A/B testing and measuring -- are you kidding me? That's how you go from 35% conversion to 36% conversion, not from 2 daily active users to 1e7 daily active users. Refinement only works if you're exploring a space that has productive alternatives in low dimensions, which is almost never the case. No amount of "progressive refinement" is going to turn Farmville into Candy Crush or Candy Crush into Fortnite. You have to look no further than those examples to see what I mean -- "Oh, look, Zygna has figured out a formula to hook users" followed by "Zygna can't repeat its success".
A developer of traditional games might think, "How can I make this game better in an aesthetic, or an educational sense?" Or, "How can I make this game a better FPS than the ones before?"
A developer of another game might think, "I want to make money, let me survey existing games and read psych literature and create the next billion dollar game. I don't care what form the game takes, or what players get out of it."
The claim is not that game developers for each game are creating an addictive game from scratch by A-B testing, but rather exactly what you said, that they're AB testing for smaller improvements in eyeball-retention, spending, and addictiveness, because all they care about are those metrics which drive profits (or valuation which can be converted into profits).
That a set of conscious design choices, on its own, isn't sufficient for extreme addictiveness doesn't change the fact that those conscious design choices are necessary, and those choices are being examined, analyzed, and documented by other game devs and by psychologists. So, of course you can take all the known lessons and apply them to a brand new game concept and fail horribly, but if your game is a success on other grounds and you mined other games and the literature for addictive ideas, you are much more likely to have a billion dollar super-addictive game than you would have been 20 or 30 years ago.
Zynga had repeated success, untill market saturated and other developers caught up. That does not make these tactics not working, that just makes them so common that they are not such a huge advantage anymore.
I'd say, until people got bored with the same single local optimum game that all Zynga games converged to despite differnt skins.
It's also how you go from unique and successful entertainment product to something completely unusable and boring is just few hundred A/B iterations.
Enjoyment is not the space where gradient descent works particularily well.
Amazon is absolutely horrible expeirience for me. It's hard for me to point e-commerece site I used that's worse than amazon.
The notion that human behavior can be manipulated and shaped by careful modulation of input is not remotely new, and it's developed considerably over many decades.
If you don't agree with or understand this, and would like sources, please say so and I'll gladly offer them when I'm at a keyboard.
Nobody fell down the stairs and discovered that the tumble accidentally linked their KPIs to engagement metrics.
And a thousand of those .1%'ers will try a second time, and the .1% that succeed that time will be hailed as geniuses and people will set about copying their open plan offices and terrible management styles in the hopes of replicating their success.
But I've played some games. I am 137% confident that some form of "level up" (roguelikes, MMOs, MOBAs, FTL) is an extremely strong addiction factor. League of Legends, that free game whose net worth is among the highest of all time.
In real life, you study->grades->job->promote->takeoverworld. Or liftweights->liftmoreweights->liftmoreweights. Or climbmountain->climbhighermountain...
In games, you do that in 1/100 the time and effort. Tap a few buttons and "you win!!!!"
A sweeping statement to the effect of "people don't understand addiction" is basically malicious and I accuse you of being the inventor of gamification or one of their PR folk.
Try playing more unsuccessful games. You'll find that most boring, unaddictive, terrible games that you'll never play more than once have the same leveling up mechanic.
Nothing is binary; I'm not going to say "either leveling helps a game all the time or it doesn't". Sometimes leveling-up mechanics help a game, sometimes it doesn't, sometimes it alienates one set of users and attracts another, and there's no magic formula for when it helps and when it doesn't, and the nature of the leveling-up mechanic itself and how fun that is can vary in effectiveness dramatically. It's like having the user's avatar wear a funny hat.
There is just something about the "Skinner box"-style of frequent microrewards in games that scratches an itch in the human psyche.
Well before then players of the original Diablo were obsessively clicking cows because of rumors and belief that there had to be something to the cow.
This is ridiculous. I mean, utterly and completely ridiculous and you should be ashamed of yourself. This all happened by chance, not by some miracle of data science. For every product that captured 6 year old brains, there are millions of failed products that did not capture any brains at all. For every claim of "hey, I just tried to stimulate dopamine addiction in my successful table game" there are a million people who tried to do the same thing and failed. This is just a ridiculous narrative that uses post hoc reasoning and ignores survivorship bias to try to justify some parochial view of what makes children become "productive" adults.
No human is going to be able to predict what will be good and bad for child development except in very narrow cases around health and nutrition (and even there we're wrong 92% of the time when we stray out of those narrow bands). The world of tomorrow is not the same as the world of today and each generation will adapt and thrive in their own way. We'll keep stumbling through trying different crazy approaches, each with drawbacks and advantages, and occasionally pushing for the "one true" approach that will be discredited before we ever get to figure out if it works.
> As much as we want it to be true, technological advancement can never be separated from social/ethical impact
Yeah, sure, of course. But responsibility for social/ethical impact is almost zero in all cases except as a retrospective tool, and trying to predict social/ethical impact is a suckers game because this is simply not possible. This idea that we can somehow have a simplified model of how humans behave and that such a model has any validity at all is a common one in marketing and social science research, but is clearly fallacious.