[motbob]

I agree that removing public dislikes is a pretty bad idea, but it's not that important of a tool in fighting misleading videos. I mean, there was once a HUGE glut of videos with misleading thumbnails/titles, and what stopped the glut was not dislikes, but rather Youtube's use of watch time as an important metric.

[nonbirithm]

Beyond the debate on whether or not downvotes are useful as a tool, I think about the higher-level implications of a change like this.

One of the most powerful media platforms in the entire world can institute a change like this under everyone's watch, and none of us have any recourse.

A couple of programmers in their two-week sprint can fiddle around with some Javascript and push it out to the masses, and suddenly large swathes of the historical record are gone forever. (If you consider downvotes to be historical, that is.)

Imagine if someone could white out a section of a book and apply the changes instantaneously to every printed copy of that book in existence.

I think about the implications of being able to look back on human history, and how that historical narrative is increasingly in the control of technology companies and a relatively small amount of people who decide to change that narrative for everyone else. I remember many chronological videos that specifically call out the dislike ratios of some YouTube videos as evidence of the negative backlash certain companies or individuals have received. With no dislikes, that point can no longer be made. That information is lost.

During a live recording of a Japanese entertainment group I watched, whose videos will be preserved on physical discs and be watched by thousands for decades to come, they announced on camera that, in the span of their two-hour performance, they had reached trending status on Twitter. All they had to mention was the word "trend," and the audience responded with enthusiasm. The influence of Dorsey's creation has permeated so many of our lives, including the people who, ten years ago, would never have cared about what bizzare-sounding technologies like Ruby on Rails would eventually enable them to obtain: a universally understood, instantaneous signal of approval.

I believe that the record of who we were and what we did will be important for us to be able to look back on, as it has been for centuries. But it seems that, in the coming decades, that will not always be up to us to decide - just a couple of people in the engagement bureau who decided for us that the alternative was better.

[Valakas]

Totally agree with you on this. We're reaching a point where the services some private companies provide are becoming too important to remain private, or at least, being served in a monopolistic fashion where the user has basically no power.

[mygo]

> Imagine if someone could white out a section of a book and apply the changes instantaneously to every printed copy of that book in existence.

Considering a fifteen year old YouTube video can resurface and the millions of votes can change the vote-ratio, digital media isn't necessarily the same as a permanent historical record.

[retsibsi]

Fair point, but I think downvotes have been useful to me relatively recently. Fake videos may not show up in recommendation feeds, but they still appear in search results don't they? (And if you're searching for something relatively obscure, you can realistically hit the bottom of the barrel, rather than only seeing the more highly rated results.)

[mrkramer]

>Fake videos may not show up in recommendation feeds, but they still appear in search results don't they?

Use report button

[retsibsi]

I would for anything truly extreme (like some horrible shock footage in the middle of a supposedly innocent video). But I don't know if all of these videos actually violate any rules; often they are of the kind <title: THING YOU'RE LOOKING FOR>, where the content is actually someone talking about the thing I'm looking for.

edit: Or another relatively common type, not necessarily fake or even strictly misleading, just cynically low-quality: a video made by pulling some text content from the web, feeding it through a speech synthesiser, and playing that over over the top of some 'relevant' images.

[mrkramer]

When you are reporting a video there is "Spam or misleading" option but then again you are left with YouTube algorithm deciding whether video is really "spam or misleading" or it is malicious report.

For the past couple of months I was regularly reporting group of spammers spamming links in comment sections but to this day I still see them running amok.

[8note]

Given its Google, I assume those get forwarded to /dev/null

[HenryBemis]

From a YT standpoint, if I am looking for "interview of HenryBemis" and there are 10 likes and 1000 dislikes, it means that this is probably NOT the video, but someone commenting on the video (e.g. I wanted to see the Oprah-Harry-Meghan video).

If YT removes the dislikes from the public eye, I will either have to scroll ahead, OR watch 5-10mins waiting for the thing to start, then go and try some other video.

YT increases user engagement (I have already watched 10mins of one video and now heading for the next.. and the next.. and the next..). So, for YT this is a win-win.. it's the user that gets tricked.

Bravo YT, well played (not!)

[retsibsi]

> if I am looking for "interview of HenryBemis" and there are 10 likes and 1000 dislikes, it means that this is probably NOT the video, but someone commenting on the video

This is the sort of thing I had in mind. Some of it is complete junk and some is probably interesting to some people, but the titles are sufficiently misleading to attract an eye-catching number of downvotes. I can't remember exactly when I last came across this kind of content, but it was definitely quite recently, so I don't think youtube has stamped it out.

[Sodman]

You see this very frequently in DIY how-to videos. There's usually about ~30-60 seconds of very helpful instructional content, buried somewhere towards the end of an 11 minute video of rambling and common knowledge. I suspect largely because the monetization gets better if your video is longer than 10 mins...

[aquadrop]

Dislikes are good not just for identifying bad videos, but also good videos. You can see it in the like/dislike ratio. The higher the better, videos with ratio around >98-99% have very high probability of being good etc

[lightgreen]

Comparison of likes to views works well enough. Actually, they don't even need likes, they can understand good and bad videos just by analysing how long viewers spent viewing the video.

Likes (and dislikes) are needed mostly for viewers who might want to feel helpful.

[nitrogen]

View time isn't adequate if a video isn't obviously bad until the very end though.

[lightgreen]

Good videos which end terribly are exceptionally rare.

Dislike mobs do more harm. They probably ignore dislikes already.

[duskwuff]

> Good videos which end terribly are exceptionally rare.

1. Videos which promise to show something that never actually appears

2. Very short videos which are difficult to not watch in their entirety

[aquadrop]

I don't need "them" to analyse it, I want it for myself.

[duskwuff]

> I mean, there was once a HUGE glut of videos with misleading thumbnails/titles

Oh, that's still happening. It never stopped.

[motbob]

If the problem still exists, it's way less noticeable than it was in 2010.

[duskwuff]

It's still absolutely rampant for some types of search terms. For example, try searching for "electronic projects", "crafts", or "life hacks" in an incognito window. The vast majority of the results for all of those search terms have misleading thumbnails and/or descriptions.

[retsibsi]

Oh if you're going back that far, then yes, it definitely still exists, though I don't doubt it used to be a lot worse. (I phrased my first response carefully in case there had been changes in, say, the past few months that I hadn't fully noticed.)

[tjoff]

The reason downvotes didn't solve that issue was because you couldn't see the ratio until after you had clicked.

There was plugins that showed it when browsing and it made such a difference.