You are viewing a single comment's thread from:
RE: The Incoherent Ravings of a Mad Man
I appreciate the passion; but I absolutely disagree.
This doesn't give bots an advantage; it massively increases the number of variables.
We humans have a two stage decision making process, so we don't have to handle the mental load of choosing everything at once; at the same time, bots are coded to check the ruleset and mana cap, and add in the team which has previously performed the best under those parameters.
I actually suggested a different, two step process a while back; but as usual, the devs have gone one step better.
Why do you think a bot would not be unable to understand a 2 part process? The 2nd part being much simpler than the 1st part? This might trip up the bots that only play a set hand on a rule, but what about bots that can run 100,000 simulations on part 1? With the limited inputs the simulations will get into the insane numbers, like high millions. The bot will always be able to make the best move.
A bot also can see what spells and items you have access to, and can see your history of play. Something humans can not do. This is a huge advantage when running simulations over millions of times.
"This doesn't give bots an advantage; it massively increases the number of variables." -- Simply increasing the number of variables to combat bots totally lacks understand of the full capability of bots. This will actually improve the higher level bots. Yes, it will retard the pick-from-a-list bots, but increase the advantage of the higher-level bots using machine learning + simulation testing. You can literally run simulations in the millions in just a mere 30 seconds scaling out in cloud functions to factually find the highest chance of winning. Proceed that by machine learning selecting which of the potential best 100,000+ item/spell combos to test... not even bji can come close to that. Also, the spell system as described actually provides a lot more data for bots than before, not less, due to surveying the selected teams prior to the second step of the application of items or spells. Pre-spells, you have 0 information past a user's collection and history. Post-spells, you are given 50% of the final information to work with prior to the final submission. That is a massive boon for bots. If you really want to beat bots, just purely make the game nothing but random luck with no maximizing the odds, lol. Just don't be fooled into thinking things will get better for humans with this change, they will get worse. FYI, there are already AI/Machine Learning frameworks for multi-process/step games. Pushing a total blind one step surprise to a more information-rich process is actually far better for the competent bots. A slight aside, if you do manage to make the game so complex to the point it actually starts to slow down a good bot, you will have made the game much less interesting for the average player. You will push the game more and more niche pushing out more and more potential players shrinking the community. It would be better to invest resources into more alternate game modes to actually attract new players and keep existing ones interested. Funny thing about doing that is that the more game modes there are, the more hindered bots would be as they then have multiple models to deal and split up their time and resources on. The only thing this newly suggested system does is improve higher-level bots and compound the pay-to-win shittyness of the game. People will now have paid for better cards AND for better items. (High level bots with best p2w items, lol, joke game) Good luck, have fun new players that are sure to stick around...