Talk:Connect Four

Source?
What source says that "Victor Allis's thesis is actually a PhD thesis, not a master thesis"? Reading the thesis on http://www.connectfour.net/Files/connect4.pdf, I find that it clearly states on the first page: Department of Mathematics and Computer Science Vrije Universiteit Amsterdam, The Netherlands Masters Thesis, October 1988

BuilderQ 00:33, 22 Mar 2005 (UTC)

Do not point to User:
Tromp has since solved all board sizes where width+height is at most 15.

Tromp: please find a way to make this citation without pointing to User: stuff.

Fplay 01:07, 12 December 2005 (UTC)

"Strategy and tactics" needs work
The image described as "A Connect Four game in progress" shows a game between orange and yellow. The text near it, in the Strategy and tactics section, describes the old image. I think the text is a useful example of a Connect Four strategy, and this new image should be changed to reflect it. The old image was more useful, but the new one looks more polished. I'm reverting to the old image for now, as it matches the text. IChrisI 05:33, 17 March 2006 (UTC)

Talking about the image, it seems to my eyes that the diferent peices are too similar in colour. then again, i do have a slight red/green colour-blindness problem. i still can't tell the colours apart easily though, and i can't be the only one with this problem. somebody please replace the image with a similar board, but different colour peices, like blue&yellow. -Grim- 00:09, 16 March 2007 (UTC)

Virtually every real (and 'genuine', not clone) Connect4 set uses red and yellow. The current svg has the right colours. The bar game image is not a genuine set. M0ffx 23:56, 4 November 2007 (UTC)

Connect 3 ('StraThreeGo')
There is a "3 in a Row" game on a 4x4 board. I have encountered it at http://www.shockwave.com/gamelanding/arcadia.jsp, and don't know if it is worthy of a full article here at WikiPedia... -- Jokes Free4Me 14:58, 24 May 2006 (UTC)

Non-standard Variations
Variations of the standard Connect Four game:

1.) wrap-around left and right sides. If you play this way, you can count a string of four pieces that wraps off one side onto the other.  Logically it is like playing the game on a cylinder.  (I guess you could also try having the top and bottom edges 'wrap around', but I think that'd  not work very well because of gravity. This type of wrapping is akin to two linked cylinders; it's not quite the same as a sphere.)


 * If you link left to right and top to bottom, you're playing on a torus.

2.) inverse Connect Four. You try to force the other player to make a line of four pieces (2-way inverse).  This version can also be played with a person who has limited cognitive abilities and/or simply likes to randomly put the pieces in the piece holder.  (1-way inverse plus random filling)

3.) checkerboard Connect Four. You play on a horizontal surface (maybe larger or smaller) if you don't have the usual vertical game set.

4.) connect 5. You aim to get 5 pieces in a row. (harder to attain on small boards and more likely to be blocked).

I didn't put these in the main article because I don't know if anyone else has played Connect Four these ways, and I don't know if minor variations on games should be in primary articles. (--Kaze0010 03:48, 15 June 2006 (UTC)


 * Those are a nice addition to the article, in my opinion. --Anand 14:30, 30 August 2006 (UTC)


 * IMHO they should be mentioned in the article, in an subsection. But drop point 2, it's senseless. --ThG 15:12, 30 August 2006 (UTC)


 * I Agree that there should be a variations section to this article (I've played 3 & 4 from above and 1 sounds interesting). There is precident, try looking at card games such as Shithead or even established games such as poker, there are many variations mentioned without reference. To that end I found a very good extension to the standard game is double move connect4 (A common pratice in spicing up 2 player games when bored of the standard rules). The first player plays 1 piece, from then on all moves consist of 2 pieces (played as independant moves). Playing like this the games are much quicker and involve many more traps, 2 pieces on top of each other are a threat! --Martin lester 14:57, 10 November 2006 (UTC)


 * I have updated the main document to include a variations section. I don't think it breaks any of the wiki rules --Martin lester 16:24, 13 November 2006 (UTC)

Nerd Players Section
This section seriously needs to be revised or deleted; I would have removed it myself largely due to the somewhat questionable inclusion of an e-mail address. E Liquere 02:44, 20 October 2006 (UTC)


 * Obviously needed deleted. It's gone now. -- Schaefer (Talk) 07:56, 20 October 2006 (UTC)

4x4x4
I've played this on a 4x4x4 grid, with downwards gravity. Is that a separate game or a variation? --Apoc2400 05:17, 1 May 2007 (UTC)
 * Found it: Score Four. Added to Variations section. --Apoc2400 05:31, 1 May 2007 (UTC)

Tell me if the theory and strategy sections are clear.
I wrote the Theory and Strategy sections, they're clear to me but since they lack pictures, they might be difficult for readers to understand. Should I add pictures or is it clear enough?

Oh and i forgot an edit summary: I put those sections because I figured people who want to learn about Connect Four should be able to learn it from Wikipedia since one can learn just about every other topic. Other topics are very informative and now this one is too.

I changed the link titles, i doubt anyone will object.

I filled in the blank lines of the picture caption, and I changed "10 minutes" to "2-10" minutes because that's more realistic.Div0ckrehnee 06:07, 12 September 2007 (UTC)


 * I've removed these sections. While the content is well-written and in-depth, it's simply not appropriate for Wikipedia. The purpose of Wikipedia is to present facts, not to teach subject matter. The Wikibooks project would be a perfect place for this kind of content. Chris Cunningham 10:25, 12 September 2007 (UTC)
 * Why did you remove them when the Chess and Poker strategy pages are not removed? —Preceding unsigned comment added by 71.175.31.106 (talk) 02:57, 2 March 2008 (UTC)

The Giant Connect 4
The Giant Connect 4 is a modification that lends itself to a social party game; simular to how Chess is often setup in a giant game setting. The giant connect 4 is fun to play in an indoor or outdoor party setting. Some pictures of it being played are at  or Google "Giant Connect 4" and look for ebay link. —Preceding unsigned comment added by UBERGAMES (talk • contribs) 15:09, 7 December 2007 (UTC)

Reverted vandalism - I hope!
Removed the following: "also it has been solved by international strong man, Andrew Nevarez on hundreds of occasions" - this makes no sense (a board game can only be solved analytically once), and is not backed up by the supplied reference. If I've missed something obvious, please correct me, but this seems like obvious vandalism (dates from 18th April 2008). Cooperised (talk) 23:33, 26 April 2008 (UTC)

Is Connect Four a board game?
Connect Four is listed as a board game on the board games page, but on the Connect Four page is says it is not a board game. I think that needs cleared up. —Preceding unsigned comment added by 12.192.16.230 (talk) 01:25, 26 August 2008 (UTC)
 * I would guess it's a matter of opinion coming down to whether the vertical frame used to hold the counters counts as a "board" or not. 94.6.24.111 (talk) 13:56, 28 September 2013 (UTC)

David Bowie?
I've seen several sources claim (without providing much elucidation) that David Bowie either owns or once owned a stake in the copyright on Connect 4. It has the classic ring of an urban myth to me, but I'd be interested if anyone can confirm or deny. Ravenclaw (talk) 11:12, 23 March 2009 (UTC)


 * This seems to have been put out by the people at TV Cream. What we need now is for a music magazine to interview him and ask him whether he owns a stake in Milton Bradley, at which point we could mention the urban legend and debunk it, with a source. -Ashley Pomeroy (talk) 17:12, 26 April 2009 (UTC)


 * The TV Cream page dates back to before March 16, 2007 (when it was first archived).
 * The claim was given further exposure by Being Human—where a character said something like "David Bowie owns Connect Four, except maybe it's Iggy Pop and Kerplunk".
 * Since May 27, 2009 the Wikipedia article on Grenfell Glen has said "David Bowie invented the game Connect Four when he lived on Pratt Street."
 * Despite the above, I cannot find any good sources to support the claim. —MJBurrage(T•C) 18:57, 6 August 2009 (UTC)

The Captain's Mistress
The article states that the game was traditionally known as The Captain's Mistress. The name seems bizarre; a couple of pages on the internet state that the name comes from Captain Cook, who used to play the game a lot whilst on long sea voyages, rather than e.g. making love to a woman, or gambling, but I can't find a really definitive source. -Ashley Pomeroy (talk) 17:12, 26 April 2009 (UTC)


 * It should not be included, it's connected to the "Harrison Heath" Wikipedia trolling (http://www.facebook.com/group.php?gid=70394073287) and it used the same referance as the "Connect Four" trademark page while there was nothing regarding "The Captain's Mistress" on that page.78.82.140.213 (talk) 14:56, 3 November 2009 (UTC)


 * Here's your vandal edit - it has no connection to the Captain's Mistress reference. I've heard the Captain Cook thing mentioned before, it's referenced a lot online and a Google book search turns up a mention in this 1994 novel, but I can't find a solid, reliable source for it. If there are several commercial versions out there under the name, though, it may be worth mentioning. --McGeddon (talk) 15:15, 3 November 2009 (UTC)


 * I have no information on whether the reference to the game as The Captain's Mistress is correct or not, but it was added to the article way back in May 2004: . It certainly appears to be unrelated to the Harrison Heath vandalism. TJRC (talk) 21:58, 3 November 2009 (UTC)


 * I've discussed this with a fellow professor at the Tom Haynes University who has just completed his sixth thesis on the playing and history of Connect 4 - he stated that whilst there are doubts about the exact link between Harrison Heath and the game, there is irrefutable proof that he had something to do with the history of the game. Whether he did indeed invent it, or was merely an early 'champion' (it used to be used as a substitute for pistol duelling, with the loser being killed by hanging), he definitely warrants being mentioned in the article. Once the professor's work is published, I will edit the article with appropriate references. Prof. Blams talk 18:58, 22 October 2011 (UTC) — Preceding unsigned comment added by 94.173.195.93 (talk)


 * For context, after trying to fake a message from a different editor the above IP also removed a link in this thread to a Harrison Heath Facebook group (a group whose mission is defined as "Basically, adding "Harrison Heath, the famous pioneer of disc based puzzles" to random Wikipedia pages, normally those of board games, but we're looking to branch out. Remember not to do too many, and keep them believable, because if the moderators find one they're probably going to find them all."). One crept back in at Mastermind, but I think we've found them all now. Thanks, professor. --McGeddon (talk) 14:06, 6 February 2012 (UTC)


 * I've taken this own of the article because it isn't properly sourced. It seems an unlikely name for a game first invented in the 1970s.--Jack Upland (talk) 02:09, 29 December 2019 (UTC)


 * The game is known as Captain's Mistress, even if the origin story is false, I have readded this to the lead.-- — Preceding unsigned comment added by 97.124.121.34 (talk) 03:24, 18 February 2021 (UTC)

David Bowie (again)
Have found reference for the David Bowie myth and added a section to the article. Paul Brookes (talk) 13:36, 20 November 2009 (UTC)

Dubious
"In recent years, Milton-Bradley added a new rule to the Connect 4 rulebook, entitled the Square Rule. This rule was originally created for experienced players whose games too often ended in ties. The Square Rule occurs when the player makes a 2x2 square out of his/her pieces." - can't find a source for this. Is there any truth in it, or (as is suggested by the now-deleted "Although unpopular in Australia, the rule has gained increasing fame across the globe.") is this just someone in Australia trying to prove they've won a game by pointing their opponent at Wikipedia? --McGeddon (talk) 14:00, 30 July 2010 (UTC)


 * I've undone all four of User:Jacksmith1978's lifetime edits pending a citation. Cheers, CliffC (talk) 16:29, 30 July 2010 (UTC)

Connect 5
I remember a game which iirc was called "connect five" which worked on the same basic principle of dropping counters into a board but the counters were much smaller, you had to get five in a row to win and there were "tricky move" bars that could move counters across the board in certain rows within the board. However my searches just seem to turn up unrelated games.

Anyone remmber this? Plugwash (talk) 00:55, 1 February 2013 (UTC)
 * Haven't heard of it. You'd probably be better off asking at the Wikipedia reference desk. --McGeddon (talk) 09:46, 1 February 2013 (UTC)

Connect Four board size variations?
In the article: There are many variations on the board size, the most commonly used being 7×6, followed by 8×7, 9×7 and 10×7.

Does anyone have citations for the 8×7, 9×7, and 10×7 board size variations? What about 8×8?

--GamePlayerAI (talk) 00:25, 21 December 2013 (UTC)

"Greatest Player"?
I think something is up with that lead section&hellip; The infobox, too. And also the Popular culture section. Seems to me like someone self-inserted. Can anyone double-check the page history to be sure? 50.135.219.155 (talk)

Algorithms section

 * "As opposed to calculating perfect play, computer algorithms that play Connect Four must be able to run in real time and find good moves in just a few seconds."

This seems to be outdated, see e.g. this website - it shows the perfect move and predicted end (for perfect play) in fractions of a second. --mfb (talk) 12:37, 11 July 2016 (UTC)

Connect Four perfect playing game solvers have been available since 2012 (Berkeley's version). Several others are cited or listed in the article. Some do play in real time, and either run or lookup alpha-beta pruning results to the end game (up to depth 42 plies). The section makes multiple assertions about Connect Four that are conjecture, and are clearly inaccurate, given perfect playing programs. WP:Blogs are unreliable sources and one has been withdrawn by its author, at the time of this writing. The content which appears encyclopedic discusses the topic Monte Carlo tree search. However, the topic by itself is of limited interest for Connect Four. I don't believe this section is fixable, so I'll suggest deleting the section. GamePlayerAI (talk) 22:45, 23 July 2016 (UTC)

The website that you showed does indeed give perfect play, but it doesn't calculate a majority of it on the spot. It uses a huge opening database for its calculations, shown by the fact that it takes more time to calculated moves later in the game than the first 10 or so moves. This is highly dubious because alpha-beta pruning at a depth of 42 plies should take significantly longer than at a depth of around 20, so seeing it take 0 seconds for the first move and over a second for later moves is suspicious. I personally have yet to find any reliable source that can find perfect moves in less than a few hours, but it's possible I'm missing something and I'd love to hear about it! Ofek Gila (talk) 23:07, 8 November 2016 (UTC)
 * The game solvers calculate and/or look-up alpha-beta results from exhaustive searching to the end game. They are reliable sources that do find, report, or play connect four perfect moves in less than a few seconds. An un-optimized alpha-beta may require several hours to to find perfect moves. However, the solvers use highly optimized Negamax with an opening database (static transposition table) and transposition tables. The opening database, pre-computed offline with alpha-beta, need not be huge. "John's Connect Four Playground" solver uses his 8-ply "Connect-4 Data Set" with 67557 entries. And the asun.net "Four in a Row Solver" with a 10-ply database manages to fit within a cellular phone. GamePlayerAI (talk) 08:16, 9 November 2016 (UTC)
 * I completely understand, and like I said, all of the existing solutions use opening databases. The (by far) most difficult and important moves are the first few, as messing up in them could leave you in an un-fixable situation. Again, I have yet to find an AI, as optimized as you want, which does all of its calculations in real time on the spot, that could even come close to existing monte carlo solutions. Let met know if you don't fully understand what I mean. Ofek Gila (talk) 18:20, 9 November 2016 (UTC)
 * Well, it simply isn't necessary with alpha-beta AI to perform all calculations in real time (when some calculations can be looked up) in order to achieve real time perfect play in the game of connect four. And perfect play is more superior to what can be achieved with Monte Carlo connect four. GamePlayerAI (talk) 21:39, 9 November 2016 (UTC)
 * Again, I understand that positional databases or tables are commonly used in Minimax implementations, but they are no-less commonly used in Monte Carlo implementations. Many Monte Carlo implementations also have opening books and closing books, instead of running until the end of the game, they run until a specific depth and then check the tables. Again, just because the AI can be completely valid using tables/databases, it still uses them, meaning that it does not do all its thinking on the spot. What I was saying was that minimax-based algorithms for connect four cannot run until the end of the game in real time, and therefore often end up with losing endgames (and this is evident in virtually every non-database-using minimax connect four ai online and for mobile devices, try it out!). In contrast, pure monte carlo solutions (without any tables) have a much higher tendency to end up with winning endgames because of the way its random simulations run. Now I'll admit, no algorithms section is complete without mentioning minimax-based solutions with tables (like the negamax alpha-beta pruning method you mentioned), but I thought that explaining the importance of looking-ahead, and mentioning how monte carlo algorithm solutions can naturally do this, would add to this Wikipedia article. Hope that made sense. Ofek Gila (talk) 23:48, 9 November 2016 (UTC)
 * Parts of the information you were interested in adding appears to be a better fit for the Monte Carlo tree search subject, and you could consider adding to that article too. This article is about the game of Connect Four, rather than programming methods and AI techniques for Connect Four computer players. Nevertheless, the earlier Algorithms section was WP:DISPUTED, because the article's cited Connect Four solvers are counter references which refute much of what was said in the section. The section also had WP:RELIABLE and WP:WEASEL issues too. I certainly understand what you're saying about "thinking on the spot", but as I eluded to before, that detail doesn't matter for this article's purpose. A connect four solver always returns perfect moves (identical result as a minimax run to the end of the game) regardless of whether the solver had to "think" about it, or not. And a game solver, by definition, will never end up with losing endgames, when beginning with any winning game position. GamePlayerAI (talk) 18:10, 10 November 2016 (UTC)
 * Fair enough Ofek Gila (talk) 22:50, 10 November 2016 (UTC)

4x4 Game and Classic Grid
I was wondering why Connect 4x4 was not added to different variations but instead has its own wiki article (See Here). Should this other article be merged into this one? If not then should this version be mentioned in this article? The instructions from Hasbro's website can be seen here.

I also noticed that the Connect 4 Classic Grid is not listed here. It contains special discs. The instructions from Hasbro's website can be seen here — Preceding unsigned comment added by 2.137.75.186 (talk) 23:46, 1 January 2017 (UTC)


 * "One board variation available as a physical game is Hasbro's Connect 4x4."
 * Appears in the "Rule variations" section. Perhaps this should be moved to the "Other versions" section?
 * Connect 4 Classic Grid is one of several (discontinued) Hasbro's Connect Four physical game board releases that support pop-out. Maybe a new list of Hasbro Connect Four version history can include Connect 4 Classic Grid by name? For now, the article just notes:
 * "Several versions of Hasbro's Connect Four physical gameboard make it easy to remove game pieces from the bottom one at a time."
 * where Connect 4 Classic Grid would be one example. The "Connect 4 Instructions" in the references section notes another game version with pop-out and with special discs. GamePlayerAI (talk) 18:34, 7 January 2017 (UTC)

Meme
Connect four's box has become a shitpost meme. We should add something to the In popular culture section. DatGuyonYouTube (talk) 15:30, 7 March 2018 (UTC)

Unexpected piece counts - 19 red 23 yellow - and the meaning of life (42)
This set was from a mental ward. At the time I thought that the meaning of life, etc, (Hitchhikers Guide to the Galaxy) was 42, and the question was "what's 6x7?". For some reason I thought I'd count the pieces and the result was unexpected. I wonder how often sets are like this - 1 in 1000? And it was the first time I'd ever tried counting the pieces. Zephyr103 (talk) 03:46, 19 April 2019 (UTC)

Rules of Connect Four
I am aware of the published rules of connect 4 by Hasbro. But, there is an interesting vagueness in the rules for which I am unable to find any clarification in this article or elsewhere on the web. The rules say that the first person to get four of their own chips in a row (vertical/horizontal/diagonal) is the winner, but, what happens if the person happens to not get four in a row, but instead they get five, six, or seven in a row? The rules "imply" that one, two, or three in a row is not a win, and we do know the game is called "connect four". Basically, 1, 2, 3, 5, 6, or 7 is not 4. So, is it possible that unless I happen to get "exactly" four of my own color chips in a row, am I not a winner? Or is it an unwritten rule that getting something that is not four in a row actually a win, like five, six, or seven in a row? (I think I made my point clear, and I hope this is a relevant vagueness that could be pointed out - if not clarified - in the article.) A793b4 (talk) 02:58, 10 June 2021 (UTC)