GEMBench ripped by AI now?

[exxos]

I'd be interested in what people think about this one.

This Reddit thread was sent to me earlier, for reference: Just a little tease of just how fast our native "ST" is...

A bit of background first. GEMBench 4 was written by its original author, Ofir Gal, who later handed the whole project over to me, source code and everything. I rebuilt it into GB6, which is still maintained, and GB7 is slowly in development. So this isn't an abandoned project, it's still very much alive, even if it moves at the speed of a 1772 seeking track 79.

Going by the video in that thread, "STBench" has clearly taken the GEMBench test routines and the RSC file and put them out under a new name. Nobody asked me about it, and renaming it doesn't really change where it came from. I think it's a bit much to rip anything with AI like this.

To be fair, it might be a bit of pot calling the kettle black, as I'm currently enhancing Super Cars II myself. The difference I see is that the game has been unsupported for over 30 years and nobody is ever going to release an update for it, whereas GEMBench is still actively maintained.

So where do people draw the line? Is it fair game to rip code and resources from software that is still actively maintained, or should you at least ask first?

I'm also a bit confused by the claim that 4.03 had to be used because it was the last version to work with TOS 1.00. I've just run GB6 under TOS 1.00 in Hatari on a stock ST in medium res, and it loads and completes the full test run without any problems.

GB6TOS100.png (28.32 KiB) Viewed 87 times

On the subject of crashes in general, GB6 works the hardware pretty hard, so when it does crash it tends to point to a hardware fault rather than a problem with GB6 itself.

[ijor]

My very personal opinion, is that ripping code from somebody else without asking the original author, is the wrong thing to do, regardless if the software is still being maintained or not. At the very least you should always ask first, and you should certainly always give credit.

Reading the thread, he claims he didn't actually copy the code?:

I'm not copying his code, just trying to create a fair comparison, doing the same tests.

To be honest, this doesn't make much of a difference for me. It might be drastically different from the legal point of view. But ethically, IMHO, if you copied an idea or a concept, even when you evolved the idea an enhanced the code, you should still give enough credit.

[exxos]

Thanks ijor, I agree, asking first and giving credit is really the least anyone should do. Ofir passed GEMBench on to me, but he is still credited for the original, and that is how it should be.

Fair point too that by that measure it applies to my Super Cars II work as well. The original game will always be credited to Magnetic Fields and Gremlin, and anything I release will be an enhancement to the original, not passed off as my own game.

As for "not copying the code", the video shows the same tests, in the same order, using the original RSC and exactly the same test text. It even shows the HGT dialog at the end. HGT is the HiSoft GEM Toolkit, a HiSoft BASIC add-on, and both GB4 and GB6 were written in HiSoft BASIC. That dialog doesn't turn up in a program written from scratch by accident. If you swap the badge on a Ford Escort, it's still an Escort underneath.

[Darklord]

Well, I think it's rude, at the very least.

[exxos]

Well, I think it's rude, at the very least.

Yep. Can clearly see that things thing the blitter tests are identical with identical shapes tested and other tests as well. Its basically GB4 with the GUI ripped out.