I want to be crystal clear for any Alphabet lawyer or majority shareholder (Google’s and YouTube’s parent company) who reads this, that I am NOT referring to any human being associated with those companies. I specifically mean that the machine learning or AI bot now in charge of the search and content recommendation function of YouTube is heavily, heavily on drugs. Hopefully it’s the fun kind of nose candy, but I suspect based on recent performance that the algorithm is flirting with more nasty Opium based AI mind funk drugs like Junky Skag, China White, Golden Brown. This blog won’t turn into a video, so it may never be able to track and scan this content to know I’m talking smack about it, and smack is just another term for the type of drugged out haze this artificial brain function has fallen into.

“Well, that’s a hell of an opening statement from a woodworking guy who doesn’t know anything this type of technology”, I picture some lawyer saying. “Oh, but I do”, would be my response.

“Hey isn’t this a woodworking blog?”

“I know. I know. It will get into a woodworking topic later. I promise”

I wouldn’t go so far as to say I’ve actually written code that went into any self learning chunk of software. However, in my former work life I evaluated vendors and products, done side by side performance testing and results evaluation, installed that software and pre-loaded the algorithm database with starter sets of information it uses in its own search engine. Then, I watched it gather more information, grow its database and invoke its machine learning algorithm while it tried to answer the needs of the requester. And then, I sat in meeting rooms while mid level managers discussed exactly how many customer service call center people could be cut from payroll as the new (Pre-AI) learning software took over the service chat and “help” function of their business. I mean, if a customer was really persistent and held out through several iterations of digital Q&A, you could speak to a human but that was always accompanied by, “Thank you for your patience, you are the 17^th caller in line to speak to a representative”.

Sort of makes you want to break out an edible just to cope with the frustration now doesn’t it? Maybe take two.

Your YouTube Experience

I’m not just a content creator; I’m a frequent consumer of the YouTube platform myself. I’ll hit it heavily during the day for entertainment now that I’m retired, and I might open it up on my tablet at night, watching a longer woodworking or car rebuild video while fading off to sleep. The all knowing and controlling YouTube algorithm uses what it knows about my user profile and my previous search history to feed me new things it thinks I want to watch. Except, there are some unwanted software behaviors happening in the background. You don’t always get what you ask for or expect based on your search and view history.

To prove that, all you have to do is look at your list of subscriptions and think about recent videos that have popped up on your feed from those channels. You haven’t seen a video from some of your channels recently, have you? Over the last 10 years, I think I’ve subscribed to more than 50 channels because I like their particular type of content. But, my tastes in entertainment go pretty wide, from woodworking to engine repair, to classic car restoration to home building, boat building to European travel, cooking to stand up comedy. Even comedians who cook in campers. (A shout out to Seinfeld’s car series). And yet, I never see video recommendations from many of my subscribed content creators anymore. I wondered if they stopped producing content. Nope. They are all still there making cool videos. I was forced to search on the channel manually of click on my subscriber listing to find them again. Why is this?

I have a couple theories as to why the algorithm is no longer consumer focused.

The first and most obvious answer is that it has to be about the money. YouTube gives preferential treatment to those channels and videos who participate with Google Ads Advertising campaigns to push those videos out to a wider audience. They call it “Targeted” advertising of your video to a specific demographic. In effect, you pay to open up a video to more eyes in the “HOPE” of getting more clicks. There’s no guarantee of a return on your investment here, but more people might see your thumbnail and find your thumbnail interesting enough to click on. The bigger your campaign, the more widely your video gets presented to the audience. It’s simply pay to play.

Except… there over 100 million active channels on YouTube with many of those channels paying to push their own videos. I suspect that the smaller channels with modest advertising campaigns tend to feel like minnows in a lake filled with hungry Bass.

My second theory is that AI is just stupid or whatever the digital equivalent of being on hard drugs is like. First it takes the easy road to recommending videos based on how much money the company makes through those recommendations. Duh. Zoned out drug users pretty much always take the easy road when a task is at hand. Rather than doing the work to review your past subscriptions and your wider interest in all sorts of subject matter, it will just flood you with video recommendations based on your most recent click.

Example: I made the mistake of taking an interest in a video thumbnail that talked about a recipe for making hard tack bread and something else called “Pemmican”. These were old fashioned methods for making camping or traveling food that would last in your saddle bags for months without needing to be preserved or refrigerated. They looked disgusting and unappetizing but it was a cool little historical video about survival staple foods when you’re on horseback out in the wilderness. Note the key word in that last sentence being “survival”. YouTube woke up from it’s opium based fog and took the easy algorithm path to flood my video feed with all sorts of nut ball survival channel videos. Dig your own bunker, doomsday prepping, how to properly stock dry ammunition, storage ideas for months worth of freeze dried food, how much camo clothing is too much camo clothing; the list went on. For the next several screen refreshes, 30-40% of all recommended videos in my feed were about doomsday prepping for an upcoming apocalypse. I’m assuming these people paid heavily for ad campaigns.

Back to Woodworking

So even though I’m subscribed to twelve different woodworking YouTube channels, I don’t get notified any longer when one of them publishes a new video. If I do see a familiar channel in my feed, the thumbnail is often way down on the list requiring me to scroll and then spot them. They never seem to pop to the top of my feed, as many other topics and unknown channels take over the top spots. I suspect that this has been my fate as well. Where a well produced woodworking video of mine would easily grab a thousand or more views on the first day of publication, that was years ago. Now I’m lucky to pull 200 views in 24 hours, even though the topic of the video might appeal to a wider audience. Then again, I’m not paying for clicks.

Woodworking is a niche topic area to be sure. It doesn’t appeal to teenagers. It’s isn’t sexy though there are a few channels out there now where the woodworker knows how to strike a click worthy pose. We used to be able to count on appealing to our subscriber list whenever we publish something new, but that just isn’t happening. With more than 22K subscribers now I wouldn’t expect all of them to be interested in every video topic I publish. But, when I can’t even pull 1% of my subscribers to check out a new video (from a channel they subscribe to), then I know the YouTube AI guy is baked.

Sadly, it’s all about the ad money and about the crappy AI customer experience. I am interested in all sorts of things, that’s true. I like variety in my video feed but appreciate seeing new content from the channels I subscribe to. Why subscribe at all when you have to channel flip and search to find a channel you already told the algorithm you wanted to see?

Last tidbit. I had published a video a couple years ago about repairing and finishing a broken cedar canoe. On a recent search, it doesn’t even show up in the top 30 video listings on YouTube, though it got way more views than others in the list (with a better thumbnail and better video quality). Wanna know what pops up on the top of the search list? See below.

I need a beer. Maybe a gummy.