A scientific analytical research conducted recently this May shows, that people don’t care about bugs that much. They analyzed thousands of reviews of different games and came to the conclusion, that bugs are mostly mentioned by people, leaving positive reviews. People, that leave negative reviews usually (in more than 80% cases) do not mention any bugs at all and focus on other things. So it can be concluded, that bugs is not the thing, that makes people dislike the game or stop playing it. It is gameplay.
Your implication assumes people who don’t leave a review are fundamentally different from those who do, which isn’t supportable. Analyzing reviews provides a good sample size and an effective cross section of the gaming population.
I never said this line, but thanks for putting words into my mouth.
What I was trying to say, though that these people are clearly more clever than 100% of people on this forum, and are 9000 times more likely to be right.
And this phrase alone proves my point of view soooo very well, because it clearly shows that person saying this didn’t even bother using common sense, before writing criticism. Why? Well because clearly there is thousands of people NOT writing a review and NOT being annoyed by bugs at the same time. So in the end we only care about percentages, not sheer numbers.
You just proven my point and youre right, you never said that line, but you implied it
“They are researchers so they are at least 9000 times more likely to be right” - first of all; this is the argumentation of a 12 year old xD - second ; you shouldnt
believe what “some researcher on the internet says just because it fits your point”
13.000 analyzed steambased reviews in contrast to 1.000.000 copies sold is NOT, in any way, representative and instead makes me question those publications, especially after looking at some of Dayi Lin other work.
Glass house and stones.
That text has not passed a serious review. It’s simplistic in its inference and has no bearing on scientific rigor.
And I’m a fierce opponent of p-hunting, which they did inversely by selecting a sample size matching their wanted power and significance. The frequentist view should never be considered when you have all the data.
It would have been interesting to put the data through an ML model with a hint of NLP to categorize sparse features. That could have said something. Now it’s just something people might quote because they find a sentence they like.
i heartily disagree. science is MEANT to be debated on, have arguments about. just because it’s a study does not mean it is true, it just presents new information that can be assessed and discussed on. There are many times in history where famous theories have been totally debunked, for example the earth at the center of our solar system.
This here is a scientific article that describes “Why Most Published Research Findings Are False”. The notes detail several methodologies and uses a probability model with references to relationship bias, and also presents suggestions for improvement over existing study models. I don’t claim to understand it because i’m not smart, but i agree with the general sentiment that we can’t take everything as truth just because it says it’s a ‘study’. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1182327/
I’m not saying the study you linked is false, i’m saying it’s folly to take it as 100% truth because “SCIENCE!”.
thing is, critical bugs are important to fix as it prevents frustration and will lead to more people promoting the game. the study only studies people who left reviews: what about all the other people who didn’t? it’s only a subset of the entire population. it’s good info, but not something to base the decision to ignore bugs on.
You’re kinda right, in a sense that “internet survey showed that 100% of people have internet”. But I don’t think that’s the case. There is no reason to think, that players that leave reviews differ from average players that much, if at all.
I don’t think they did any p-hunting. I think they just selected a sample size large enough to extrapolate it for the whole audience. I mean why would they need p-hunting, they would get meaningful results no matter what.
You would get a representative sample long before you’re done training the model. Without training the results will be very inaccurate.