Hacker News Reader

> In particular I block Youtube, not because they aren't sometimes correct, but because I don't want videos polluting the regular results - it just takes too long to get info from videos.

Funnily enough, lately I've been prioritizing YT videos more when searching. So many sites now are just regurgitated SEO farms with minimal quality, and easy to see why: it's minimal effort to produce and cheap to host. But making a video takes time and effort, so has a much higher barrier to use as a click farm.

More than once when traditional search failed me, I went to YT and found some video from 2009 clearly and eloquently explaining what I'm looking for in detail, and without any distractions because the person authoring the video clearly didn't specialize in the media format or show interest in experimenting.

I've found it to also be a better source when looking a product to buy. Want to know which fan to get? Turns out there's a channel from a dedicated guy who keeps finding ways to test different fans and their utility and with multiple videos demonstrating his approach and findings. The mainstream channels aren't all that useful, but there's a ton of "old web" style videos (some even recent) passionately providing details for almost anything you'd think to search. And they're a gold mine.

Would a browser feature that skipped to the relevant parts of the video based on closed captioning and understanding search intent be useful? It seems like this would be a good way for Google to fight to stay relevant in UX vs having the chat bots just quickly spitting out a readable answer. Hunting through ad laden webpages is annoying. Seeking to the relevant section of the video is a solvable problem, especially for videos above some viewership threshold.

I've definitely seen Google do this already: https://searchengineland.com/google-tests-suggested-clip-sea...

Google seems to be taking much more advantage of YouTube's transcription feature lately. The first addition was the (ok, gimmicky) animation on the Subscribe button when someone says the dreaded like. Hopefully a sign of things to come.

Overall AI summaries are very welcome for a certain subset of YouTube which is sadly dominated by sponsored, clickbait, and ad-driven content.

Didn’t Google try this already? It seems useful to me, at least. IMO the next frontier of search is not better hypertext, it’s podcasts, audio, and video.

> Seeking to the relevant section of the video is a solvable problem

...and it has already been solved, though partially: SponsorBlock allows people to add a "Highlight" section to a video, which denotes the part of the video which the user most likely wanted to see (sans the "what's up guys", "like and subscribe", etc.)

Of course, it's not perfect: it relies upon humans doing the work, though some may see that as a positive over something more computerized.

Do you have some tips for finding concise videos that answer the question you are asking? I am finding more and more obvious LLM bullshit in results, so I am willing to try some other tactics. But I am not ready to spend the minutes watching videos to see if it is actually relevant or a waste of time, always artificially long to increase ad revenue.

For me, it really depends on the type of video. For fixing cars, I'm usually looking for something specific enough that there isn't a lot of chaff. It was probably recorded and edited on a phone just to splice the clips together. Probably the default thumbnail that youtube extracted from the video.

For product videos, if Project Farm did it, look there first. Otherwise, I look for someone has a lot of videos for competing products with basically the same format, not over 10 minutes.

Tech videos are the hardest, I often still prefer text. Maybe look for links to the docs in the description? I still get duds though.

I don’t know much about fixing cars, but yeah, YouTube is a treasure trove for tacit knowledge.

Wish I did, but here you're at the algorithm's mercy, unfortunately. One possibility is subbing/accruing watch time on channels that you find provide you the right value, so that the algorithm might recommend similar channels on other subject matters.

> But making a video takes time and effort, so has a much higher barrier to use as a click farm.

> The mainstream channels aren't all that useful, but there's a ton of "old web" style videos (some even recent) passionately providing details for almost anything you'd think to search. And they're a gold mine.

This won't be the case for long. YT is already starting to be polluted with spam and AI generated content, which will get more and more common. The same thing that happened to the web in text form, will happen to videos.

I think the only solutions are using allowlists for specific domains, and ironically enough more AI to filter specific results. Or just straight up LLMs instead of web search, assuming they're not trained on spam data themselves.

Yeah. I was recently looking for videos comparing two smartphones and among top ranked videos there were videos that just show the phones side by side and the video consists of showing specs side by side and videos that just have LLM-generated text, added to the video with TTS.

One critical difference is the date attached to youtube videos. It's easy to verify that a video was made before this tech was available, but you can't do that with websites, or search engine result pages.

It does limit utility for more modern needs, unfortunately.

Note that the problem of filtering bad data out of learning material isn’t inherently easier than filtering same out of search results.

That's curious, I generally hate video due to inability to glance over content, and the few attempts I made to actually find useful information I searched for resulted in... spammy extra low effort video content that did not answer my questions.

Depends on what you’re looking for. A blog post about how to play Search and Destroy by The Stooges is not as useful as a video of James Williamson himself showing you the riffs!

Well, I don't think I'd be able to learn much just from watching the concert: teaching is fundamentally different from doing.

So I think even that example does not universally hold. I'd still appreciate a write up with tips on what's important and if there are any transitions to focus on with only the bits on video where some of that is demonstrated.

Now, I can barely contort my fingers into one riff, so I lack the knowledge to understand what I am missing, but I'd still have a hard time learning that from video.

I’m not talking about concert footage, I’m talking about James breaking the song down and showing you the riffs at quarter speed.

Until a recent YouTube video I was playing the song incorrectly. It’s blazing fast and the mix is sort of insane so it’s very hard to hear exactly what is going on. And the tablature isn’t going to let you see how his body fits into the groove.

This is tacit knowledge we’re talking about, not book learning. Guitar instruction is always hands on.

Almost everything is hands-on (everything apart from things you really can't do hands-on, like exploring black holes): I don't remember seeing someone come out of reading a book on programming and being a master programmer.

But video is not hands-on any more so than text: if it was, live concerts and sports games and other performances would not be such a big deal. Sure, video is richer in some signals (audio/video), but poorer in others (introspection, pacing and focus...).

That does not mean I can't read to understand a new topic or to be prepared to look for subtleties in a hands-on performance.

If anything, to a great student, they should be complementary, but still, each student will have one or the other contribute more to their learning, and that depends both on the teacher, but also on the student.

> Though I do sympathize with folks who reminisce about 2008 Google Search results, there were probably orders of magnitude less content out there and a complete ignorance to how valuable your place is on your business and thus no SEO.

That was a decade after Google was created and people certainly understood SEO and Google was constantly updating its algorithm to punish people who were trying to game the algorithm.

The wikipedia page on "link farming" for example references it happening as early as 1999 and targeting SEO on inktomi:

https://en.wikipedia.org/wiki/Link_farm

I remember some internal presentations at Amazon around ~2004 about how boosting Google SEO on Amazon web pages increased traffic and revenue (and Amazon was honestly a bit behind-the-curve due to a kind of NIH syndrome).

At the time it seemed like Google was winning, though. SEO seems to have gotten really good, or maybe Google just gave up.

I have a hard time believing it's so difficult for a search engine to distinguish between a credible, respected website that has been around a while with some generated garbage that exists to be a search result. We humans can tell them apart, so in principle, computers can too.

Yes, this should be table stakes for a classifier - a company with the resource of Google can definitely solve that problem if they weren't themselves in the business of spam (advertising) and benefited from spam sites (as they often include Google ads/analytics).

Google is quite quick in plugging holes in AdSense but AdWords.

> table stakes

Always “table stakes”. Do you think in buzzwords also? I’ve always wondered this. Or do you think normal words and then translate it into this bandwagoning / membership proving garbage ?

Hey, please don't cross into personal attack on HN. It's not what this site is for, and destroys what it is for.

If you wouldn't mind reviewing https://news.ycombinator.com/newsguidelines.html and taking the intended spirit of the site more to heart, we'd be grateful.

Edit: unfortunately your account has been breaking the site guidelines in a lot of other places too—here are some recent examples:

https://news.ycombinator.com/item?id=38825624

https://news.ycombinator.com/item?id=38825543

https://news.ycombinator.com/item?id=38783196

We eventually have to ban accounts that post like this, so if you'd please stop doing that, we'd appreciate it. On HN the idea is: if you have a substantive point, make it thoughtfully; if not, please don't comment until you do.

I guess this brings up the question of how good are humans at doing this across a wide number of domains on average?

The other question I have is how long do these garbage results stay up for a particular query on average?

Google's PSE is neat but there isn't a good way to manage switching between them. They could easily add a little dropdown to let you select which one to use as part of the public link UI they provide for each one individually. Giggle[1] gives me this ability and I run it locally (alongside Kagi) for more specific things to target domain lists I've been building over the years.

1. https://github.com/dan-lovelace/giggle

I'm a big fan of the non commercial site search engines because of the gaming aspect. If you're not generating revenue from the clicks the game mostly goes away.

I'm not saying people aren't entitled to make some money, but it clearly incentivizes user hostile behavior.

Maybe make it an option because legitimate sites like journalism also use this model.

Subscription model like Kagi seems to work pretty well against gaming the results.

Their only remaining incentive is to be good enough that people keep paying for the service.

It works not because they're somehow smarter or have more resources than Google at detecting spam/SEO, it's because unlike Google (and other ad-supported search engines), they make money from result quality and have an interest in blocking spam.

Google on the other hand makes money off ads (whether on the search results page itself or on the spam sites), so spam sites are at best considered neutral and at worst considered beneficial (since they can embed Google ads/analytics, and make the ads on the search results page look relatively good compared to the spam).

Black-hat SEO has been around since the early days of search engines and they managed to keep it at bay just fine. What changed isn't that there was some sudden breakthrough in malicious SEO, it's that it was more profitable to keep the spammers around than to fight them, and with the entire tech industry settling on advertising/"engagement" as its business model, the risk of competition was nil because competitors with the same business model would end up making the same decision.

The same reason is behind the neutering of advanced search features. These have nothing to do with the supposed war on spam/SEO, so why were they removed? Oh yeah because you'd spend less time on the search results page and are less likely to click on an ad/sponsored result, so it's against Google's interests and was removed too.

Kagi works because there is no incentive for SEO manipulators to target it since their market share is so small.

Super tinfoil hat to believe Google wants to send users to blog spam websites (e.g. beneficial to Google).

Anytime there is money to be made, there is an effectively infinite amount of people trying to game the system.

Google is a complex system so “want” can just include we are making money from the blog spam and while we don’t like it other things take priority over fighting it as effectively as we could.

It's never tinfoil-hat to assume that a corporation is, at very least, making sure not to fight too hard against any activity that brings it more revenue.

That's why taking the money out of the click is effective.

There can be other models for making money, but methods that really on casting a wide net and driving low quality traffic is the thing that shouldn't be indexed or at least labeled as such

But the author tried Kagi and the results don't appear to be noticeably different, filled with scammy adspam just like Google and Bing. Kagi's results seem to mostly aggregate existing search engines [1], so this isn't much of a surprise. Perhaps a subscription-based service that operates an index at Google's scale might help, but no such thing exists to my knowledge.

[1] https://help.kagi.com/kagi/search-details/search-sources.htm...

Right, but Kagi has built in tools to make it easy to fix that. Blocking those spammy sites from ever showing up again. Moving certain sites up the ranking, and so on. These features mean that over time my Kagi results have become nearly perfect for myself.

This is addressed in the article. As Hacker News readers and expert computer users, we have a bag of tricks that we can reach into in order to make our searches perform better. With a similar level of effort and an expert user's intuition you can get good results out of any search engine. Not so for the average user. In fact, again paraphrasing the article, Google's original claim to fame was that you didn't have to spend a lot of time doing exact keyword matching and fancy tricks in order to get good results.

> it just takes too long to get info from videos.

I can’t wait until video transcripts get fed into LLMs just to eliminate the whole “This video is sponsored by something-completely-unrelated, more about them later. What’s up Youtube, remember to like, share, subscribe… 5 entire minutes pass on similar drivel… the actual thing you want, but stretched out to an agonizing length”

You need SponsorBlock.

Usually people leave a "highlight" marker which tells you where you're supposed to jump to. Along with the regular "This video was brought to you by <insert>VPN".