I made a trailer/Preview

Why make a trailer almost a month after releasing the full video, you may wonder? It has to do with math. Algorithms that is. YouTube algorithms. Well, that, and because the video is the length of a feature film, by Gad, a preview is helpful, if not necessary.

As a Preview

I’m using Filmora X as my video editor. Incidentally, I recommend it. Once you get the hang of it, it’s powerful, and the process is more intuitive and comfortable than a couple other video editors I tried. Well, it has a new feature where it can automatically generate a 60 second trailer. I tried it, and for my particular video and tastes, it sucked. So, I decided to make my own, and without the one minute constraint.

The goal was to show that it’s not your garden variety art history video. The visuals are layered and never strictly perfunctory. There’s a conspicuous aesthetic element and fairly consistent style. And I need to show that it included video footage of experts opining, as well as analysis of written articles. I included a few zingers, and hints of my more extravagant visual experiments. I could describe it more, but the video is only a few minutes, and it’s worth watching.

The Algorithmic Hammer

People who are serious about making money off of YouTube videos tailor their content to appease the Algorithmic Gods. I’ve looked at the research on how to do this, and largely disregarded it. The whole point of me bothering to make a video (or art, or a blog post] is to do something different, more personal, and that doesn’t follow the fashion, or subordinate itself to market imperatives, or dominant belief systems.

But most video-makers who address issues of art — art history is the most relative in this instance — cater to the template, both in terms of presentation and subject matter. Best to keep your videos around 15 minutes or less (you’ll see why], and hit popular topics. Multiple people made videos about Frida recently, as well as Hokusai’s “The Great Wave”, and Van Gogh’s “Starry Night”. This isn’t coincidence. One of the tactics of getting views is to find a popular video, do something similar, and make a few tweaks so it will appear near the other popular video, or even above it, in suggested videos, along the sidebar, or searches.

I don’t do any of that crap. My biggest mistake was to make a 2 hour and 24 minute video, but not because it was too long. That appears to not be the real problem, because people who are interested in the topic will go the distance. The issue is how it makes the video rank in the unconscious mind of the algorithm. Behold this revealing graph:

There are two critical stats here, one is outstanding, and the other is shit. The shitty one is that the average person only watched 13.7% of my video. This is because around 50% of people didn’t go beyond the first minute. That sudden waterfall dive is common to all my videos, and to my understanding is typical for YouTube videos across the board. It’s people who click and click away, or just click to see what it is and come back later, etc. After that initial drop, however, there’s a very gentle slope to the end of the video. This signifies that people who got beyond the initial clicks and samplings watched a good deal of the video (perhaps watching it in more than one session, which also counts against me]. I’m not very good at math, but I am at logic. If your video is 10 minutes long, and you lose 50% of your viewers in the first minute, they at least watched 10% of your video. But if your video is 145 minutes, if someone watches one minute, it’s not even 1%. So, long videos will be slammed by the mere fact that virtually all videos start with a nose dive of people clicking away in the starting gate.

The other statistic is superb. The average visitor watched nearly 20 minutes of my video. Viewers have watched the video for a cumulative 520 hours. If the goal was to keep people on the YouTube platform, it would be a smashing success, holding peoples’ attention for a third of an hour a pop.

Unfortunately, YouTube doesn’t care about the second statistic, but only the first. And it doesn’t care that the video is 98.45% up-voted: 68 up-votes and one disgruntled down-vote from an inveterate pud-whacker (possibly a moderator on reddit]. The algorithm determines that if people watch, on average, 13% of the video, it must be horrible. And so YouTube doesn’t recommend it much, nor suggest it, not rank it highly in search results. When I logged into Chrome as a “guest” and did a search in YouTube anonymously — so it wouldn’t tailor my search to favor my own content — my video did not come up at all in a search for “Salvator Mundi”, even though videos with as few as 3 views and 0 likes did.

Short videos are better for ad placement at the start of the video. A long and successful video works at cross-purposes with ad delivery, and profits. The more people watch my 2.5 hour video, the less ads they are clicking on. I suspect that also has a lot to do with it. Whether a video succeeds or face-plants on YouTube is up to whether they, or rather their algorithm, decided to promote it or not.

My videos soared for a while when YouTube opened the flood-gate, but then hit baseline after they closed the valve. I gather, by extension, that whether one is successful on multiple platforms depends on the grace of big tech and their various filtering mechanisms: how much attention they determine your content will get.

In short, YouTube buried my video alive, dead on arrival. Of the 1,500 views it received, more than half were “external”, meaning from my blog, Facebook, Instagram, or reddit where I directed the traffic myself [of course not in the art history subreddit where I was banned by inveterate pud-wackers and furious fappers].

So, my video formula was a recipe for failure, and that’s why one of the most successful art history channels is called “Great Art Explained in 15 Minutes”. I like the guy’s videos, though he does stick rather closely to wholly established views, in which case there are no surprises for me. Frida is good. Warhol is good. Hokusai is good. Starry Night is good. Duchamp’s “Fountain” is good. I mean to say the formula is very carefully crafted to sync with the algorithm, but the content is also well above average, if entirely predictable.

And thus I made the trailer as a possible way to bypass the algorithm that hates long videos, and direct people back to the main video. So far it hasn’t worked. And it may be that the 13% average viewing time has afflicted, on average, all of my channel, dragging down the average for the other videos, in which case all of my content is virtually invisible unless you go looking for it very specifically.

This may change over time if the video gains more grassroots interest, in which case the people clicking on it know what they are doing, and intend to engage with it for more than 1 minute.

Only time will tell. But for now, no matter how much people like it, I shot myself in both feet and one nad by disregarding the formula and the algorithm’s indifferent and unconscious judgement.

Eh, I’d rather make a good video than a successful one. And 1,500 views isn’t bad for the 6th video on a channel, though usually one’s 6th video isn’t as ambitious.

~ Ends