He's also still giving some details on his twitter account, but as last time with the tutorial, he contradicts himself in several ways.
He talks about how it's not just about statistics, then come back about statistics, makes a weird shortcut about how games with smaller rosters sold better than those with bigger ones (which seems like a weird jump to conclusions), and mixes up regular Tekken and Tag (the latter typically selling to fans rather than a bigger audience).
Not really. In either case he's still saying it's about analytics, but he's saying that their are more nuanced factors than just usage in one region etc.
If anything, you are making inferences about the demographics and target audiences of Tekken games, which you have no data on (saying Tekken Tag, isn't the same audience as regular Tekken).
It's not really that complicated. He's saying they tried to include the most popular characters, with some consideration to which were the most popular, within each region. Obviously there are other, logistical challenges too, like the dev team only having time to work on a limited number of characters. I think in general, there isn't anyone in the Tekken 7 roster that I anticipate is likely to be seen as a wasted slot.
Speaking more generally. I appreciate Harada's openness about their process and I think the negativity that's often associated with the use of analytics, and other user research methods such as playtesting often dissuades most developers from offering that level of transparency. I think their are a lot of misconceptions about how user research and data science influence game development, and in general, the gaming community (NeoGAF and similar) help spread those misconceptions resultant, misplaced negativity.
Wouldnt usage data be dictated by a characters tier? If a character is weaker than ithers, even if you love them, it wont be used as much. Doesnt Harada and team dictate control this? Essentially, they are controlling who gets used the most due to tier.
Kind of nonsense.
Just to speculate, but I doubt it's as clear cut as Hareda described in a tweet. Their are statistical methods you could use to examine character usage while mediating the impact of other variables such as win-rate. Additionally, while tier lists are a significant factor in competitive tournaments, they often don't have the impact you might suspect on the community at large. As an example, Ryu and Ken are pretty crap in SFIV, yet they were still the characters you would see most commonly online. Same with characters like Juri in SFV, Juri sucks, rare at high level competitive, but she appears common at low and mid levels of play (obviously, just my own observations).
Additionally, Hareda admits that if they weren't popular enough, it's his studios fault. That means, he's aware that the design of the character, including their overall efficacy, has an impact on selection rate.
That's silly because you are completely dismissing what the consumer wants. Of course characters, modes etc costs money (they had years to produce this title after the arcade release anyway) but I believe they didn't include it due to ”statistical data". That's just an excuse. They didn't include these things because they were lazy and are now backtracking with ”internal data" due to the backlash.
Perfect example. You don't have a clue, and yet you assert the developer was 'lazy'. 100s of members of staff likely worked through crunch to even get the game released. People need to stop with this nonsense.
And yet there are not one but TWO bears in this game. In separate character slots! No one likes them!
Likely because the bears overall are popular, and the overlapping moves makes it very easy to put two in. If Kuma already existed, adding Panda is no where near equivalent to the work of that adding Jill would be, for instance.
