Patchwork Medium AI (Ute) on Android

[CameronHamill]

Hi Fellow Members,

I'm wondering if many of you also play Patchwork on the Android App? (I'm not sure if it's also on the Apple App Store).

I mostly play against the Medium AI Bot (named 'Ute') on the official Patchwork app. And around 20 rated games.

I'm just wondering how you would rate the Medium AI bot's ability compared with ranked human players on Board Game Arena? I realize it's subjective. I'm not sure on the best way to become quite good Medium AI level Patchwork. I would classify good as a 50% Win Rate. Very good as let's say a 75% win rate against medium AI and excellent with a 90% win rate against medium AI. At the 75% win rate possibly moving it up a level and playing against hard AI.

I have a poor win rate of only 30% against medium AI after 320 or so matches. I've been playing Patchwork for a few months. I'm not sure on the best strategy to try and get my win rate increasing. I seem to have hit a brick wall!

Essentially I'm asking what Patchwork ELO from Boardgame Arena would you estimate Medium AI on the official Patchwork App is?

[adeodatus243]

1. Value pieces numerically based on their benefits (space filling, button income) against their costs (buttons & time).

2. These values change as the game progresses. In the early game (about through the first four button incomes) button income is much more important because your button supply is limited, so you need that income to buy future patches. You also have more button incomes to benefit from that income, and filling spaces won't really help for a very long time (the end of the game). Try using the formula (button_cost+time_cost)/button_income to compare piece values. You'll probably end up passing several times or being constrained on what pieces you can buy by lack of buttons, so valuing one time as one button is a fair approximation.

In the middle game (about through the next three button incomes) time is your main constraint. This is when you fill the bulk of your board. Try using the formula (2×squares_covered+button_income×button_incomes_left–button_cost)/time_cost to compare piece values.

In the late game (about the last two button income sections, the last 12 time track spaces, though this can vary) space often becomes the main constraint. You're also close enough to the end that it can be possible to map out all possible endgames and decide what to do. My valuation formula is 2×squares_covered+button_income×button_incomes_left–button_cost–time_cost.

3. This is a 2-player zero-sum game. Anything that's good for you is bad for your opponent and vice versa. Don't be afraid to pass instead of taking an okay piece that gives your opponent a good piece. If you're winning (meaning if both players passed until the end and the last player took the best of the three pieces available you'd win) then don't lose—don't make a move that gives your opponent a better move.

[CameronHamill]

Thank you very much for taking the time to write-up such details! I'll re-read all of your points and learn your formulas! I've played over 320 games against Medium AI and I thought I had it all figured out! Lol. I will try all of your tips!

Thanks very much.

I will like to establish my ELO on BGA so I can see what my actual playing level against humans is. I'm only 520 RP within the app though I've only played around 20 games against humans. I've never played the physical version.

[Turtledude3333]

I was not familiar with the Patchwork app until your post. I downloaded it and played a couple of games against the AI. I would guess that the medium AI on the Patchwork app corresponds to about 100-150 Elo on BGA. Some observations against the medium AI opponent:

Weaknesses
1) Does not appropriately plan ahead.
2) Will take low value pieces unnecessarily.
3) Does not consider the board state it leaves for the opponent.
4) Inconsistently identifies when to correctly pass the turn and may pass unnecessarily.

Strengths
1) Does value ensuring the board has open space to play pieces.
2) Understands that passing may be the best move.

If you are looking to improve, I would recommend playing more games on BGA. The BGA interface allows you to more easily view all of the pieces so that you can better plan ahead.

I would also recommend viewing game replays from some of the top-rated players. Try to identify what their next move will be before clicking "next move." If they made a different move than you would've made, stop to think about why.

[CameronHamill]

Thanks very much for your analysis Turtledude3333!

I appreciate very much your estimate of Medium AI's ELO. I agree with you about not being able to see the tile selections all at once. Like on BGA or a physical game.

My win rate against medium AI is only at 34% so I think I will have to boost my plays against people on BGA. Again thank you for your analysis and also for your suggestions.