Hacker News Reader

Comment by nomilk | Hacker News Reader

I searched youtube today for a particular hardware product. The video I clicked on was 20 minutes long, and by every indicator I could see, the presenter looked like an expert at what they were talking about. Yet by the end of the video (which, thankfully, I watched at 3.25 speed), I hadn't gained a single new piece of information, and my brain hurt from dealing with so much repetition of facile hearsay.

The key to the success of this video (and others) was that the presenter was nothing more than a salesperson, not for the product they were talking about, but for themselves.

They're trying to sell themselves as experts, and if we're not careful, we end up trusting them and paying (with our time).

I got duped, but it's a continual arms race between viewers looking for useful content, and facile content creators trying to game the system by imitating the hallmarks of great content without any substance to back it up.

View on HN

This is the reason for my strong dislike of online content moving to video. It's way, way harder to skim a video to see if it's decent quality.

I feel like this is the inevitable result of video content having a lower entry bar than written. And even when writing's bad, it's more immediately obvious.

>This is the reason for my strong dislike of online content moving to video. It's way, way harder to skim a video to see if it's decent quality.

It depends on what people mean by "hardware". If the hardware is comparing something like graphics cards, then yes, text content with table of performance numbers and some charts is faster to parse.

But if the hardware is more physical hands-on like mechanical tools, drones, grills, etc, videos of people demonstrating its use and/or reveal the flaws is easier and faster to digest. ("Picture is worth a thousand words.") E.g. a Youtube video comparing a bunch of miter gauges is better than a magazine article that's just text and photos. Text+photos is too static so the reader can't see and hear the slop in bad products that machined are out of tolerance. Conveying that type of quality is more effective in videos.

If you're viewing Youtube on desktop instead of mobile, the fastest way to to skim is to press the keys 1 to 9 on the keyboard which jumps to timeline at 10%, 20%, 30% ... 90%. There's latency for 2160 4k 60ps videos but for lower bandwidth videos, it works well. If the video looks good, press 0 to rewind to the beginning.

It’s also monetized better

Not the gp but the "Enhancer for YouTube" extension for firefox supports up to 5x (given enough bandwidth) and MediaPipe for Android up to 3x.

Yeah but then you don't understand what they say, so it's not useful.

I guess this is what the Internet has come to.

It took me a while to come across an actually-convincing example [1] of where 5x speed is useful. But the current top HN item [2] is stunning at 5x. Not to mention that it clocks in at a minuscule and almost entirely procrastination-guilt-free 35 seconds!

[1] https://www.youtube.com/watch?v=jh_ukt8g53c

[2] https://news.ycombinator.com/item?id=30312694

But the truly important question: "Did you like and subscribe, bro??"

I'm sure you can get quality information on YouTube and even TikTok, but it's likely to be buried under all the highly ranked LikeAndSubscribe spam. That's the problem the interviewee in the article points out:

> What's really sad about all that—in my opinion—is the way that the algorithm works on TikTok is based purely off of views and likes and comments. It's not based on actual good information. The people who post good information don't get a lot of views. It's the people who post ridiculous stuff that get a lot of views.

Curation based on popularity doesn't work, and we seem to fail to learn this over and over: Media content likes, product reviews in online stores, star ratings in app reviews, upvoted/downvoted comments. You can't crowdsource quality--you can only crowdsource popularity. And as soon as you introduce an algorithmic black box, you are handing the top spots over to whoever can game the algorithm the best.

Oh, and if you like this comment, remember to smash that upvote button!

I ran into this at a higher level recently. The YouTube channel HardwareUnboxed made a few videos last year where they needed to face questions about the impact of RAM parameters on software performance. They demonstrated a complete lack of understanding yet continually made up some bullshit and claimed they were an authority on the matter. The right thing to do in their position is to identify gaps in their knowledge and research it. Instead they are confidently wrong and selling the idea that they are the go-to for computer knowledge.

This is one of the many issues with YouTube removing dislikes... With the dislike to like ratio, you would have spotted this immediately, and wouldn't need to waste your time.

This was probably by design

"[search terms] + reddit" does usually the trick for me. But I agree that it is getting harder by the day

Yep, site:reddit.com provides consistently better results than a plain search for reviews. An undomained search ends up with an endless list of sponsored/generic "top 10" sites which if you know anything about the products are clearly not written for the benefit of consumers.

I wonder if there could be some mechanism to counter this, like some kind of a button that you can press and other see that you're not happy with the content ... but that would probably be impossible to implement right, no way we're there technologically ... maybe some day ....

What add-on are you using to watch at 3.25x speed?

It's called 'Video Speed Controller'. I try to minimise addons but I couldn't get by without it. 'd' key speeds up by 0.25 and 's' slows down by 0.25. It works on videos on most sites, not just youtube. Also handy are Youtube default shortcuts: right arrow to skip 5 seconds forward and left arrow for back 5 seconds. Tap them a bunch of times to skip product placements or long intros. Just a warning when you use these a lot it will make regular TV unbearable (if it isn't already).

To skip intros and sponsored content, I use SponsorBlock extension. It skips those segments automatically.

First time seeing this. Are the timestamps crowdsourced or AI generated? In your experience how accurate is it? I notice it's got <1000 reviews. I can see this really taking off in the next few years. Thanks a lot for the recommendation, I've installed it and look forward to seeing it in action.

Timestamps are crowdsourced, not AI generated (AFAIK). It is 100% accurate of most of the videos. Since it is crowdsourced, less famous videos or videos with less views won't be skipped during the sponsor content, but we can be a good citizen and add the timestamps ourselves to help others.

Reply on news.ycombinator.com