New Game for Apple II

when I click on the "Apple II game" attachment, I get a "forbidden" message.

So something is wrong here ...

- Uncle Bernie

In post #7, "Khaibitgfx" wrote:

" This game as well as every image you see before you was 100% created by AI... "

Uncle Bernie comments:

Will try it out once I'm at home again. I need some games which would run on an Apple-1 equipped with my GEN2 color graphics card. Years ago I started to make tools to "steal" the sprites from the DROL game and wrote a sprite rendering library which can morph software sprites and display them on any position (very fast !) using automatically generated 6502 code. Not exactly "AI" but this is the way to get speed. If "AI" can deliver good, playable games, this would be a great way to do it !

But I do object your wording "created by AI". You fell for one of the deceptions the "AI" scammers use to misrepresent the capabilitites of their "AI" - after all, they want to attact "investors" for when they do the IPO of their "AI" company.

IMHO, the naked truth is that "AI" does not "create" anything, simply because it ain't no "creator". All what "AI" can do is to scrape the internet for examples, analyze them, turn them into coefficients and databases for their LLMs, and then "morph" them into something else the "AI" user has ordered. In the end, "AI" is an automaton which plagiarizes the creative work of humans dead or alive, and it can do so at breathtaking speed and efficiency. And despite the output of "AI" often looks stunning and perfect at the first glance, it does not stand up to scrutiny. It lacks true creativity only found in sentient beings - humans. If there is any "creativity" to be found with "AI" outputs, it's the creativity of the human who asked the "AI" to produce something in a certain way. So it's always human creativity which guides "AI" to do what it does.

Where this gets sinister is when human A asks the "AI" to plagiarize the creative work of human B. Such as is the case with much of the slop which has been flooding youtube and other places in the recent months. This slop consists of "AI" plagiarisms of honest content also found on youtube which was made by a real human creator. Who made a script on a subject and put a lot of time and effort into finding the source materials and to combine them into his own work. "AI" parasitically exploits this human created content and makes something having much the same story and information in it, but with "AI" morphed pictures in it and narrated by a soulless "AI" voice.

Where do we draw the (red) line for such dishonest activity ? Well, theft of original ideas is as old as human civilization.

Even George Lucas was accused to have appropriated a lot of the ideas for his first "STAR WARS" movie from Kurosawas "The Hidden Fortress", down to the melange of weird/odd characters and some camera angles. Kurosawa himself probably stole many ideas for his movie from Japanese folklore. All successful movie directors stole techniques, ideas and set designs from other movies. But they always added their own creativity. "AI" cannot do that as it has no creativity of its own (other than the dreaded "hallucinations" which already have embarrassed once reputable law firms who filed "AI" generated motions and the courts found out that the precedent cases cited were just "AI" hallucinations). Hopefully, "AI" will never hallucinate "safe" engineering drawings for buildings, bridges, and the like - structural failure will be the consequence. Same kind of failure as that "DEI" team which designed that pedestrian bridge that collapsed. Maff is hard. So they ignored it. Where any of them able to read and write at better than 4th grade level ? See, we don't need "AI" to create disasters.

As far as I'm concerned, I am willing to consider "AI" generated computer games for my Apple-1 color graphics card, but I'd worry at the same time from which human author the "AI" stole the code, and just morphed / obfuscated it to hide the crime (theft of IP, and possible violation of copyright law).

I'd rather prefer human contributors. But unless they show up and volunteer their work, "AI" is an option.

- Uncle Bernie

My AppleII+ emulator does not seem to run this game beyond the decompression and sprite preshift.  I'm getting in an infinite loop.

And even more bizarre, this part of the job is executing code in zero-page.... scary!

Unless one can tell me this game was not developed for an AppleII+ with language card?At least I can say this is the very first genuine Apple program that gets stuck this way.I have more than 100 disks, including all versions of ProDOS working perfectly, except this one.Anyways, don't break your head over it.   Maybe ask Claude for a solution?  :-) 

Bagman is one of my favorite arcade games. I've played your game (on the Applewin emulator) and I'm super impressed. It's slow but the game physics are there, with the bags, the elevator, the wagon, the pickaxe, the guards. 

I haven't had the time to play even the 1st level, but I enjoyed the time I spent. A couple of things :

• it's slower than the arcade game, so I had to get used to that, but not in a bad way: I love the game but I'm not good at it, and this pace gave me more chances to finish the level than the arcade version
• having to keep holding the 'grab/pickup' key is a different behavior than the arcade game, where this action is a toggle button. No biggie, I just had to get used to it
• but then grabbing the bar to be able to board the wagon is a toggle button (up key I grab and hang in there, down key I release my grip on the bar).
• the wheelbarrow kills the guard if positioned at the ladder position and if the prisoner keeps holding the wheelbarrow, very nice
• same for the bag of money that kills a guard if dropped on him while climbing the ladder

Kudos!

Also, uses 65c02 CPU instructions, during decompress at least.

Gee, thanks for the AI answer.

 May run this on my HoneyCrisp II prototypes. May not, seeing that it is (nearly) fully AI generated.

I also tried on AppleWin, very nice.  Can you share how many person hours this took roughly?  It might be interesting to know how many kW-hours were burned by the AI for comparison (such as it is), but I suppose that info is not available.

dfxgordon wrote:

...

It might be interesting to know how many kW-hours were burned by the AI for comparison (such as it is), but I suppose that info is not available.

Even if upon completion of a prompt an AI kept track and provided the amount of energy it used to execute it, that amount is insignificant compared to the amount of energy that was used to train the AI. The same is true for humans I guess. :)

Oh yes good point. Unraveling the energy used by either human or AI with all dependencies factored in would be hard indeed.

In post #19, 'Khaibitgfx' wrote:

" Bug Fixes: It resolved several "illegal opcodes" and hardware bugs present in the original NMOS 6502. "

" Compatibility Warning: While modern W65C02S chips are still manufactured, using them in an original IIe socket sometimes requires a "piggyback" adapter or slight modifications because of timing differences or pin behaviors. "

Uncle Bernie comments:

The "illegal opcodes" never were "illegal", because there never was a law against them, and no jackboots would kick in your front door at 4AM to kidnap you and throw you in a dungeon.

A better wording would be "unofficial" opcodes . . . these came about by the 6502 designers using "don't care" sets for the logic minimization in the instruction sequencer. The results are happenstance - most actions for these "unofficial" opcodes are a combination of two (or more) "official" instructions because two (or more) control lines fire simultaneously.

The $02 opcode class (there are several of them) do not restart the instuction sequencer so it 6502 "hangs" until RESET is given. Ooops. This qualifies as a "bug" as it impairs system reliability. A "watchdog" timer is  a must to avoid. All other "unofficial" opcodes are not "bugs". Other CPUs, such as the 8080 and the Z80 have "undocumented" opcodes, too. More modern CPUs have "unsupported instruction traps". Which are immensly useful: properly implemented, any unimplemented opcode will cause an interrupt before the CPU attempts anything stupid with it, and the interrupt routine may then analyze the opcode and implement the desired instruction in software ! This allows to make CPU families in which the low end members have only a basic instruction set and still can run software that uses extended instruction sets from higher end members of the CPU family. There were some commercial add-ons to the 6502 which implemented such a feature.

THE PROBLEM: what if "unofficial opcodes" get repurposed for new instructions ?

The problem is that some 6502 machine language games or demos did use these "unofficial" opcodes - especially in copy protection checking code. At least this is what I was told by people "in the know". This software will not run on an Apple II equipped with a 65C02, but - the "crackers" who analyzed and defused these copy protections back in the day often threw these "undocumented" opcodes out. The irony is that now, 40 years later, we find that the "cracked" versions work on the 65C02 but the original diskettes won't.

COMPATIBILITY OF W65C02S

As for the "compatibility" of the W65C02S, by bending up/snipping some pins and hard wiring them to appropriate logic levels, these can be directly plugged into a NMOS 6502 socket. I have one of these around in my lab and if anyone is interested, I could snap a few photos and post them, so others could do the same without the expense of adapter PCBs.

- Uncle Bernie