yep, cpu at level 3 and 4 lost to your program! Poor C64

Last thing left now is extending the board structure to store the "blown door" state and the make_move(...) logic.

Your code was written very well and smart !
Thanks again

Well, I think there is only one rule that you didn't get right: It is OK for an attacker to enter a destroyed enemy base (the piece simply goes back to the corresponding base at the other side of the board). I've played enough games against the program that I am fairly confident I have the rules right.

This is my current code:
#include <iostream>
#include <algorithm>
#include <ctime>
// A function to measure time in seconds. I usually implement it wigh
// gettimeofday, but I am posting this one because it's standard.
double now() {
return static_cast<double>(std::clock())/CLOCKS_PER_SEC;
}
typedef unsigned u32;
bool test(u32 x, int bit) {
return (x >> bit) & 1;
}
// Count number of set bits
unsigned int popcount(unsigned int w) {
w -= (w >> 1) & 0x55555555u;
w = (w & 0x33333333u) + ((w >> 2) & 0x33333333u);
w = (w + (w >> 4)) & 0x0f0f0f0fu;
return (w * 0x01010101u) >> 24;
}
int lowest_bit_set(unsigned int w) {
static int const de_Bruijn_table[32] = {
0, 1, 28, 2, 29, 14, 24, 3, 30, 22, 20, 15, 25, 17, 4, 8,
31, 27, 13, 23, 21, 19, 16, 7, 26, 12, 18, 6, 11, 5, 10, 9
};

w &= -(int)w;
return de_Bruijn_table[(w * 0x077cb531) >> 27];
}
const u32 holes = 00022002200u;
struct Board {
u32 net[2];
char base[5][2];

Board() {
for (int i=0; i<2; ++i) {
net = 0;
for (int j=0; j<5; ++j)
base[j] = 3;
}
}
};
std::ostream &operator<<(std::ostream &os, Board const &b) {
for (int i=0; i<5; ++i) {
os << "*0123"[1+b.base[0]] << ' ';
for (int j=0; j<6; ++j) {
int bit = i*6+j;
os << "->< "[test(b.net[0],bit) + 2*test(b.net[1],bit) + 3*test(holes,bit)] << ' ';
}
os << "*0123"[1+b.base[1]] << '\n';
}
return os;
}
typedef char Move; // 1-31 are out-of-base moves, 32 is advance up, 33 is advance and 34 is advance down
int generate_moves(Board const &b, Move *m, int player) {
Move *original_m = m;

// Move figures on the net (put these moves first because they are often the best)
if (b.net[player] != 0) {
*m++ = 32;
*m++ = 33;
*m++ = 34;
}
// Now generate moves bringing out figures from bases
char mask_of_non_empty_bases = 0;
for (unsigned i=0; i<5; ++i)
mask_of_non_empty_bases |= (b.base[player]>0) << i;
char move_order[] = {1,2,4,8,16, 3,5,6,9,10,12,17,18,20,24, 7,11,13,14,19,21,22,25,26,28, 15,23,27,29,30, 31};
for (unsigned i=0; i<31; ++i) {
char c = move_order;
if ((c & mask_of_non_empty_bases) == c)
*m++ = c;
}

return m - original_m;
}
void make_move(Board &b, Move m, int player) {
static u32 const target[2] = {04040404040u, 00101010101u};

u32 &bnp = b.net[player];

if (m<32) {
for (unsigned i=0; i<5; ++i) {
if (test(m, i)) {
b.base[player]--;
bnp |= (077u<<(6*i)) & target[!player];
}
}
}
else {
// Loop over the figures that are boarding the enemy bases
for (u32 boarding = bnp & target[player]; boarding; boarding &= boarding-1) {
int row = lowest_bit_set(boarding) / 6;
b.base[row][!player] = -1;
if (b.base[row][player] >= 0 && b.base[row][player] < 3)
b.base[row][player]++;
}
bnp &= ~target[player];
bnp = player==0 ? bnp << 1 : bnp >> 1;

if (m==32)
bnp = ((bnp & 00000000077u) << 24) + (bnp >> 6);
else if (m==34)
bnp = ((bnp & 00077777777u) << 6) + ((bnp & 07700000000u) >> 24);

bnp &= ~holes;
}

b.net[!player] &= ~bnp;
}
int eval(Board const &b) {
static int const base_value_table[5] = {-200, 0, 500, 1000, 1500};
int on_base_diff = 0;
for (int i=0; i<5; ++i)
on_base_diff += base_value_table[b.base[0]+1] - base_value_table[b.base[1]+1];

int on_net_diff = 400 * (popcount(b.net[0]) - popcount(b.net[1]));

int noise = -32 + std::rand() % 64;

int score = on_base_diff + on_net_diff + noise;

return score;
}
int negamax(Board const &board, int alpha, int beta, int depth, int player) {
static int const depth_reduction[35] = {
0,
1, 1, 2, 1, 2, 2, 2, 1, 2, 2, 2, 2, 2, 2, 2, 1, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
1, 1, 1};
if (depth <= 0)
return player==0 ? eval(board) : -eval(board);

Move moves[34];
int n_moves = generate_moves(board, moves, player);
for (int i=0; i<n_moves; ++i) {
Board copy = board;
make_move(copy, moves, player);
int score = -negamax(copy,
-beta,
-alpha,
depth-depth_reduction[static_cast<int>(moves)],
!player);
if (score > alpha) {
alpha = score;
if (score > beta)
break;
}
}

return alpha;
}
Move search_root(Board const &board, int player) {
double start_time = now();
Move moves[34];
int n_moves = generate_moves(board, moves, player);
if (n_moves == 0)
return 0;
int depth, alpha;
for (depth = 1; depth < 5 || now() < start_time + 2.0; ++depth) {
alpha = -999999;
for (int i=0; i<n_moves; ++i) {
Board copy = board;
make_move(copy, moves, player);
int score = -negamax(copy,
-999999,
-alpha,
depth-1,
!player);
if (score > alpha) {
alpha = score;
std::rotate(moves, moves + i, moves + i + 1);
std::cout << depth << ": (" << 0+moves[0] << ") " << alpha << ' ' << now()-start_time << '\n';
}
}
if (std::abs(alpha)>500000)
break;
std::cout << depth << ". (" << 0+moves[0] << ") " << alpha << ' ' << now()-start_time << '\n';
}
--depth;
std::cout << depth << ". (" << 0+moves[0] << ") " << alpha << ' ' << now()-start_time << '\n';
return moves[0];
}
int main(int argc, char **argv) {
int seed = argc == 2 ? atoi(argv[1]) : std::time(0);
std::cout << "seed = " << seed << '\n';
std::srand(seed);

Board board;
int computer;
std::cout << "Which player am I? (0=pirates (left), 1=sailors (right)) ";
std::cin >> computer;
int player;
std::cout << "Who goes first? (0=pirates (left), 1=sailors (right)) ";
std::cin >> player;

while (1) {
std::cout << board << '\n';
Move move;
if (player == computer)
move = search_root(board, player);
else {
int i;
std::cin >> i;
move = i ? i : search_root(board, player);
}
if (move == 0)
break;
make_move(board, move, player);
player = !player;
}
}


Well, I think there is only one rule that you didn't get right: It is OK for an attacker to enter a destroyed enemy base (the piece simply goes back to the corresponding base at the other side of the board). I've played enough games against the program that I am fairly confident I have the rules right.[/quote]

The rule works like this:
As you already wrote, attacker can enter destroyed base but an attacker cann't come back to destroyed base.
E.g.: attacker is at height 3 beside enemy's base, he can go through it, doesn't matter destroyed or not, but if his home base at height 3 is destroyed he will be also destroyed.

As i see now in your source code above you already implemented this rule correctly. So base[]=-1 means destroyed base (and is drawn with "*").

OK, then. I guess I didn't understand your wording of the rule and I initially implemented as "when boarding an enemy base, the boarding piece is destroyed if either the enemy base or the base where it is normally respawned is destroyed" (since you "enter" both, in some sense).

As you see your original description of the rules was full of holes, but I am happy that eventually I managed to understand the game. It is actually kind of neat!

Yep, when it comes to describing rules i am lost
You did it very well !

My next step is to make the code iterative, the problem:
I will run this on smartphones (iOS, Android) where i cann't use multi-threading. One challange i will be facing is to somehow update the screen while the CPU is "thinking".
My first idea is to call screen-updates from the Negmax-function "periodically" without losing too much CPU-time for the A.I.

What do you mean you can't use multithreading on smartphones?

https://www.google.com/search?q=multithreading+ios
https://www.google.com/search?q=multithreading+android

I have to check this, i am working with AirplaySDK (www.airplaysdk.com) and last time i checked, it wasn't supporting multithreading.

Edit:
Looks like it is supporting now multithreading ... so i will go for it.