Thoughts on connect 4. 4ir forum
8 replies. Last post: 2010-02-03Reply to this topic Return to forum
wccanard at 2009-02-06
I’ve been playing 4ir for about a month now. Initially I had thought that it would be all about manipulating yourself into a position where you would win the endgame, and that one would have to learn the ins and outs of complex endgames (e.g. “yellow has a threat on the 3rd row, and then, in another column, a threat on the 4th row and a minor threat on the 7th row, but red has a threat on the 8th row too, and on the 5th row in another column, so who wins?”). But, at least on an 8x8 board, it doesn’t seem to be like that at all, at least at the top end of the spectrum.
Here’s how I currently think the game works though, based on looking through games between experts and slowly beginning to understand their moves.
1) Opening theory. There is a vast amount of opening theory on an 8x8 board and the majority of it is “secret”. All the good players know all the natural openings, many of which run 16 or more moves deep (8 moves each). As far as I know there are no places on the web to read about standard openings---you just have to figure them out yourself. It’s very very easy to lose the game, as player 1 or as player 2, on your 7th or 8th move, by deviating slightly from the standard opening that you’re in. Anyone who looks at a few championship games deduces quickly that “a common opening is 4,4,4,4,4,4” (or d1 d2 d3 d4 d5 d6 in chess notation) but if you go down that route you’d better know the next 5 or 6 moves to play or you’re in trouble against an expert. If a standard opening is played then there are a couple of standard “key points” that you need to know at the end of it, and if P2 keeps their head and knows the key points, they will almost certainly win. Oh---this brings me to
2) On an 8x8 board the game is almost certainly a win for player 2. Take a look at the records of any of the top players; they win almost all of their games, and tie a few, but the vast majority of the games they lose are games they lose as player 1. All the top players seem to be agreed on this point.
3) As player 2, the most common way to win seems to be the simplest way. Play a standard opening and then (a) make an even threat and, simultaneously, (b) make as many odd threats as your opponent. If you play a standard opening out for the first 16 moves then there are almost no places on the board where odd threats can form! So you don’t need to understand complex openings, you just (as P2) have to make that even threat, which is typically quite easy.
4) As player 1, the most common way for an expert to win against another expert appears to be to try and find an opening novelty that your opponent doesn’t play very well. It seems to me that basically all games are won and lost in the first 16-20 moves of the game, and if he’s lucky, player 2 might not even have to think. Hence there are some games going on on this server where P2 (a) plays perfectly and (b) doesn’t ever think for more than a few seconds per move! He just plays out the standard opening which leads to the standard win. P1 will win by tricking player 2, either by deviating from the standard line, or by knowing a standard line for a few moves further than P1. This is very different to a game I understand much better---dots and boxes, where opening theory is almost non-existent and against an expert I will almost always have to spend 15 or more minutes on a move at some point (and sometimes an hour).
5) Very very vaguely speaking, it seems to me that sub-2000 players are the ones that lose as P2 because they don’t know the standard opening lines for long enough, and 2000+ players are the ones that lose as P2 because they know all the standard lines but don’t know the strange variants that P1 tries to pull off.
David Scott at 2009-10-22
I pretty much agree with the gist of what you are saying. I haven’t played for a while, but when I did, it was essentially like you say. There is definitely a large body of opening theory, and you will find that many of the main lines actually transpose to the same exact positions later on in the game. I used to spider the site every few months to download the most recent games of high rated players, and put them into an openings database to help me understand the recent trends. To be honest, I can’t really understand why the ‘main lines’ are played so often – they are a little dangerous for red, but are so well trodden that it is tough for red to go wrong. I won practically every red game that I played – only the very toughest players would have even the remotest chance of winning or drawing with yellow. Of course, when I played the other tough players at the time, it was similarly difficult to win or draw with yellow. I would look for games played by my opponents where they made mistakes in the past (especially if they went unpunished), and aim for those positions, although this would rarely work. Often the outcome of monthly cups would depend on who got more reds in the final round against the tougher opponents. It is possible to win with yellow, even against strong players, but you can only really improve the low chance that they will screw up. If your opponent sticks to the main lines a lot, try to deviate early, and avoid transpositions to the main positions. If your opponent plays his/her own openings, stick to the main lines. Deviating from opening theory doesn’t necessarily mean automatic loss, but you have to be a very good player to check that your opening novelty doesn’t lead to a forced loss, and you probably don’t have time to invent novelties for every game. With red, stick with the tried and true main lines.
Rex Moore ★ at 2010-02-02
What is meant by “even” and “odd” threats?
KPT ★ at 2010-02-02
pair an unpair
Rex Moore ★ at 2010-02-02
Thanks, but I don’t understand what that means. :)
Is it possible to illustrate an even or odd threat here?
x <--Odd threat with x on the two rows?
x o o <--Even threat with o o?
Rex Moore ★ at 2010-02-02
Let me try that again:
....x <--Odd threat with x on the two rows?
..x o o <--Even threat with o o?
halladba at 2010-02-02
As far as I know the following is an odd threat:
If x is able to play on the third (odd) row it wins.
The next diagram is an even threat for x (on the second row):
FatPhil at 2010-02-03
I think you can use <tt> for monospace<tt>