Human versus Bot – The machine revolution in Splinterlands
There is always a lot of discussion about the issue of bots in Splinterlands.
On the other hand, in many ways, the profusion of bots can damage the game's economy and several measures have already been taken by the developers to mitigate this problem.
But it is not about all this controversy that I want to talk about today, but about the combat between bots and humans in practice, in battles.
For the time being, to our delight, the Splinterlands bots don't play very well and we humans still have a facility to beat them. But not for a long time?
One thing that is very clear to me is that in all types of games, the tendency is for engines, the computer software, to become superior to the best humans.
I believe this is not a possibility, but a certainty, we just do not yet know the exact time for bots to fully dominate humans in a specific game.
In 1997 a supercomputer, which occupied the space of a room, called Deep Blue created by IBM, defeat the world chess champion, Garry Kasparov, losing a few games.
The advantages of the machine are many, in addition to the infinitely superior calculation capacity, it does not get tiring, it can play hours and hours without any change in its performance, it does not go through the ups and downs inherent to a human, such as fatigue problems, and state emotional and psychological.
Also Go, the Chinese chess, a game in which humans until recently had a great advantage against machines for requiring a non-linear way of thinking, is currently also being dominated by engines.
At the moment the baton is with Google DeepMind the artificial intelligence department of Google, which created the AlphaGo software.
In 2016, AlphaGo challenged Korean Lee Sedol, best GO player, world champion for 10 consecutive years, considered a true legend.
Lee started by losing the first three games, and in the third, the machine was able to do a movement at first sight simple but so creative that it eliminated Lee's chances of victory.
Lee starts the 4th match determined to regain his pride and that of his millions of fans. However, everything pointed to another defeat.
After 30 minutes of reflection, Lee made a movement so unusual, beautiful, amazing and effective, that it was later called "Touch of God".
After Lee's move 78, AlphaGo responds inconsistently, and, for the first time, Lee manages to emerge victorious against the machine.
His move also has a 1 in 10,000 chance of being played by a human, but this would also be his last victory, after which the frustrated champion decided to retire.
To learn more about this confrontation, I recommend watching the AlphaGo movie on Netflix.
Since then, the developers of Google DeepMind have created a new version of the software, AlphaGo Zero, which only needed 70 hours of training playing against itself to play at a superhuman level, and start making plays "never before seen".
For now, its only limitation is its requirement for fixed rules.
The software is not able to do well in situations where the input data is random and changeable.
That is, their superhuman abilities are restricted to games that have a set of fixed rules and that do not change.
Now, so going back to our beloved Splinterlands, what will happen when the software of this caliber starts playing it?
Will we only have machines in the TOP 100?
What will be the solutions? ban bots? A separate league just for bots?
Please leave your opinion to enrich this fascinating discussion...
See you soon!
Images: tecmundo thenextweb & @splinterlands