Posts:
4
Registered:
Aug 4, 2026
Age:
67 Home page
Looking for beta testers — a Pente app for iPhone, iPad, and Mac
Posted:
Aug 8, 2026, 4:30 PM
Hello all — first post. I checked with rainwolf before writing it.
I've spent the last stretch building a Pente app for iPhone, iPad, and Mac. It's called Lapilli Stones, because Pente® is Hasbro's trademark and I'm a one-man operation with no appetite for a letter from their lawyers. The game underneath is the real thing: 19×19, pair captures, ten stones to win by capture, plus Keryo and the tournament rule. Gabrel, Tesh, and Braunlich are credited by name on the site — it seemed wrong to build this and not say where it came from.
It's in TestFlight now and I'm looking for testers who will be hard on it — specifically on the computer opponent. Here's my problem: as much as I love this game, I'm not good enough to give Advanced a real run for its money. It plays a genuine search, and I can't tell you how strong it actually is, because I'm not strong enough to find out. That's what I need people here for. "Your Advanced level hangs a capture in this position" is exactly the kind of report I'm after.
There's also two-player on one device, and correspondence play over Game Center if you'd like a game against another tester.
To be plain about it: this isn't a rival to pente.org. No server, no accounts, no ratings, no game database — nothing that competes with what's here. It's a board in your pocket for when you're away from one.
If you're willing: https://lapilli-stones.com/testers.html has the setup steps, or just email lapilli-stones@proton.me and say which devices you play on. Free during testing, and I'll make sure testers end up with a copy of the released version.
There's also a page for what's coming next — recorded games in a notation you can export, a review mode that walks back through a finished game, a practice board. https://lapilli-stones.com/roadmap.html No obligation to care about any of it, but most of what's on that page started as an offhand remark from someone playing the game, so if you look at it and think "that's the wrong thing to build next," I'd want to hear that too.
Posts:
9
Registered:
Oct 29, 2021
Age:
72 Home page
Re: Looking for beta testers — a Pente app for iPhone, iPad, and Mac
Posted:
Aug 12, 2026, 5:19 AM
I'm in almost exactly the same place - I'm implementing Gomoku and some related games for Boardspace.net, and I'm using Gomoku+capture2 as a variation name.
Posts:
2,293
Registered:
Mar 4, 2007
From:
San Francisco
Age:
45 Home page
Re: Looking for beta testers — a Pente app for iPhone, iPad, and Mac
Posted:
Aug 12, 2026, 11:04 AM
hi im a pente master. i would be interested to test but i use android and pc. we also have a app which has AI level 12, and does VCT mode as well. victory by continuous threats, meaning if it hits 12 moves, if there is a forcing move, it will keep searching beyond 12 as long as each move is a threat. we could face them off with each other. the AI has trouble spotting some stuff. example, if only way to win is with a diagonal post 3. meaning X _ X _ X. and meaning to form one. the AI wont see it and will take a losing path. the app also allows to inject opening book files. and to do analysis and create data bases of moves with value markers placed on moves if desired. also show threats highlighted if wanted.
Scire hostis animum - Intelligere ludum - Nosce te ipsum - Prima moventur conciliat - Nolite errare
Posts:
4
Registered:
Aug 4, 2026
Age:
67 Home page
Re: Looking for beta testers — a Pente app for iPhone, iPad, and Mac
Posted:
Aug 12, 2026, 1:47 PM
Thanks for writing — and thanks for reading the post.
First, a clarification before anything else: I'm not a Pente master and don't want to be mistaken for one. I'm a hobbyist who fell for this game in a disco in 1979, bought Braunlich's book, played it with my wife and kids for years, and finally, after retiring from a software career, built the version I'd been wanting. I'm the developer, not a strong player. You'd almost certainly beat me badly, and my engine too.
On that: your engine sounds meaningfully stronger than mine, and I'd rather say so plainly than posture. Mine is iterative-deepening alpha-beta with a transposition table, capped at 6 plies with about a 1.5-second budget — tuned to be a good opponent on a phone, not to be a strong Pente engine. No VCT. I've measured what more time would buy, and it's less than a ply per extra second; getting to your depth isn't a matter of spending longer, it's threat-space search, which I haven't built. So depth 12 with continuous-threat extension is a different class of thing, and I know it.
Two things I'd genuinely value from a face-off, if you're up for it:
Rules correctness. That's the part I'll defend. If your engine finds a position where mine adjudicates a capture wrongly, mishandles the tournament restriction, or misjudges a terminal position, that's a real bug and I want it. Strength differences I'll accept without argument; rules bugs I'll fix the same day.
Blind spots. You mentioned yours misses the diagonal split three — X _ X _ X. Mine happens to handle that shape, because the evaluator scores by sliding a five-cell window and counting stones rather than looking for solid runs, so a broken three scores like a solid one. But my forcing-move reflexes only recognise contiguous open threes, so my blocking logic is weaker than my evaluation. Different blind spot, same family. I'd be curious which of us misses what.
How we'd actually play it: simplest is hand-relayed — you tell me your engine's move, I play it against mine, I send the reply. Pente games are short enough that this isn't painful. The app exports finished games as plain-text move lists in standard algebraic notation, so we'd both have a clean transcript to argue over afterward. I'd suggest Tournament rules and two games with the first move swapped. If it proves interesting I'll write a proper headless match runner to take the typing out of it.
Your analysis tooling sounds like a different and more serious animal than what I've built — opening book injection, position databases with value markers, threat highlighting. That's a study suite; mine's a game. I'd be interested to hear what it's called and whether the databases are shareable, entirely apart from the face-off.
Posts:
2,293
Registered:
Mar 4, 2007
From:
San Francisco
Age:
45 Home page
Re: Looking for beta testers — a Pente app for iPhone, iPad, and Mac
Posted:
Aug 13, 2026, 11:14 AM
to my understanding its written in java version 6. files are .sgf (smart game files) its open source. AI by mark mammal, pente pro app written by dweebo. some adjustments i think made by rainwolf. im sure he will chime in to clarify, maybe. contains data base of this site, and brainking. com games. and my personal move files, that no one is allowed to ever get their hands on lol. as to testing or head to head, we are in middle of masters tourney currently. so maybe after. again i currently just use android and pc. technically my daughter has a ipad she uses, so, maybe.
Scire hostis animum - Intelligere ludum - Nosce te ipsum - Prima moventur conciliat - Nolite errare
Posts:
2,293
Registered:
Mar 4, 2007
From:
San Francisco
Age:
45 Home page
Re: Looking for beta testers — a Pente app for iPhone, iPad, and Mac
Posted:
Aug 13, 2026, 11:28 AM
rainwolf owns the app currently. he open sourced it i think? unsure. you can ask him. im just a beta tester of it. he has the ability to supply data bases, if he wanted to. not sure if hes giving them out for free or selling or what not. while i can technically physically export them, i think you should just ask him for those files. i did supply him with the brainking data base tho. thats something i can provide if needed. altho it may need to be converted from one file format to another.. if u want i can dig up a sample game, where at certain position, player one can only win by forming a diagnal x_x_x. and you can see if yours can find it,... or does it pick losing path. im also currently testing another pente app... which is hush hush stuff. web based. so is why im hesitant to start testing another one at same time. that and the tourney im in. so maybe in bit.
Scire hostis animum - Intelligere ludum - Nosce te ipsum - Prima moventur conciliat - Nolite errare
1. on 12th move for white only 1 move can win, which is the move shown in the game. which i found on my own.
2. when the position is put through AI, level 12, VCT on, it can not find the move. it instead suggests any other move which all other moves are sure loss. and this is why AI can not be used at high level of play. AI is flawed so far. and it does not have the creativity that humans can do.
Scire hostis animum - Intelligere ludum - Nosce te ipsum - Prima moventur conciliat - Nolite errare
Posts:
2,293
Registered:
Mar 4, 2007
From:
San Francisco
Age:
45 Home page
Re: Looking for beta testers — a Pente app for iPhone, iPad, and Mac
Posted:
Aug 15, 2026, 5:48 AM
this was a tourney game. human vs human. after the round, i poked at certain games in post game analysis, using AI12 vct on. certain key moves to see how it would react, same or different. and in this game i discovered the flaw of the AI at the move position as mentioned. i realized it simply could not see a diag post 3 shape to even check at all. it chose losing paths instead, i found this interesting. humans can spot that easy. AI struggles.
in chess, the game is solved, but in pente, human can play way stronger and more creative. AI is not very creative.
Scire hostis animum - Intelligere ludum - Nosce te ipsum - Prima moventur conciliat - Nolite errare
