But the links in this article, the ones generated by a script, are worse than useless. Meaning, they’re a waste of time that makes the text worse.
First of all, they are set to open a new tab. Rude. Second of all, a link to a wikipedia page for a term, for example, is just stupid noise. If I want to know what wikipedia has plagiarized about a certain topic I can go there and look it up.
Links are not an afterthought, something that can be usefully tacked on to the finished text. Each link should be genuinely helpful, or back up an assertion. Ideally, each one should be non-obvious.
- links aren’t dead: they’re vitally important. Links lend authority. [...] Don’t link to stuff you don’t trust, SEO or otherwise.
- Linkoln normalizes the post, replacing each wikilink with the best corresponding hyperlink it could find on the web, using Google
To that end, I keep a list of every interesting thing that I read. I plan to index that list, and make it searchable. Reliance on a search engine is a stop-gap to keep scope sensible at this point. There's nothing stopping other people from taking this idea and running with it.
Given your personal involvement in this area, I understand your apprehension. Like you, I want to see a denser, more interesting web. But without tools to overcome the static friction of linking things together, I don't think we'll see density and utility of links increase any more than it already has. That's what I hoped to highlight in this piece.
Like most HN readers, I've written my own content/information management system, and this makes keeping track of sources and inserting/formatting links much easier. Almost a pleasure, in fact. But there are plenty of open-source and other products out there designed to help with this task.
You could do that, or you could use one of the reference managers that already exist and do this for you. At the very least, try out a few and see what features they offer. If none of them work for you out-of-the-box, perhaps try out what extensions and customizations exist, or create your own. EndNote, Mendeley and Zotero are popular. EasyBib.com and RefWorks also come to mind. Most of these work with browsers and website to make adding and cataloging references easy.
Forcing to open on a new tab breaks conventional behaviour, it should be left as the user's decision (either explicit every time, or a setting in their browser. This setting already exists btw, if you like more tabs you can enable it in your browser).
(Maybe a case could be made if the website is actually an application and there would be data loss if not for the new tab, but I would argue design your application so as data loss is not possible in that case)
---
As for myself, the browsing history is something very useful to have, I like keeping the amount of tabs down to near one per activity, so I regularly use the history to go back to what I was reading before. Forcing new tabs means I have to reach for more advanced tools like tab trees, which are often not worth the cognitive effort.
This debate raged in the early days of the web: what is the better practice? A consensus emerged among web designers that leaving it to the reader makes more sense. I’m proud that I convinced J. Zeldman that his early practice of using `target='_blank'` hurt the user’s navigation, and so played a minuscule part in nudging designers in general away from this.
But if the link is hardcoded as a new tab, the user has no choice. There's no way to open a new-tab link in the same tab. :( Not to mention that there's generally zero indication it's a new-tab link anyways, so not only does it take away choice, but it does so non-predictably. Ugh.
Personally I have set `browser.link.open_newwindow = 1` in Firefox to force all links to open in the same tab, then I can middle-click to open in a new tab. This way I can do what I want without worrying about what the website author thinks I wanted.
It does have some downsides though. Some sites don't use <a> tags but instead <div> with JS to open the link, then you can't choose to open in a new tab (on the other hand they can't force you to open a new tab so it is sort of even on average). Some sites break because they are trying to launch a new window that communicates with the parent. But overall I think it is a big improvement to the web.
I sometimes link to the Wikipedia article when there are many things by a certain term, or when I'm using terminology that doesn't show up in search results (but is common in colloquial use). Should I do this a different way?
I agree. As bloggers continue to link only to good stuff between them, those blogs will become infinitely more valuable than the flood of trash that people will post from GPT.
It seems hopeless that people will find that good stuff, but word of mouth will do it, I think.
Ad apologists will claim "but how will we fund content creation!" and the answer is, if the content is actually good, people will be willing to pay for it--probably even if not forced to do so: lots of content creation is being funded by donations right now. "But that doesn't scale!" cries Hacker News. But what you mean is, it doesn't scale for a centralized entity that wants to collect rent on other people's work, and frankly, fuck those centralized entities. They're the ones creating the filtering problem. Paid, quality content creation inherently doesn't scale because it's inherently serial: a person/team creates one piece of content at a time. You can scale distribution, but you can't scale quality creation.
And the fact is, people will create because they want to create, even if they aren't paid to do so. Maybe they are paid, but that's secondary to the human need to create. As Quincy Jones said, "When you chase music for money, God walks out of the room."
If I'm being honest, most "content" I've come across on the web doesn't deserve to be funded and its creators are better served creating something else if they're looking to be compensated.
people will create because they want to create
What great content I have come across is usually created within a similar ethos to FOSS. People know they've learned a great deal from others in the past and want to pay it forward, or they want to share something with the world to see how others might iterate upon it.
if the content is actually good, people will be willing to pay for it--probably even if not forced to do so: lots of content creation is being funded by donations right now
I've also seen great content from people connected to a patreon, so there is definitely room for compensation, if the content is good enough.
I'm going to have to start quoting this; it's a much more succinct argument than I've heard before, and it captures the essence far better than I would be able to do even with many more words.
I can feel with your sentiment, but don’t see a practical way to do that.
Define “ad”. Is “read more in my blog” an ad? Is “see Foo’s blog at …” an ad? “Read Foo’s excellent blog”? “Read more here” in a blog entry?
If you say “when it is paid for it’s an ad”, how do you detect that? And what is “paid”? In a webring (https://en.wikipedia.org/wiki/Webring) people reciprocally ‘pay’ each other by providing incoming links, thus all gaining reputation. The difference with a citation ring (https://scienceintegritydigest.com/2022/03/23/citation-stati...) is subtle.
And for true ads, would you want to block a starting blog author from promoting their work online?
The problem is, payment is hard.
For authors:
- Micropayments are still ridiculously expensive
- Users want it to be as simple and hassle-free as possible, which means you're stuck with Stripe, Ko-Fi, Patreon or PayPal as that is what people have, and they don't want to trust Joe Random Blogger with their credit card information or their real name
- PayPal has been known to randomly close down accounts and hold your money hostage for months and years, especially if you're dealing with "morally grey" content (e.g. if you write about sex or anything related to crime)
- Accounting is annoying because each micro-donator needs a formal invoice, you have to deal with dozens to hundreds of micro invoices at tax time
- You're virtually stuck with credit cards and their limitations (again, see above, most of PayPal's ridiculous policies comes from the big CC networks) if your audience is international, but guess what, less than half of Germans have one [1], and outside of Europe the situation is even worse.
For users:
- Fuck no I don't want to fill out a form with CC number, CVC code, my real name and my address just to give a dollar for Joe Random Blogger's coffee cash
- Fuck no where did I just put my phone for 2FA
- Why is Joe Random Blogger now blasting my email address with newsletters?
- Uh, I did not consent Joe Random Blogger to suddenly draw in 20$ a month instead of 5$
The centralized entities (Paypal, Stripe, ko-fi, Patreon, whatever) take care of a ton of the associated bullshit, but they are expensive and, as I wrote, prone to randomly ban you without recourse - and not just ban your commercial entity but also your personal account - for life.
In contrast, ads are easy: you embed some piece of code on your website (or set some identifiers in your WordPress theme/plugin) and get a check or bank transfer every month. That's it.
Unfortunately, a lot of what folks are saying to you is true. It's very hard for GPT to cite its results because the way it is trained it is by design made to garble the inputs down into a simplified concept of guessing the next words given some input.
It essentially just has a procedural generator for words that is seeded by your prompt. If you give "alfalfa" it will head off in a direction towards "farm", "hay", and "grazing" along with other connective words to form sentences. Because its concept of data is all around just word spaces, it can't really go "oh, I read about alfalfa on a Wikipedia article with x sources". It just knows "alfalfa is like the word grazing". I am simplifying to make a point, but this is in essence how these algorithms work, directions of traversal, guided by probability, towards word clouds floating in a grouped space.
This is sort of changing though. Bing and Google (as well as many other researchers) are using specialized databases to provide further context that is fed into your prompts that come from real search results. Theoretically, they could get this tuned enough that GPT and other LLM can have the right data to provide a connection to cited facts alongside the hallucinated glue language.
I feel like what you're asking for is valuable, but might take a bit before we really get it relatively accurate.
IMHO links should be chosen "wisely", with the intent of helping the reader to find more information on important parts of the article or referencing other work etc.
Linking to the definition of rabbit holes is not really useful, if I'm not talking about varying sizes and depths of rabbit holes due to different breeds of rabbits and their prefered style of rabbit hole digging.
I used to link to reduce redundancy in what I had to write. Now, I tend to duplicate large parts of documentation to remove links, because I found that people actively avoid links. I'm guessing this aversion is related to a unconscious awareness of the "doorway effect", which exist even in virtual environments, and I assume, to some degree, on web pages:
> The doorway effect is a known psychological event where a person's short-term memory declines when passing through a doorway moving from one location to another when it would not if they had remained in the same place.
Something which could expand incrementally. It would be harder to write, but it would be excellent for documentation. Maybe an LLM would be good for such a thing. Say, instead of just linking to the documentation of another function, you could "zoom" into a description of it, tailored to screen estate or number of words:
zoom a box of 10, 20, 50, 100, 200 words
So, to bring something to attention, you need to both quote enough and provide a link.
For one's own archival purposes, it makes sense to locally store a copy if any important enough web page. No, that text or image is not guaranteed to be there tomorrow.
Incredible.
People enjoy the power wikilinks have and love sharing their knowledge graphs[0].
[0] https://preslav.me/2022/09/01/logseq-come-for-the-graph-stay...
ps. no need to downvote, its just an advice :)
How does specifically linking the first or last word (of ...the sentence? paragraph?) make things any clearer?
Wouldn't it be best to just link the words that most describe what the linked-to page is about (maybe even the linked-to page title?), no matter where they appear?
But yes we should all also be adding links in our writing and publishing on our own platforms.
How about allowing the readers to participate? Posts could be in a pre-publish state and the community could add link suggestions and annotations.
(defun my/org-insert-wikipedia-link ()
(interactive)
(apply-to-region (lambda (string)
"Convert a string to a link to English Wikipedia"
(concat "[[https://en.wikipedia.org/wiki/" (subst-char-in-string ? ?_ string) "][" string "]]"))))
This seems disingenuous - clearly a pun is intended
As a writer, I'm continually inspired by things I read or watch, and when I find something, I add it to my references. I use Zotero[1], but it's not the only solution. When I start writing, I already have my notes and the links to sources and related content. I don't add in links as an afterthought, the writing is informed by the related material.
Maybe, once in while, when I've done a couple of drafts and am getting close to the final version, I'll go back and add in links, but rarely do I spend time poking around on the web. I go to my references. Granted, the reference database takes effort to create and maintain, but the end result is far better than what I'd get just doing a web search for a term.
Love the idea of using a script to help with this too, though it's usually the harder to find stuff that's more valuable to link to, not things you ca immediately find with a quick Google search (like the Google home page).
> The script itself is… pretty dumb.
This is where the article lost me. If it feels like a lack of discipline, it probably is.
By placing links in an article, the author asserts that those were intentional choices just as much as were any of the other writing decisions made while making the post. The same thing goes for decisions surrounding the theme of the blog and what DNS name the blog is found under. The author is presenting an idea; more or less effort can be put into this, but it will show if effort isn't put into it.
Not spending time on links is the equivalent of not spending time on citations, arguably one of the most important parts of an academic paper. It is almost as important a part of the writing as the writing itself.
https://meadhbh.hamrick.rocks/v2/design/why_is_my_web_site_s...
- they allow me to back up claims with cold, hard facts instead of "pure hearsay"
- they defend me from getting held liable... say someone files a libel suit/C&D order against me. If I have a link to a proper, accepted medium (e.g. a newspaper or a TV station), I can defend my claim by pointing to the original source(s).
However, it is becoming increasingly difficult to link to stuff:
- publishers decide to put interesting content behind paywalls, leading to a constant noise of "can't read, paywall"
- publishers decide to re-launch their website, but not set proper redirects so all my archives are dead now
- German public TV/radio has to take web articles and TV/radio archives offline after a few days or months, because private media got a court decision and then a legal provision forcing them to take down content [1]
- some media absolutely LOVE live tickers, but you can't directly link to posts in these for posteriority
If you're referencing a study, link to the study. If you're citing a tweet, link to the tweet. Otherwise save spurious links as they add zero value and are a distraction and an unnecessary decoration making text less readable.
As an aside, I chuckled seeing the link to Atwood's "be a bigshot blogger" post where he recommended that people blog constantly about everything. For those who haven't kept track, that was a failure model. It made people basically give up on "blogs" because there was so much low value content, with people writing on a schedule rather than because they had actually interesting content. Now everyone just hopes that the rare useful post appears on a social news or media site.