“Theoretically perfect game” challenge

[ChioneWinterfury]

I spent way too much time on this but it was actually quite fun:

https://docs.google.com/spreadsheets/d/ ... edit#gid=0

As mentioned in the sheet, my key self-imposed restriction is to build EVERY turn if we can (have any energy). Repeatedly picking to build only lvl 3s is not interesting.

Check my math, see if you can do better

EDIT: the grand total I arrived at was a whopping 148 points. This surprised me, but I guess that's what happens when you assume perfect luck.

Step 8 seems wrong. The tier 3 score 2 VP cards are 5B > trg _bRU_2vp and 5R > trg_bYB_2vp, card id 317 and 318.

there's no 5Y > trg_bBR_2vp in the game. the 3 5Y are discount tier 2, file when building BR & 2 VP when filing.

And you seem to have added additional rules to ignore maximum of 4 tier 3s. There are 7 in your computations

[Fletcheese]

I spent way too much time on this but it was actually quite fun:

https://docs.google.com/spreadsheets/d/ ... edit#gid=0

As mentioned in the sheet, my key self-imposed restriction is to build EVERY turn if we can (have any energy). Repeatedly picking to build only lvl 3s is not interesting.

Check my math, see if you can do better

EDIT: the grand total I arrived at was a whopping 148 points. This surprised me, but I guess that's what happens when you assume perfect luck.

Step 8 seems wrong. The tier 3 score 2 VP cards are 5B > trg _bRU_2vp and 5R > trg_bYB_2vp, card id 317 and 318.

there's no 5Y > trg_bBR_2vp in the game. the 3 5Y are discount tier 2, file when building BR & 2 VP when filing.

And you seem to have added additional rules to ignore maximum of 4 tier 3s. There are 7 in your computations

You are absolutely right on all accounts - I did not consider conveniently ignoring the 4 level 3 limit. Guess it's back to the drawing board...

[ChioneWinterfury]

In the theoratical perfect RNG game, the pick Gizmos have better economy than Build Gizmos, since they return 1 extra gizmo, and you don't need to care about availability and randomness of the pick, since perfect RNG.

The 2 VP omnis is a must for high score, and the last 2 tier 3 cards will definitely include 1X 5 pt VP2. The question for the last tier 3 is then which is better 8 X 2 or 6 X 3. By math 6 X 3 is better hence the last tier 3 card would be the omni capacity instead of the other 5 pt VP2.

Hence my ideal opener is 2 same color picks, going into move 6 to build 2 pt build, then 2 pt pick in move 7 to get back 4 energy to start setting up the 4X 3 pt VPs and the 1X 5 pt VP2.

https://docs.google.com/spreadsheets/d/ ... sp=sharing

I had an initial setup lining up the 4X 3 pt VPs first then going into the VP2 last with 2 doublers for the 3 omni cards at the end. Score 132. But I do also realise we can optimise the sequencing to fit the VP2 earlier so as to set up 4 VP per move as early as possible.

Hence the 2nd worksheet for "Setup 4 VP". My theoratical high score is 144.

[Kodiak-Spirit]

I just got 84 points, it's closer to a perfect game than I've so far. https://boardgamearena.com/table?table=554991122

[Le Penseur]

Theoretically, the maximum number of points mathematically achievable in a game of gizmos, given perfect luck and no opponent to hinder or end the game prematurely, is 347. Of course deducing 3 points per turn would drop this number by a lot: just picking one Energy at a time and then building as soon as you have enough, it takes 64 turns *(-3)=-192 points.
Still, we get 155: an impressive amount, and definitely we could do better giving up some points during the game to reduce the number of turns.

As an example, mindlessly tweaking 2 cards saved 15 turns dropping 0 points, effectively raising the total score to 200. I'm positive that with the correct adjustments somebody could get a lot higher.

Edit: My latest calculations resulted in 332 VPs obtained in 42 turns, for a grand total of 206 VPs.

[Carhou]

I guess I did it.

First, I rebrand the challenge. The perfect game is when you always win. Deduct 3 points for each turn, doesn't make sense... Or you are faster and bigger than your opponent when the game ends, or not.

So, there's a curve with tempo. Doesn't mind if you will get 100 points with 16 Gizmos in 18 Turns, if I get 40 points in turn 12 and the game ends before, when you have just 30 points at that turn, I win, and is just like this.

So, the perfect game when you win, is when you have the bigger score in the shorter time. My proposal is 43 Points in 12 Turns, triggering game end at that point with 4 Level 3 Gizmos. But then, the curve of the combo machine just falls down, and doesn't scale. Is not made for that, is not a space probe, is made to win. I will call it the Rocket. Fast big and explode. Can you beat that? I encourage you to try, I've already thought in one possibility, but now I don't want to test it.

Here's the execution plan::

Code: Select all

TURN	ACTION	CARD      .	COLOR	ENERGY	POINTS	VP	SUM
1	Pick	     .     .	.	U	0	.	0
2	Build	Build Y → Pick 	U	0	1		1
3	Pick	     .     .	.	Y	0	.	1
4	Build	Pick R → Draw	Y	RX	1		2
5	Build	Build R → Pick 	Y	RRX	1		3
6	Build	Build UY →Pick 	R	RXX	2		5
7	Build	Pick RU → Draw	Y	RRXXX	2		7
8	Build	Build BY → 2VP	R	RXX	5		12
9	Build	R R → X X	Y	RRXXX	2	2	16
10	Build	Build BR → File	Y	RXXX	5	2	23
11	Build	4 8 no Research	Y	RXXX	8	2	33
12	Build	4 8 no Research	B	X	8	2	43

Short explanation. When writing energy, X means the wildcard of the Draw trigger effect, means any color. And the legend for colors is: Black - B. Blue - U. Red - R. Yellow - Y. I guess that the name of each Gizmo is obvious, and also its color. The combo triggers, you can follow yourself.

Fun fact. As you see, the points of the curve starts following the Fibonacci sequence, until it breaks in Round 7, due to the limitations of the game, regarding possible combinations of colors, trigger effects in cards, and also the limitation of energy in hand, etc.

Or let's say, and here follows another interesting challenge maybe, which will be the "theoretically perfect DESIGN of the game", to allow the points to follow the sequence? Would be curious to think about it.

By now, I leave here the other challenge.
To beat the ROCKET!!!

[Le Penseur]

"I guess I did it"

I mean... you set up your own rules and declared yourself winner of the game you invented. This format has never even been agreed upon, the thread was about a totally different game.

Don't get me wrong: the rules you suggested are much more interesting than the ones given in the main challenge... but you can't declare closed a game nobody ever played

Btw, let's define a keyword:
- Strike X (of power Y): given a queue of gizmos moves which satisfies a end-of-game trigger, we'll call it Strike X of power Y, where X is the length of the queue and Y the final score of the moves sequence.

Given this term, we can define your problem as follows:
Find the longest Strike N which for any Strike M of power P, given M <= N, has a score at Mth move greater than P.

Am I correct? Did you mean something like this?
Feel free to correct me if I misunderstood something.

Writing down this problem I noticed the theoretical possibility that three strikes may be in a triangular stale.
An hypothetical first strike may end with 20 points at turn 10.
Another may have 22 points at turn 10 and close with 25 points at turn 11.
A third one may have only 18 points at turn 10, but 26 at turn 11.

In this way first strike would lose against the second one, the second against the third one and the latter would lose against the first one.
I fear the problem as you suggested may be flawed and might not have a solution, unless you can demonstrate such strikes' triangles cannot exist within gizmos rules.

[Carhou]

Ciao Le Penseur!

I explain you. The initial challenge was this one:

[...] "the highest scoring run possible, in the shortest amount of time. Build speed would be the biggest constraint. Deduct 3 points for every turn spent. [...]

My modification just was eliminating the rule of deducting 3 points for every turn, because it has no sense to me, is not a rule from the Gizmos game and is not needed at all by my point of view.

Now, as I understand the challenge, it doesn't modify any other rule from the Gizmos game. So, triggering end-of-the-game is the main thing to take in mind, when thinking about "shortest amount of time. Build speed would be the biggest constraint".

So I'm not defining a new complex problem, I'm just playing the game with no luck involved. Also in a previous post, you can see Fletcheese acceptind and confirming this way of see the challenge:

You are absolutely right on all accounts - I did not consider conveniently ignoring the 4 level 3 limit. Guess it's back to the drawing board...

As I saw, your proposal in a previous post is 332 VPs in 42 Turns. Is a good one! But you are forgetting the other important parameter, which is speed, and I guess your proposal loses against mine. What are your VPs at turn 12? Are bigger than mine? I invite you to share your VPs, and also to focus for next proposals in shorter games. Real games from top players with luck involved end between Turn 18 and 20, most of the time.

Regarding the triangular stale you propose as a problem, it doesn't exist according to Gizmos rules, which clearly define the rule for winning, with no tie options. We need just to focus when the game ends, and when we need to check for the score (neither before or after the game ends, just in that turn). Let's say that in your example, game ends at turn 10, so with VPs 20, 22 and 18, clearly the one with 22 points win. There's no need to check what happened at Turn 11, because game ended.

[Le Penseur]

My modification just was eliminating the rule of deducting 3 points for every turn, because it has no sense to me

So I'm not defining a new complex problem, I'm just playing the game with no luck involved.

Yes, you are: the victory condition of the thread challenge was different and you just set up a different challenge and declared yourself winner since all other players suggestions did not win your game, but this was not their main focus.

The goal of the challenge was maximise the points of a gizmos match, not winning one: if you don't like it just create another thread.

Also in a previous post, you can see Fletcheese acceptind and confirming this way of see the challenge:

You are absolutely right on all accounts - I did not consider conveniently ignoring the 4 level 3 limit. Guess it's back to the drawing board...

He did not mean anything like that: he just said he played more than 4 L3 cards in his try, forgetting that would have triggered the endgame condition.

As I saw, your proposal in a previous post is 332 VPs in 42 Turns. Is a good one! But you are forgetting the other important parameter, which is speed, and I guess your proposal loses against mine.

I absolutely considered the speed parameter: in the challenge I was playing it was deducing 3VP per turn.
You are playing a totally different game.

Regarding the triangular stale you propose as a problem, it doesn't exist according to Gizmos rules, which clearly define the rule for winning, with no tie options.

No, you are not getting my point. This is a mathematical issue, not an opinion.

Unless you are able to demonstrate that within gizmos rules cannot exist 3 different Strikes with those properties, your challenge is flawed by definition, since in 3 different matches A would win against B, B would win against C and C would win against A.
This means an unbeatable strike could not exist and your challenge could not have a correct answer.

[FSKFSK]

If your opponent isn't even trying to win the game, you can:

- Buy the two level 3 gizmos that give you 2 points when you build from file.

You're always going to file one turn and then build from file your next turn, to make sure you milk these points. You also can pick one at a time to get the marbles you need, because your opponent isn't trying to win.

- Buy the two level 3 gizmos that give you a point when you file.

- Buy the two level 3 gizoms that give you 2 points when you build.

- Buy the 4 level 2 gizmos that give you a point when you build.

- Buy 3 of the level 1 gizmos that give you a point when you build.

- Buy the 2/1/2 level 2 gizmo to increase your hand size to 7.

- Buy the two level 7 gizmos that give you a point bonus for every point token. Notice this will active all your other point gizmos.

You also need good luck of the marbles available so you always have what you need.