Hacker News Reader

Comment by fuckcensorship | Hacker News Reader

Check out Marginalia[1]:

>The search engine calculates a score that aggressively favors text-heavy websites, and punishes those that have too many modern web design features.

I have found some excellent websites using this which I would have never found via Google or elsewhere on the modern web. Very reminiscent of exploring the “Old Internet.”

[1]: https://search.marginalia.nu/

View on HN

I know hundreds of interesting websites and I can discover more thousands with ease.

My problem is not with information discovery but in finding the EXACT bit of information that I am interested in. The old Internet wasn't better in this regard. I've came around information that was of interest by lots of digging and sheer luck.

Already now that site gives me significantly better results for certain queries (try searching for something along the lines of

  dual boot windows linux

  git fixup

or similar on marginalia to see what I mean.)

Warning: it seems to be a one person thing and my experience is that results might differ somewhat from week to week.

Edit: someone usually point out that SEO spammers will arrive shortly. I say that if SEO spammers start optimizing their pages by making them useful, lightweight, readable and without most tracking and ads I see that as a win.

The results are a bit hit and miss, I think it has to do with the fact that either there isn't content that match my criteria in some topics, or the spider is stuck in some local minimum and simply hasn't discovered it yet.

> Warning: it seems to be a one person thing and my experience is that results might differ somewhat from week to week.

Yeah, it's still very much in development and I only have one server so unfortunately I have to do some of the develoment and testing in a way that is very visible. I can usually do preliminary tests off another computer, but it doesn't have nearly the same resources so it can only deal with a sample of about 2 million documents, makes the test results pretty hard to interpret sometimes.

> Someone usually point out that SEO spammers will arrive shortly. I say that if SEO spammers start optimizing their pages by making them useful, lightweight, readable and without most tracking and ads I see that as a win.

I'd also like to add to this that since some of the things I use as quality measures are opposite to the quaility measures used by Google, that means that any site that attempts to optimize for both metrics underperforms in both search engines.

Please be aware that already at its current level it beats its competitors hands down in not being annoying :-)

Edit:

> I think it has to do with the fact that either there isn't content that match my criteria in some topics, or the spider is stuck in some local minimum and simply hasn't discovered it yet.

This is totally fine with me:

Telling me the search engine doesn't know of any good matches for my search is not only a valid answer but also a far better answer than a list of a billion pages that some dumb AI somehow can construe as being related.

In fact the list of non-matches has a negative value since I end up sifting through a hood bunch of them in a year hoping that this time there was something usable there.

Nice search engine! Is it yours? Pretty accurate and nice for content discovery. Just read some interesting stuff after searching for 'bicycle carts', which is not your everyday topic I suppose.

I don't understand this obsession with text-heavy content.

One of the virtues of this very website is its text-heavy design.

It's a proxy indicator for a certain character of website. As well as a direct indicator of the content itself.

Many people enjoy reading.

Text is super easy to reuse, manipulate, cut, copy, paste...

Text heavy usually means content heavy.

A disturbing amount of the web is "How To Make A Made for Adsense Clickbait Page In 2021" containing lots and lots of pictures to pad out the ads and amazon referral links, but very little prose.

I'm not sure it's an obsession, but I do like the simplicity of just text content. There's fewer distractions and annoyances (off the top my head: the aggressively annoying habit some sites have of modifying the DOM a few seconds after the page loads, seemingly timed just right so that the link you just clicked hits something else entirely).

There's lots of good things about the media-heavy interactive websites of today, but sometimes it's nicer to just read what people wrote. As mentioned before, this site is a good example.

They're not modifying the DOM. They're loading ads from external sources which then pop into the page. I hate this too but a good adblocker solves this for 95%.

I had a discussion with a website developer recently asking why they insist on loading all these ads from external URLs. All the different loading times cause this. But he said the ad provider needs this in order to validate someone looked at the ads, they don't take the web owner's word for it.

So blame the ad industry, not the website :)

To get a little pedantic, the DOM changes in that the size of some element gets modified after the page loads, causing other elements to shift around visually.

I can empathize with the developers, but I would prefer if they sized out those elements in advance so the page elements didn't get shifted all over the place once something loads.

Isn't that the draw of news aggregate comment sections such as this?

> I don't understand this obsession with text-heavy content.

Sites with it have a positive attitude towards literacy, and that often goes with other good things.

I just love text because it's ready to consume at my own pace. I can glance through the sections that don't interest me, skip the parts I already know, I can slow during the really deep parts and consider how they fit in to my situations. All in my head.

If I do that with a YouTube video, I have to constantly fast-forward and pause, and it's much harder to pick up the thread again after you've jumped a few minutes. Also, YouTubes (and the same with online courses) tend to be really slow to make sure even the slowest viewer doesn't disengage. And tend to have cutesy 'funny' bits to keep people 'interested'.

I have nothing against people that enjoy those things. To each their own. But I really don't work that way myself and I hate it when people impose it on me.

I just started a new job and my boss gave me a whole load of YouTube links to study and also podcasts (which I hate even more).. I really struggle wading through that kind of content :( so I proposed he just give me the topics and I research them myself.

> I just love text because it's ready to consume at my own pace.

This! Especially when my reading speed far outpaces most Youtubers talking speeds even at 2x. It drives me up the wall.

Counterpoint: I do not understand the apparent massive appeal of video content.

To avoid being too brief:

• For entertainment: I have never cared for most television or movies. I've watched the entirety (or even majority) of fewer than ten series in my life, and movies in the low double digits.

• Video is a terrible medium for covering most kinds of information. Its speed is often fixed, and it's barely searchable! It is good for showing process or sequence (e.g. some kinds of how-to content), and for showing 3D perspective it's an acceptable substitute for a freely movable camera on a digital model, but for almost anything else informative, text can convey more, and more quickly, and more responsively.

This is a band-aid on top of the searching issue you described, but it's a godsend when a youtube video is nicely subtitled, so you can open up the transcription and ctrl+f to find exactly when a specific word/phrase was said. Youtube also has limited speed controls at this point. I think the fact that those features exist highlight your second point. Searchability and fixed speed make video content less accessible.

I believe that chromium is still the state for the art for realtime processing of playback rate increase at speeds greater than 4x.

Forgive my stackoverflow bodge job (this isn't my bread and butter) but this is what I came up with, when I had to speedrun a couple of courses.

I use the shift key to scroll the page sideways (which it can't) so that the script registers the event correctly without moving the page. I just scroll up to get through the sluggish parts and scroll down for the meaningless trivia that inevitably ends up being crucial insight when you least expect it. I would have attempted to implement acceleration with my limited skills, but my scroll wheel has mad detents so it wouldn't have helped. Not sure if it would work on youtube.

  javascript:window.addEventListener('wheel', function(event){ if (event.deltaY < 0) {  console.log('scrolling up');for (vid of document.querySelectorAll('video'))  vid.playbackRate += 0.1   } else if (event.deltaY > 0) {  console.log('scrolling down');for (vid of document.querySelectorAll('video'))  vid.playbackRate -= 0.1 }});

But why should I have to search subtitles instead of Google doing it automatically?

Not to mention that you can copy and paste from a text site, while a video leaves you laboriously retyping everything.

I watch loads of series, movies, yet agree with text as preferable for most web circumstances.

Most people are barely literate. Literacy figures are based upon grade 6 reading levels, and that's average reading levels. So when you see 99% literacy rate, yes, but that's for reading stop signs.

It also pains some to read. You are doing them a disservice, to ask them to read.

Most adults, pre-internet, never read books, and only some read newspapers. Of those newspapers, many only read tabloids (high school reading level, barely), or looked at magazines with large pictures, quizzes, and such.

The rest often watched 4 hours of TV a night, no reading required.

The modern internet does not want to exclude, profit is reduced if so.

So picture books for all!

Video is less useful to disseminate information but is good for entertainment and for learning purposes.

I hate videos.

HEY GUYS WELCOME TO ANOTHER EPISODE OF $ASPIRING_INFLUENCER_BLOG

<60 second intro movie/theme song>

<2 minutes of content>

THANKS FOR WATCHING AND DONT FORGET TO HIT THAT SUBSCRIBE BUTTON

Don't forget to buy a $DEFINITELY_NOT_A_SHADY_VPN subscription and use promo code OMEGAGRINDSET to get a 5% discount!

Text is the ultimate killer app.

I have a personal site I'm trying to pursue an old-style feel with, but it never qualifies by the heuristics people use, mainly because:

* image-heavy (with careful expressive intention! https://maya.land/latin/)

* I have javascript that, horror of horrors, is separated into multiple files (so I can manage it without a build step) for such malefic purposes as accessibility (https://maya.land/prefers-reduced-motion/) and displaying the current moon phase

But I think it's kind of good that there are different people pursuing different visions of The Good, even if many of us from different corners are all inept calling it nostalgia without better language to articulate our visions. This blogger is presenting cesspools like voat and raddle as a positive vision. Despite the fact that I think they're wrong about a lot, I think it's good that we're both out here on the web seeing and doing things differently. Maybe it takes some people to be obsessed with text-heavy content to act as a counterweight to FB misrepresenting video view counts. Maybe it takes some people like me going wild with GIFs to balance out Leo Babauta's web ethic (https://mnmlist.com/w/), and both of us to balance out the standard wordpressy norms.

ETA: I realized this looks like self-promo, so if you are suspicious of my using my own site as an example, https://sadgrl.online/ isn't me but illustrates a lot of the same thing I'm trying for

It's pretty hard to quantify quality due to the inherent category error in attempting to do so, so the measures are pretty much always going to be blunt and inexact.

For what it's worth, while my marginalia /barely/ tolerates the technical aspects of maya.land, it's really teetering on the precipice of getting completely rejected, at the same time it ranked very highly in the search results due to its graph-adjacecy to the indieweb (it's domain #182 in that ranking).

One thing that I really admire/respect about your search is that it's really good at capturing a particular kind of good web page. I wonder if there's more precise language than "quality" to express that: inexactness is only inevitable because we're trying to lump everything good into one bucket. Things I've considered improvements to my site are considered degradation by marginalia search's heuristics if we're trying to fit them all on a good-bad spectrum, but maybe that isn't fundamentally what we're asking a search engine to do -- even an opinionated one.

The point isn't that it should be text heavy.

The point is that it should be light on tracking, ads and unnecessary use of scripting. (I have earned good money writing client side code the last few years so I see there is a place for it, but most websites I look for information from should not need it.)

Well if it's not for you, then it's not for you. I'm not particularly into the marvel cinematic universe, but I don't rain on anyone's parade if they like that type of stuff.

Text-heavy content isn't for everyone, and that is fine. The problem is that the modern Internet primarily caters to the lowest common denominator of viral mass appeal, which leaves the long tail, those who want more substance, yearning for more. That /more/ is out there, but it has massive discoverability issues.

Reply on news.ycombinator.com