alt account. Then challenge it for evidence. Now your position wins by default.

Re: Any claim without a URI should be treated as suspicious

#177

As many of those who have read academic papers know, even claims with citations are often suspicious. Time and time again I've found cited sources that don't say what was claimed. Sometimes, citations are not specific enough (e.g., citing a huge book rather than the specific page with the evidence). So, better than giving a URL would be to give a URL with a "fragment" if possible, i.e., #id. https://en.wikipedia.org/…

> As many of those who have read academic papers know, even claims with citations are often suspicious.

Yes, I've encountered that numerous times. One problem is the chain of strengthening citations; citation X supposedly proves Y, but it cites something less conclusive that cites something that has very weak evidence for Y.

Then there's the self-citation: citation X shows Y, but citation X is by the same author and cites something else by the same author and there's nothing solid at the bottom of the chain.

Then there's the authority-strengthening: a journal cites a less-rigorous journal which cites a book that cites a random person.

The point is that a citation is only the starting point for evaluating a claim. (The other conclusion is that I wasted a lot of time in the library researching claims for stupid internet arguments.)

Re: Any claim without a URI should be treated as suspicious

#178

As many of those who have read academic papers know, even claims with citations are often suspicious. Time and time again I've found cited sources that don't say what was claimed. Sometimes, citations are not specific enough (e.g., citing a huge book rather than the specific page with the evidence). So, better than giving a URL would be to give a URL with a "fragment" if possible, i.e., #id. https://en.wikipedia.org/…

This is disturbingly common in the legal world too. A judicial opinion will cite some earlier case as supporting a proposition, but if you dig into it, you'll find that the earlier case is merely tangential. (Or totally unrelated, or even completely contrary.)

Re: Any claim without a URI should be treated as suspicious

#179

Earlier quoted context omitted.

The extension’s author here. Just wanted to let you know that—apart from Chrome—it’s also available for Edge, and a beta version based on a (still unfinished) polyfill for Firefox. The same code also works on Safari 14, we hope to release it there, too. Install links: https://github.com/GoogleChromeLabs/link-to-text-fragment#in...

How is it different from this bookmarklet? https://twitter.com/austegard/status/1228150401427529728

The big difference is that the original intended text is encoded in the URL. An element-id fragment is better than just a URL but will often be less granular than the exact text being linked (e.g. a heading).

If you're looking to link a citation to a statement, having the text is far more useful since the person following the link doesn't have to now find the statement in a potentially large and broad section. Also, if the page has since changed and the text no longer substantiates the cited claim, that'll be immediately apparent.

Re: Any claim without a URI should be treated as suspicious

#180

Earlier quoted context omitted.

but then you run into the old saw about damn lies and statistics. the argument needs to stand on its own and be taken in good faith, otherwise discussion is futile. (edit: that's not to say provide no links, just that it won't make the argument, per my other comment, https://news.ycombinator.com/item?id=23899088 , ironically, a citation.)

If your argument is based on some fact (e.g. some number being above some threshold), then you absolutely need to back up that number to make the argument resolve into "it's most likely true". If you don't back the number up, then the argument resolves only to "if the data is indeed as argued, then it's most likely true". Keeping beliefs with free variables in them is fine, IMO, as long as you're careful about not ac…

sure, i'd go one step further and note that there are no absolute certainties, so those free variables always exist, even implicitly. in that vein, most threshold numbers like that should at least come with uncertainty bounds (damn lies and statistics).

said another way, outside of math and formal logic, real-world arguments resolve at best to "if the data is indeed as argued, then it's most likely true" (perhaps gödel is relevant here, as @marcinzm referenced downthread[1], another citation!).

[1] https://news.ycombinator.com/item?id=23899111

Post reply on HN