Zachtronics is a contender for the greatest game studio you've never heard of.
If you're a fan of Factorio, you owe it to yourself to play SpaceChem, TIS-100, Shenzhen I/O, Opus Magnum, Exapunks, and Last Call BBS.
(In fact, since Factorio was inspired by a Minecraft mod, and since Zachtronics' Infiniminer was the direct inspiration for Minecraft, there's no Factorio without Zachtronics!)
I think the best introductory recommendation would be the missing item: Infinifactory.
It may not be as nakedly algorithmic as some of the others, but the visual impact of a solution is very satisfying--you might even be able to show it of to family members without their eyes totally glazing over. :p
I tend to recommend Opus Magnum as an introduction, it's a lot easier to plan with, nice to look at, and has a somewhat more engaging story than most.
I'm a big fan of all of the puzzle ones, though. SpaceChem had a huge impact on me, my most-viewed YouTube video is this one little clip from January 2, https://youtu.be/dlJmKqi6EEc . But very hard to explain why it occupied my whole soul for a week!
I tend to recommend that people ignore any recommendation to start with a specific Zachtronics game (unless they're getting the recommendation from someone who knows them personally). Different flavors of nerds seem to have vastly different preferences for the games, so it's probably much better to take a few minutes and check how interesting the core mechanic of each one is to you and pick based on that.
Personally I enjoy working with electronics and assembly, and that translated to me really liking TIS-100 and Shenzhen I/O. Meanwhile Opus Magnum wasn't anywhere near as interesting to me and felt like kind of a slog.
Another couple of games in the same vein are Human Resource Machine and its "sequel" Seven Billion Humans (by roughly the same people who gave us World of Goo many years ago).
Yes! Spacechem is my favorite game ever, I finished the game a few times on tablet version (which has a few less puzzles than the desktop one if I remember correctly). Their other puzzle games are great too, but there is some minimalism and the simple tablet compatible UI (unlike say TIS100 which is minimalist in a way, but the interface is not as great - Spacechem can be played by kids easily) which for me makes it superior to the others.
But the aesthetics are completely different and in a way different game mechanics. Both computer puzzles. No doubt TC is literally the representation of a computer as the goal of the main storyline.
I liked nearly every games by zachtronics (I even enjoyed Ironclads tactics which I don't think was as popular as some others) but not infinyfactory. I couldn't find the added value of 3d in a puzzle game.
Well, then you're about to discover something that might very much fit your interests.
I am not a huge fan because it feels a bit too close to actual work, but TIS-100 has been on my to-play list for a long time (I've bought it ages ago).
TIS-100 is great. The “mesh of many tiny cores” architecture is cool, and also somewhat mind-bending — but the simplicity of the TIS design makes it just about possible to get your head around it.
After playing TIS a bit I found it really interesting to read about the Transputers and the Connection Machines, two similar real-world architectures.
David Ackley’s T2 Tile project[0] and Movable Feast Machine[1] look similar to me too, but they take the idea much further; the aim is to create an infinitely scalable and totally decentralized architecture. I only know a little about it, but it’s super cool stuff.
If you liked Transputers, you might want to also read about Adapteva and their Epiphany core for a more recent attempt at something similar-ish.
I still have two of their prototype machines from their Kickstarter - two ARM cores to run Linux, with an Epiphany chip with 16 cores in a 4x4 grid. But their goal was scaling it up to 64 cores or up to I think 4K cores on a board. Each core had a small amount of on core RAM and four buses to each side in the grid, and you could access the memory of every other core with a predictable latency (one cycle per "hop"), so if you planned things carefully, you could have them working in lockstep.
It's an interesting space, but hard because the first difficult question you need to answer - which strips away a whole lot of potential use-cases and many of the most profitable one - is "why not a GPU?".
Just before Epiphany, there was also the Tilera, which had a lot in common with the Transputer. Our lab got one and we played around with it, but it was a pain to program. Transputer had OCCAM, Tilera chased after the C model and shared coherent memory.
The Tilera TILE architecture lives on in NVIDIA's DPU.
I've finished Exapunks. Should go back and finish TIS-100.
One thing I both love & hate about Zachtronics is the Histogram of Doom that appears at the end of each problem. There's no better feeling than ending up in the top ranks for efficiency or speed. And no worse feeling than refactoring everything and still not budging from the middle of the pack.
Considering you're being ranked against other players, in some ways this makes a Zachtronics game one of the more vicious multiplayer games. I mean, you're never playing against another player directly. But you're always being compared to others and it can be humbling at times.
I wonder how hard of a problem detecting this would be, with the goal of unskewing the scores?
Could they simply hash your solution and see if you match someone else exactly? What if you give someone credit for exact matches because they don't have exact matches on other problems and sometimes it's just more likely? What are some other ways this could be evaluated?
It wouldn't be too hard, you'd need a model for baseline improvement which should be obtainable by those who do improve slowly and incrementally.
Instead of using a probability density field on the final submissions I would use one based on the likelihood of a person getting to that point within the time they took ~ skill * time.
You could ask on submission if it's a worked solution so then you're only looking for false submissions, but since there is no prize for getting a high score there would be little incentive to lie. Those using worked submissions marked as their own for their own ego would have an unusual submission pattern.
There are many layout options for the same algorithms and it's likely that a rough percentage of worked solutions could be obtained where a specific layout appears much more often than it should.
Worked submissions are also much more likely to be final submissions.
Not sure if I remember correctly, but maybe you need to give a solution before progressing. I would make it possible to progress without giving a solution to take away that incentive.
As a software developer who occasionally having fun with assembly coding, I think this Shenzen/IO are the finest coding games by Zachtronics (and probably in general, ever).
I'd love an updated/improved version of CoreWar, but probably it's too geeky for most people. Oh well...
Corewar is fun. Someone should include it as an optional feature in a cyberpunk game. So to crack open electronic locks you need to fight computer warriors.
Exapunks just never clicked for me in the way that the other Zach games did. Perhaps it was too visual, or perhaps the problems didn't sit well with how I problem solve.
I don't like to look at guides, yet it took me a long time to hit a wall in TIS-100 . But the wall I did hit was "Signal Window Filter" which involved delayed state.
A lot of the exapunk problems felt much closer to that style of problem, so perhaps it's just a blindspot of my own reasoning.
Would them sharing their steam page with all the badges showing they are in the top 1% of most the solutions kind of similar to sharing their GitHub profile?
One thing I enjoyed about TIS-100 was trying to get the low cycles or low instruction count. I remember the "Aha!" moment when I discovered that it often not possible to get both at the same time.
I still haven't finished all the levels, I should really finish it one day.
I feel the same about Factorio, but I love the Zachtronics puzzles. I think it's because each problem is self-contained, so the active scope is generally what can fit in my head at once. Otherwise I need to plan things out and it starts being work.
Opus Magnum is a good entry point. If not that then Last BBS. I know most his games but only dedicated myself to these two. I’m trying to “finish” the main stories and then move to TIS100 or Shenzen
If you're here, you enjoy zachtronics games. We talk a lot about their programming games, but also check out Eliza, the visual Novel they made. It's great.
If you're a fan of Factorio, you owe it to yourself to play SpaceChem, TIS-100, Shenzhen I/O, Opus Magnum, Exapunks, and Last Call BBS.
(In fact, since Factorio was inspired by a Minecraft mod, and since Zachtronics' Infiniminer was the direct inspiration for Minecraft, there's no Factorio without Zachtronics!)
It may not be as nakedly algorithmic as some of the others, but the visual impact of a solution is very satisfying--you might even be able to show it of to family members without their eyes totally glazing over. :p
I'm a big fan of all of the puzzle ones, though. SpaceChem had a huge impact on me, my most-viewed YouTube video is this one little clip from January 2, https://youtu.be/dlJmKqi6EEc . But very hard to explain why it occupied my whole soul for a week!
Personally I enjoy working with electronics and assembly, and that translated to me really liking TIS-100 and Shenzhen I/O. Meanwhile Opus Magnum wasn't anywhere near as interesting to me and felt like kind of a slog.
http://thesiteformerlyknownas.zachtronicsindustries.com/kohc...
For me the motivation just never materializes. Contrast this to 100s of hours w/ each of Factorio, Satsifactory, and DSP amongst others.
We are on HN
xkcd lucky ten thousand
If getting 1 in 10,0000 right makes you a contender.
>Zachtronics
>Never heard of
Pick two.
To discover things we haven't heard of. That's why it's Hacker News, not Hacker Common Knowledge.
I keep running into his games whenever I'm on the lookout for games, and he is heavily targeting engineer-minded people.
I am not a huge fan because it feels a bit too close to actual work, but TIS-100 has been on my to-play list for a long time (I've bought it ages ago).
After playing TIS a bit I found it really interesting to read about the Transputers and the Connection Machines, two similar real-world architectures.
David Ackley’s T2 Tile project[0] and Movable Feast Machine[1] look similar to me too, but they take the idea much further; the aim is to create an infinitely scalable and totally decentralized architecture. I only know a little about it, but it’s super cool stuff.
[0] https://t2tile.com/ [1] https://movablefeastmachine.org/
I still have two of their prototype machines from their Kickstarter - two ARM cores to run Linux, with an Epiphany chip with 16 cores in a 4x4 grid. But their goal was scaling it up to 64 cores or up to I think 4K cores on a board. Each core had a small amount of on core RAM and four buses to each side in the grid, and you could access the memory of every other core with a predictable latency (one cycle per "hop"), so if you planned things carefully, you could have them working in lockstep.
It's an interesting space, but hard because the first difficult question you need to answer - which strips away a whole lot of potential use-cases and many of the most profitable one - is "why not a GPU?".
One thing I both love & hate about Zachtronics is the Histogram of Doom that appears at the end of each problem. There's no better feeling than ending up in the top ranks for efficiency or speed. And no worse feeling than refactoring everything and still not budging from the middle of the pack.
Considering you're being ranked against other players, in some ways this makes a Zachtronics game one of the more vicious multiplayer games. I mean, you're never playing against another player directly. But you're always being compared to others and it can be humbling at times.
Could they simply hash your solution and see if you match someone else exactly? What if you give someone credit for exact matches because they don't have exact matches on other problems and sometimes it's just more likely? What are some other ways this could be evaluated?
Instead of using a probability density field on the final submissions I would use one based on the likelihood of a person getting to that point within the time they took ~ skill * time.
You could ask on submission if it's a worked solution so then you're only looking for false submissions, but since there is no prize for getting a high score there would be little incentive to lie. Those using worked submissions marked as their own for their own ego would have an unusual submission pattern.
There are many layout options for the same algorithms and it's likely that a rough percentage of worked solutions could be obtained where a specific layout appears much more often than it should.
Worked submissions are also much more likely to be final submissions.
Not sure if I remember correctly, but maybe you need to give a solution before progressing. I would make it possible to progress without giving a solution to take away that incentive.
I'd love an updated/improved version of CoreWar, but probably it's too geeky for most people. Oh well...
Do you think TIS-100 is better or worse? It's a strange omission from your list.
Is it that you never played it, or do you think TIS-100 is superior (asking because I haven't tried TIS-100)
Exapunks just never clicked for me in the way that the other Zach games did. Perhaps it was too visual, or perhaps the problems didn't sit well with how I problem solve.
I don't like to look at guides, yet it took me a long time to hit a wall in TIS-100 . But the wall I did hit was "Signal Window Filter" which involved delayed state.
A lot of the exapunk problems felt much closer to that style of problem, so perhaps it's just a blindspot of my own reasoning.
EXAPUNKS kinda felt like Robocode, a bit.
I still haven't finished all the levels, I should really finish it one day.
1) just give me a real ide (my issue with pico8 as well)?
2) If I'm going to play a game that's close to doing FPGA stuff why not just learn to program an FPGA, that would be more useful
10/10 asm is life
This game is great, but is geared more towards digital logic instead of coding.