How to succeed with not very interesting videos

I think I've cracked it.

I've taken "optimising a not particularly interesting QBASIC game", perhaps one of the driest and most unappealing subjects I could have found short of the British summer of 2026, and had two hits with it.

Perhaps the second is more "respectable for a channel of your size" rather than "breakout", but then as a certain model railway enthusiast once said, "you can't top pigs with pigs". Even the Pass The Pigs score sheet tells you it's impossible.

This comes on top of two Sawyergame-related videos which have also done ridiculous numbers for me. Even despite my incredibly sporadic upload schedule. The difference this time is unlike the other times I've had breakout hits, I can point to why these have been hits and other notionally similar videos in the past have not been.

The short version of this is "don't do things which piss the least dedicated audience off" but if you're coming here for a short and to-the-point version of anything then... it's nice to see someone new for the very first time but I fear a sense of crushing disappointment is about to ensue.

Chartsengrafs

YouTube's "Audience by watch behaviour" is one of the most vitally important metrics to pay attention to in understanding why videos do what they do. Here is a big success:

And here is an utter bomb in comparison:

This second one was already off to a bad start, since regular viewers, the people who watched my videos in 6 or more months of the previous year (and given my release schedule, that's people dedicated enough to re-watch old videos), weren't really that interested.

YouTube was actually pretty forgiving with this. It did whatever the uncaring machine equivalent is of going, "OK, but you've made decent videos in the past, maybe we should try with the casual audience?" (Casual: watched videos of mine in 1-5 months of the previous year). Not that much of the casual audience, since the feedback from the regulars wasn't great, but enough that it had a crack at the new viewers. And it died there. Nobody was interested in the topic, and since by definition new viewers don't know who I am, the presentation style isn't a draw either.

Compare to the success, which followed a much more conventional path: regular viewers absolutely loved it, then casual viewers also enjoyed it, but the drop-off from there to new viewers was not much at all so the impressions just kept coming. There's not that many more regular viewer impressions than the bomb, but the casual and new impressions in comparison are massive.

Based off this the conclusion might be "CTR on the less successful video is crap, so repackage it with a new thumbnail and title and you'll be off to the races" but I need to pull in a graph at this point.

Success:

Not success:

Some of the spikiness is due to lack of data from just not that many views overall, but even the weeks with reasonable viewcount follow the same pattern. What's surprising given the usual view of the algorithm is YouTube is really trying to find this one an audience. After it completely bottoms out in late October there's a spike in both views and the duration shown here, but it doesn't take long for YouTube to run out of lookalikes and it dives again.

Put simply, people gave a lot of signals that they really did not want to watch this video and eventually YouTube's digital auditors agreed with them. There's people that loved it, indeed watched all or nearly all the entire thing, but as soon as YouTube tried to turn that into a wider audience the average view length went plunging back down to a tiny fraction of runtime. There just weren't enough people who wanted to watch this.

People

The most useful advice I got about YouTube came from a video I watched a couple of years ago which said, simply, "whenever you feel yourself about to use the words 'YouTube algorithm' substitute that phrase with 'people'."

It took a long time for that to sink in. And even longer for the mode of thought that summation is supposed to engender to kick into action. The shift in thinking is to stop considering YouTube's recommendation process as some vast conspiratorial empire which arbitrarily picks winners or losers and delights in crushing the spirit of the everyman. Instead, it's just something following how people respond to your video. If they do things which indicate they're not interested or not having a good time, YouTube assumes other people will also not be interested or, worse, will not have a good time.

Remember, YouTube is in the business of not getting sued by media conglomerates with enormous teams of copyright lawyers. But just after that in their list of priorities comes providing an experience which has viewers hooked for long enough that they watch multiple adverts or, even better, consider the service high-value enough to purchase a subscription. And then keeping those advertisers on board. I think a lot of what feels like creator hostility is simply us being fourth in line behind the people who pay the bills, the other people who pay the bills, and the people who could sue them into oblivion if it looks like they've dropped the ball on copyright infringement for one solitary moment.

I saw an interesting comment about a non-YouTube video service the other day, which was "they've gone so far in favour of what creators want to happen that it's actually quite a bad experience for me as a viewer". That's what they're avoiding when they kill your video because five people on the same day clicked away at the point where you stumbled over a word twice and then said "er".

Yes. People. YouTube's algorithm didn't coyote-into-painted-tunnel your video's momentum for some ludicrously petty reason. People did.

Impatience

When I watch people I know and have met in Actual Real-Life Situations on YouTube I am one of the most forgiving viewers around. I will patiently sit through slow sections, shrug off all sorts of vocal fluffs, maybe even let a research error or two slide, and even in extreme cases sit there watching something I'm not enjoying because I don't want to wreck their engagement stats.

When I watch channels I regularly view I'll give them a chance, and maybe they might get away with a few dull videos or some questionable editing here or there, but eventually if they make enough rubbish or do something like decide "we're doing AI thumbnails and AI-generated musical skits now" then I eventually drift off and realise after a while YouTube has noticed that and I don't see their stuff any more. But largely I am a reasonable man who understands these people are only humans trying their best to entertain me.

When I watch things that are completely new to me I have the patience and level-headedness of a toddler. Who's just dropped their ice cream. I am the one who thinks, "that's three times you've rephrased the same sentence without getting to the point" and is immediately back to the home screen. The one who goes, "would it kill you to just say what the game is rather than trying to clickbait people" and clicks Don't Recommend Channel. The one who goes "did you even listen to your own audio?" or nopes out because someone held a shot for 12 seconds longer than my attention span. Not that I would comment to that effect, I'm not a monster, but I will banish from my electronic kingdom all who needlessly pad videos, engage in a journey without a clear narrative destination, or just feel a bit tedious and tryhard.

All of which I have done myself. Even if the one time I forgot to mention the game in the title was an accident.

What I claim to have "cracked" is the realisation is that toddler-brain me, the supposedly grown man who is so infantile he will cancel you for excessive use of the pole_swipe_whoosh sound effect... that is who is in charge of the success of my videos.

YouTube starts the regular>casual>new ramp up early. If the performance with early cohorts is strong enough, you're already being tested with a completely new audience before even 5% of your regulars have seen your video in their home screen. In my bomb example, YouTube skipped the regulars entirely because the casual audience response was much stronger. But then it hit new... the toddler brain.

Plan 9 video: no hook in the first 20 seconds, not clear what this is about -> 💀
First Encounters video: irritating shaky camera -> 💀
Wreckfest video: no actual Wreckfest in the first 30 seconds -> 💀
Black & White video: who the hell has 51 minutes to watch this idiot? -> 💀

These are competently recorded videos with solid scripts, good audio and a decent level of visual polish. Doesn't matter. You made me drop my ice cream! Fail consistently with that new audience and YouTube will eventually decide this one is for the casual viewers at best or maybe even only for the regular viewers. Maybe it will try with a new audience at some point in the future, but it's entirely likely it won't.

All because of one stupid thing that got enough people to instantly quit. It didn't matter that in all those cases it was something small and with a bit of tolerance and patience those viewers would have found something they enjoyed, although that Black & White video does get mighty repetitive 40 minutes or so in. When I am a completely new viewer my tolerance and patience are zero until the creator proves themselves to me.

Then we get to the Transport Tycoon Deluxe retrospective and the Locomotion retrospective and they don't have that immediately obvious "well fuck this guy" moment within the first 30 seconds. They're not perfect, but it's clear this video has a premise matching with the thumbnail/title and it's going to deliver on it.

Same with the "Was QBASIC just too slow to write games?" vid. It's instant - first 30 seconds and it's clear this video is what the thumbnail and title said it would be about, you're straight into the story of developing the code. The toddler brain is pacified, at least temporarily. Pacify it long enough and you might eventually start getting away with stuff, but even then not for long.

If anything, this might hint that the lower performance of the most recent video is not purely a pig-stacking issue: it goes into a history of PC graphics cards and it's 5 minutes before we even get to the details of the VGA and nearly 9 minutes before any code appears on screen. Any reasonable human would go, "well yes, but it was an entertaining and concise history and it is related to the code which eventually gets produced". Toddler brain: I wanted to watch programming and there's no programming!

Critic

This is what I've realised. I've made my peace that I'm doomed to make videos that are generally OK. It took a friend telling me during the recent extended hiatus words to the effect of, "you're a well-spoken and an entertaining chap, you should be making videos on whatever interests you" but I have finally accepted this is another thing I tried to just enjoy without getting any good at it and failed. Perhaps I should muse about that the next time I'm halfway down a mogul field and enjoying myself in a hobby that was only supposed to be me cruising lazily down blue runs on the way to an après-ski bar.

If I want to make videos that succeed as much as the chosen topic allows them to, I need to change the way I look at them while I'm recording and editing them. I need to stop looking at them in Hey I Know That Guy mode and watch them in impatient toddler brain mode. Learn to go "well that bit's going to piss me off and make me click away and vow never to come here again". Tell a coherent and well-paced story that I would accept from some channel I've never heard of that has already irritated me in 3 different ways at the point I've reluctantly clicked on the thumbnail because I do kind of want to know about that topic.

Also probably make more than about one video every six months. One day I'll manage to act on that. Maybe.