Monkey Lover, on 27 April 2017 - 07:48 PM, said:
Im not going to get into if the way they are refunding is good or bad . What I want to think about is how much is this going to hurt them financially.
Im getting back over 500k gxp. this is over 100 bucks worth.
Im getting so many gsp back i will never have to grind again.
They're giving people back consumable MC refunds.
Then we are getting some cbill module refunds.
Tons of people are going to skip spending cash for the mc-xp to gxp converter.
They are going to skip buying few extra mech bays because they got free mc now.
Lastly they're going to skip buying a mech pack because they have the cbill to buy a few other mechs now.
I just can't see this being a good business plan.
TLDR: Not only is it a bad business plan, its a bad plan in light of their own goals
4 Stated goals of the skills tree were originally:
1) Increase mech diversity,
2) Improve player choice,
3) Provide means to aid players into transitioning from a 3 mech model to a 1 mech model,
4) dramatically reduce quirks.
Goals that are implied by more recent statements and the latest PTS announcement:
5) improve long term balance by determining a new base line of mech performance.
6) ensure player retention and satisfaction by providing a refund scheme that lets players of all types maintain their XP and module based "progress" on all their current mech without flooding accounts with excess c-bills with which players can buy mechs with rather than impulse purchase with real money.
Now those are the stated and implied goals.
Does the skills tree and all its accompanying dross (nerfs, decouple, refund, etc.) achieve any of these goals? Are these goal ALL even achievable buy the system proposed? Answer is No, and No.
By drowning the community with HSP and GSP there is no motivation to even play the game, and certainly not to play or "grind" mechs that are in any way less than optimal or meta.
By flooding players with all this artificial currency, the goal of increased diversity is blown (goal 1), improved player choice is avoided (goal 2), there is no need to worry about 3 mechs vs 1 mech since all those currencies ensure it will be no problem to acquire 1 or in some cases 100s of mechs and level them with clicks rather than by play or purchase (goal 3). Right there three of the four original PGI stated goals are tossed aside by this refund scheme. Don't believe me? Consider the poke'mech player or even someone that likes to collect a nice cross-section of mechs. They have now all this GSP just sitting there. No need to buy 3 but so what, they can buy every...say Vindicator variant...with little effort (they may even now have plenty of cbills too) apply some fraction of their GSP hoard to leveling them, and viola: Fully skilled out low tier mechs that they have no need to "grind". Maybe they play once or twice for the novelty, but this player knows Vindicators are garbage tier so why bother? The collection has been added to, the mechs are all instantly leveled. The player can go back to their Hunchback IIc with a smile on their face and the ability to say "yeah, I leveled every Vindicator in the game". Has this increased diversity of mechs being played? Did some in game motivations encourage the player to try something new, to make real choices that affected the game in a positive way? No? Well then there is a problem with a couple of the goals we are trying to achieve here is there not?
Now consider the two implied goals of a new data driven base line, and player retention and satisfaction by refunding all this artificial currency.
If player can now acquire a mech and level it without playing it...or at least not be required to play it via the "grind"...what kinda data is PGI going to get from what little low tier, less desirable mechs are resultingly played? Really f**king skewed data that's what. What do ya suppose that highly restricted data set is going to do for the establishment of that baseline? Now add to that the nerfs that represent the fourth original goal of the skills tree (the drastic reduction of quirks). Bad mechs made worse are not going to get played...EVER in an un-fully leveled state, and even leveled only rarely and probably only by those players playing them as a mere novelty. Now back to that "data" and "base line" and "balance". Good luck with that.
And now the final implied goal of player satisfaction and retention. So in this system, most of your fairly stable population of players can play whatever they want in a fully leveled state with no need to experiment or try new things. Well that sounds like a plan for a diverse experience just full of viable options... or maybe it is actually the essential definition of stagnation. If you think it is going to be a satisfactory gaming experience to provide a system where only a select few of the best mechs (mechs that are being made better by original goal #4 relative to all others) are played then I would like to sell you some PGI stock, just as soon as they go public.
Edited by Bud Crue, 28 April 2017 - 03:54 AM.