The idea that they need to keep the death rate so high in order to "gather data" seems really weird to me. I mean its been at 10% for quite some time now. How much data do they need to collect about it? What are they going to learn? That the rate's too high? Just playing the game for one hour will tell you that. Is it really necessary to subject the entire player base to it for so long?
Or maybe 10% is actually the desired death rate and its not gonna change at all.
----------------------
As an aside, this kind of goes with an impression I've gotten from reading various dev comments here and there, that they really seem to rely an awful lot on running simulations and collecting loads of data for all their balancing decisions. Not that that's necessarily a bad thing. Its great to have solid data to back up your decisions, but it is possible to overdo it and rely too much on data in place of human judgement. I've seen a ton of forum threads where players are complaining about some aspect of the game feeling off, and the dev response is something to the effect of "We've run hundreds of simulations that show this is balanced just right." But simulations and data can only tell you so much. A lot of times you just have to go by feel. At the end of the day, all that really matters is "Does this play well?", "Do these game systems feel good?, "Is this fun?"
I was reminded of this by a recent Jason Schrier article in Bloomberg about Amazon Game Studios and how they've struggled to put out any decent games despite having some of the best game devs in the industry and all the money in the world. One big problem is Amazon's data-driven culture.
That's why modders are able to run circles around TW devs and develop major new features in a fraction of the time that TW can do it. They don't need to run weeks of tests for every new idea they have. They just figure out something that they think would make the game better and they put it out there. If it turns out to be unbalanced, they adjust it until it feels right and its done. I think TW could use a little bit more of that approach