zlacker

[parent] [thread] 13 comments
1. mattma+(OP)[view] [source] 2024-09-16 00:24:05
I don’t even think he’s gaming an algorithm. He doesn’t have to.

He’s just making videos people will click on and then watch.

It’s almost like he’s trying to make something people want. I’ve heard that before somewhere…

replies(1): >>simonw+d1
2. simonw+d1[view] [source] 2024-09-16 00:40:57
>>mattma+(OP)
If you read the full PDF it’s clear he is very carefully gaming the algorithm: he includes charts showing exactly when people drop off from watching videos, and explains how he has an exact set of rules for how the thumbnail, first minute, 2-3 minutes, 3-6 minutes and 6-end minutes of any video should work.

I find the lengths he has gone to in order to design his videos specifically for how YouTube works to be extremely impressive.

replies(2): >>nitwit+p2 >>gruez+D8
◧◩
3. nitwit+p2[view] [source] [discussion] 2024-09-16 00:58:14
>>simonw+d1
Statistics about when people drop off, or what thumbnail or content is appealing, is studying human viewer behavior. There's no algorithm telling the users to find it interesting and keep watching.
replies(1): >>simonw+25
◧◩◪
4. simonw+25[view] [source] [discussion] 2024-09-16 01:35:01
>>nitwit+p2
Talking about “the algorithm” always feels a bit foolhardy to me because it’s undocumented and constantly changing.

Given that, it’s pretty clear to me from the full PDF that MrBeast is “gaming” it to the best effect possible given no perfect information.

The thing he cares about is if YouTube is going to recommend his video for people to watch, even beyond his own subscribers.

He believes that the key to this recommendation mechanism is having a high AVD and AVP (defined on page 5). Given that he has the highest rated account on all of YouTube now I’m inclined to defer to his expertise.

replies(1): >>mattma+DV
◧◩
5. gruez+D8[view] [source] [discussion] 2024-09-16 02:21:29
>>simonw+d1
>If you read the full PDF it’s clear he is very carefully gaming the algorithm: [...]

How is this different than any other technique to maximize engagement/readership, eg. inverted pyramid format for newspaper articles? It's probably designed to draw people in and sell copies. Is that also "gaming the algorithm"?

https://en.wikipedia.org/wiki/Inverted_pyramid_(journalism)

replies(1): >>simonw+99
◧◩◪
6. simonw+99[view] [source] [discussion] 2024-09-16 02:28:40
>>gruez+D8
“How is this different than any other technique to maximize engagement/readership, eg. inverted pyramid format for newspaper articles?”

Because it’s extraordinarily effective?

He made it to the top of YouTube with it. If it’s the exact same thing as other existing techniques how come others haven’t been able to match his success with those classic formulas?

replies(1): >>gruez+M9
◧◩◪◨
7. gruez+M9[view] [source] [discussion] 2024-09-16 02:37:07
>>simonw+99
Presumably because journalism is centuries old, and techniques like this eventually become "industry standard" and you don't notice it. Once people figure out what the strategy is, they're going to try replicating it to capitalize on his success. Afterwards I suspect he'll still have a first-mover advantage, but he's going to be nowhere near as popular (comparatively). It's not any different than say, the reality show "format" being eventually copied by other production companies/networks.
replies(1): >>simonw+V9
◧◩◪◨⬒
8. simonw+V9[view] [source] [discussion] 2024-09-16 02:39:04
>>gruez+M9
Did you read my summary and not the full PDF?

If you’ve only read my summary then we are discussing this with completely different mental models of what he actually does.

replies(1): >>gruez+ea
◧◩◪◨⬒⬓
9. gruez+ea[view] [source] [discussion] 2024-09-16 02:42:15
>>simonw+V9
I skimmed the summary and it describes every aspect about his production company, whereas your "summary" only described one aspect (ie. figuring out how to keep engagement up), so I only responded to that. You can't treat the entire document as "gaming the algorithm". For instance, the document also mentions only hiring A players, which could hardly be described as "he is very carefully gaming the algorithm".
replies(1): >>simonw+ma
◧◩◪◨⬒⬓⬔
10. simonw+ma[view] [source] [discussion] 2024-09-16 02:43:34
>>gruez+ea
Go and read pages 6 through 10 of the PDF: https://drive.google.com/file/d/1YaG9xpu-WQKBPUi8yQ4HaDYQLUS...

They do not describe the same process everyone else uses to make content. They are much more specific than that.

replies(1): >>gruez+Oa
◧◩◪◨⬒⬓⬔⧯
11. gruez+Oa[view] [source] [discussion] 2024-09-16 02:49:44
>>simonw+ma
So far as I can tell his "gaming the algorithm" is having a few short clips near the start to hook people in (ie. an summary/abstract), and periodic bursts of excitement to keep people engaged. The first is so banal that it's hardly worth discussing. Articles in scientific journals have abstracts/summaries. It's not anything nefarious. The rest seems like standard narrative/storytelling advice, eg. hero's journey[1], or how broadcast TV shows have cliffhangers/plot developments to get people to watch the next episode or ad break. Do you think 24[2] is "gaming the algorithm" by presenting 24 action packed episodes where there's always some new/unresolved plot point at the end of each episode?

[1] https://en.wikipedia.org/wiki/Hero's_journey

[2] https://en.wikipedia.org/wiki/24_(TV_series)

replies(1): >>simonw+6d
◧◩◪◨⬒⬓⬔⧯▣
12. simonw+6d[view] [source] [discussion] 2024-09-16 03:18:00
>>gruez+Oa
What he’s doing is very clever and very effective.
replies(1): >>mattma+E41
◧◩◪◨
13. mattma+DV[view] [source] [discussion] 2024-09-16 12:06:17
>>simonw+25
I don’t dispute his expertise, I dispute your interpretation of what he’s doing if you think it’s gaming an algorithm. Perhaps we’re debating semantics.

These are metrics one might use even if there’s no algorithm, in fact historically they have. TV shows used to use Neilsen data for similar purposes long before there was YouTube. TV producers would measure audience dropoff and then use that to help writers write more gripping episodes.

Google’s hope with their search for decades was that their algorithm was ungameable and that the way to get your site to the top of any result was to make it the best. That’s why they made it a black box and changed it whenever SEO caught on and used it to push junk to the top.

That’s had mixed results on the web for sure but it’s probably worked much better with video because you can track these metrics in a way you can’t with text. Also with the web, the page you land on may make Google further money (with ad sense, inspiring more Googling, using a Google product directly, etc.) or it may not, they don’t always own the ad service at wherever you land when you click a search result link. They don’t have the pure financial incentive of just showing you what you want, something you want a little less might make them more money.

With YouTube they own it all. The more you watch YouTube the more they make. You’re only clicking ads to other YouTube videos.

Everybody on YouTube knows you want a compelling lead in to get the click over to your video, a hook to keep them watching, etc. He’s codifying what they all already know and do. He just is better at it.

◧◩◪◨⬒⬓⬔⧯▣▦
14. mattma+E41[view] [source] [discussion] 2024-09-16 13:17:07
>>simonw+6d
Correct but that doesn’t mean it has anything to do with intentionally gaming an algorithm. TV never had an algorithm and some people were a lot better at making TV that others wanted to watch than others.

You seem to be of the belief that for anyone to be the most successful at this field they have to be gaming an algorithm. But perhaps there’s really no algorithm, or perhaps (my opinion) the algorithm is so good at showing people what they want that you can instead just focus on making videos people want.

[go to top]