Wednesday 22 February 2023

ChatGPT bandwagon: recommend 1990s games

The ChatGPT bot has a good reputation for making lists and summarizing things, so I asked it to make lists of 1990s computer and video games.

First question

"Give me a recommendation of important 1990s video games, including computer games. Make a balanced list between games from Japan, Europe and the United states. Use popularity and influence as a metric, how well known the games were and how much they influenced later game design. The list could include 20 games."


Sure, here are 20 important video games from the 1990s, grouped by country of origin:


  • Super Mario Bros. 3 (1990, Nintendo)
  • Street Fighter II (1991, Capcom)
  • Final Fantasy VII (1997, Square)
  • The Legend of Zelda: Ocarina of Time (1998, Nintendo)
  • Metal Gear Solid (1998, Konami)


  • Lemmings (1991, DMA Design)
  • Worms (1995, Team17)
  • Theme Park (1994, Bullfrog Productions)
  • Tomb Raider (1996, Core Design)
  • Championship Manager 2 (1995, Sports Interactive)

United States

  • Doom (1993, id Software)
  • Myst (1993, Cyan)
  • SimCity 2000 (1994, Maxis)
  • Half-Life (1998, Valve)
  • Starcraft (1998, Blizzard Entertainment)


  • Sonic the Hedgehog (1991, Sega)
  • Pokémon Red and Green (1996, Game Freak)
  • Resident Evil (1996, Capcom)
  • Shenmue (1999, Sega-AM2)
  • Gran Turismo (1997, Polyphony Digital)

These games were all highly popular and influential, and helped shape the gaming industry as we know it today. Each one has had a lasting impact on game design and player experiences, and many of them are still beloved by gamers today.

My observations:

A nice list, not sure why there's an additional Japan/Europe grouping.

I feel the games are well chosen for both popularity and being prime examples of works that helped establish a genre. There were many significant JRPGs before Final Fantasy VII, but it is the game that cemented the popularity of the series also in the west. Tomb Raider made third-person 3D action games interesting, and so on and on.

"Game Freak" is listed as publisher for Pokémon, although "Nintendo" could have been used.

Championship Manager 2 sounds weird to me, but then again I don't know the genre and its significance at all. Trust the bot with this one?

Second question:

"Please also make a list of obscure computer and video games from 1990s, that many do not know yet still shaped the industry."

Certainly, here are 10 relatively obscure computer and video games from the 1990s that had a significant impact on the industry:

  • System Shock (1994, Looking Glass Technologies): A first-person action-adventure game that pioneered many gameplay mechanics still used in modern games, such as audio diaries and RPG-like character progression.
  • Thief: The Dark Project (1998, Looking Glass Technologies): A first-person stealth game that introduced many now-standard stealth gameplay mechanics, such as hiding in shadows, leaning around corners, and pickpocketing.
  • Ultima Underworld: The Stygian Abyss (1992, Looking Glass Technologies): A first-person action-RPG that was one of the first games to feature fully 3D environments and non-linear gameplay.
  • Syndicate (1993, Bullfrog Productions): A real-time tactical game that popularized the concept of controlling multiple characters simultaneously and the use of cyberpunk themes.
  • Betrayal at Krondor (1993, Dynamix): A role-playing game that featured a complex story and well-developed characters, and was one of the first games to use 3D graphics for its world map.
  • Grim Fandango (1998, LucasArts): A classic adventure game that used a unique art style and a combination of 3D and 2D graphics to create a memorable, film noir-inspired world.
  • Gabriel Knight: Sins of the Fathers (1993, Sierra On-Line): A point-and-click adventure game that blended horror, mystery, and history into a compelling story, and featured some of the best writing and voice acting of its time.
  • Descent (1995, Parallax Software): A 6-degree-of-freedom first-person shooter that allowed full 3D movement and was a technical marvel for its time.
  • Battlezone (1998, Activision): A hybrid first-person shooter and real-time strategy game that used a unique blend of tank combat and base building.
  • Star Control II (1992, Toys for Bob): A sci-fi adventure game that combined space exploration, combat, and role-playing elements to create an immersive and unforgettable experience.

These games may not be as well-known as some of the more mainstream titles from the 1990s, but they all had a significant impact on the industry and helped shape the future of gaming in their own unique ways.

My observations:

These were actually well known games at their time, but I wouldn't except younger people to know these now. I like how the first list included some brands and franchises widely known even now by "non-gamers", whereas none of the same games appear here. The list has become exclusively about PC games.

Interesting that the bot remembers some of my qualifiers from the first question, but chose not to give a Japan/Europe/USA balance here.

I'm maybe less qualified to say whether all these games were both obscure and influential. System Shock and Ultima Underworld alone were portents of things to come, even though the ideas in them were largely abandoned in favor of simpler games like Doom and Tomb Raider.

I didn't check the release years one by one, but I can see the bot gives publishers for some entries and developers for others, a mismatch that one could easily make when compiling such lists.

I'm not sure why I needed to say "please" here.

Third Question:

"Now, one more list with really non-commercial, public domain, fringe and cult computer and video games from the 1990s that were not sold in high street stores."


Sure, here are 10 really obscure, public domain, fringe and cult computer and video games from the 1990s that were not sold in high street stores:

  • ZZT (1991, Tim Sweeney): A shareware game that allowed players to create their own levels and stories using a built-in text editor and simple graphics.
  • ADRIFT (1995, Campbell Wild): A text adventure game creation system that allowed users to create their own interactive fiction games using a simple interface.
  • Captain Blood (1990, Exxos): A surreal and trippy adventure game that takes place in a strange alien universe.
  • The 7th Guest (1993, Trilobyte): A puzzle game with live-action video sequences and a horror theme, which was one of the first games to be released exclusively on CD-ROM.
  • Bolo (1993, Stuart Cheshire): A real-time strategy game that focused on tank combat and was one of the earliest games to support multiplayer over a network.
  • Angband (1990, Alex Cutler): A roguelike game that features randomly generated levels and permadeath, and is still actively developed and played today.
  • Slime Volleyball (1998, Quin Pendragon): A simple and addictive one-on-one volleyball game featuring blobs of slime.
  • WinTrek (1995, Dan Baker): A multiplayer game that allows players to command their own starship and engage in battles with other players over the internet.
  • Nethack (1993, The NetHack DevTeam): A classic roguelike game that features randomly generated levels and a vast array of items, spells, and enemies.
  • The Incredible Machine (1993, Kevin Ryan and Jeff Tunnell): A puzzle game that challenges players to use various everyday objects to create Rube Goldberg-style machines to complete a goal.

These games were not sold in high street stores and are relatively unknown to most people, but they still had a significant impact on the gaming industry and helped inspire future generations of game developers.

My observations:

Ok, that's pretty obscure and included also games I haven't heard of. Still, the "not sold in high street stores" part was not taken very strongly, because Captain Blood and The Incredible Machine were commercial games.

The bot may have considered the categories as broadly inclusive, because certainly Captain Blood is a cult game.

Here I felt the need to actually check the list entries.

One entry I consider a full error. ADRIFT appears to be an adventure-authoring system, and not a game as such. Besides, it was released in 2002 according to Wikipedia. I didn't challenge the bot about it, but presumable it would argue the author already was working on it in the 1990s.

Captain Blood was, according to many sources, released in 1988.

Bolo might have been originally released in 1987, but I'll give the bot a pass as the later Macintosh version is probably the more significant one.

Angband probably has more credits than "Alex Cutler".

According to sources I could find, Wintrek was not authored by Dan Baker but by Joseph Jaworski in 1992.

Slime Volleyball has apparently a somewhat contested authorship and year of release.


The bot did a good job in my opinion, it mostly didn't invent bullshit or non-existing products, kept in 1990s almost entirely, and made relevant suggestions. For the last category, there seem to be more uncertainty about authors and release years.

I didn't ask for game descriptions, but these were actually relevant for the less known games and as far as I see they are quite correct too.

Sunday 19 February 2023

Multipaint 2023

With no Multipaint "yearly" release for 2021 and 2022, I thought it was high time to get a version out.

The hiatus was at least partly because of Covid. But maybe not in the way one would expect. You see, for some reason 2020-2021 was at first a time for ambitious new ideas and concentrating on hobbies. This led me to create a huge internal overhaul for Multipaint that eventually became something of a burden, locking me out of a release for a long while.

Often it was like "It would be nice to get that pan tool feature out, but the script features are far from being finished!"

GUI and functions

Indeed from the user point of view one of the biggest additions is the pan tool. In magnify, you can either use the "hand" icon, or hold down the middle mouse button to drag the view.

In the full screen view the pan tool doesn't do anything, which is not that intuitive. Programs like GIMP allow you to drag the full image around, which is kind of useful when preparing to zoom in.

The Multipaint viewport may need rethinking in the future, but I would also like to preserve the 8-bit/16-bit style "full screen" drawing.

The text in menus and buttons now also uses both upper and lower case, which I hope is slightly more readable than the past UPPER-CASE menus. The characters were tweaked a little too, but as it's a monospace, there are always some compromises.

Multipaint 2020 scaled the contents of the window depending on how the full screen fit into the application window. Resizing the window switches between 1-3x scales. Now it's also possible to force one of the scales from the Other menu.

This can be helpful for some situations where you need to fit Multipaint to a low resolution laptop screen. The viewport may overlap the tool palettes, which I didn't fix for now. Using the forced scales, the user is responsible for setting the window size anyway.

The file selector caused some problems for Windows users, which I already knew from Marq's PETSCII editor. Instead of revamping the themed file selector code I disabled it for Windows altogether, which according to one test helped enough. From prefs.txt it is possible to force it back on, in case someone wants to test if the slightly better file selector works.

I adjusted the grid terminology a little. I had happily used "8x8" grid for what was technically a "4x8" grid in wide pixel modes, now it ought to be more consistent.

New Modes

A kind of Amiga mode has been "unofficially" part of Multipaint for a while. However, Multipaint 2023 finally has a resizable, bitplane-adjustable mode that I could properly call Amiga.

Currently it only exports IFF, and cannot load it in, as I found it to be somewhat complex. What kind of IFF would I support? Possibly a Deluxe Paint 3 output would be enough.

Obviously it can handle the Multipaint "*.bin" project format and png load/save, so it's not something to worry too much.

The resize-function is still rather limited, and there's an invisible memory limit to which the mode has to fit into. I consider this an Amiga "low resolution" mode although you can use 1 bitplane and create a 640x400 bitmap.

I guess this would also work as an Atari STE mode, although I haven't written a Degas exporter specific for this mode.

Why even do an Amiga or Atari ST? I no longer remember. Multipaint originally concentrated on 8-bit modes, and it might have been good to hold onto this constraint. For a while, Sinclair QL was my favorite retro machine, resulting in a new mode. Then Atari ST was done as it could be done just as well, and then I began to think what an Amiga mode needed.

VIC-20 hires and multicolor modes were a late addition. I have wanted VIC-20 modes for a long time, but was always somewhat confused with the different implementations, some of which required extra memory and even extra hardware. Aleksi Eeben (thank you!) supplied practically a white paper with specs and examples and a format that could make most out of the unexpanded VIC. I could not resist adding the modes.

Extras Menu

I added an Extras dialogue in the File menu for running less complete, experimental or non-standard modes. I couldn't live with the idea the main platform selector was cluttered with these, but I still wanted more easy access to them for the user. In the future I may revamp the platform selector and again combine the menus somehow.

One of these is the ULAplus direct mode, which is like ZX ULAplus, but the mode doesn't attempt the notorious "live" palette adjustments of the original mode. So it's bit like working on a normal ZX Spectrum with 64 color adjustable palette.

If you turn over to the real ULAplus mode, Multipaint will then convert the image to the real platform limits. Although I've allowed 64 colors, I'd recommend using maximum of about 16, especially because there seems to be something a little wonky about the internal conversion still.

I added a C64 multicolor free mode, similar to the C64 hires free mode. This is meant for experimentation with non-standard C64 ideas, such as mock-up sprites or border effects, or create larger than screen-size bitmaps. Or make pretend you are in a FLI or some such mode.

I also enabled the resize-function for these two modes. Admittedly Multipaint is currently not very good at handling larger screens. I have some ideas about how to expand the usefulness of these modes.

I was asked to add an Amstrad GX4000/plus mode, and it turned out so easy to create I put it to the Extras. This computer enjoyed some popularity in France, but it is not well known for the rest of the world. I do remember it was advertised in Finland. The mode is 160x200 with 16 colors out of 4096 maximum.


The biggest change to Multipaint is not really visible to the user at all, the external script functionality.

All the export/import functions were previously written inside the Multipaint source code. Although I could reuse components, this approach began to look like a growing problem if I were to add more modes. Likewise, the component-approach meant that if something broke it could affect multiple other modes. The text exporter was also getting to be a little messy.

Now, Multipaint loads a boot.txt script that defines what other scripts are attached to the current platform File menu and what role they are to play. These scripts in turn use a primitive BASIC-like language to convert the Multipaint internal format into an 8-bit format or executable.

Of course the interpreter takes room in the Multipaint source, but adding new import/export scripts doesn't clutter the source and each platform has a self-contained script. Well, the Amstrad CPC disk baker uses two layers of scripts which I don't like, but it was necessary.

I also encountered the fact that a lazy implementation of a script interpreter isn't especially fast in Java/Processing as the string appending, clipping and concatenation can be costly there. Making it faster took some while, and I can only hope the export/imports are rapid enough.

I'm not yet prepared to say it's a good idea to create your own scripts, but it is possible, they are located in the data folder just as the prefs.txt is.

Into the future

There's actually a bunch of to-do ideas hanging around for a long while, yet I didn't get them into this release version. Now that the scripting idea works I could consider some more lightweight additions.

One future idea is to enable user-creation of modes that are similar to the already existing ones. For example, Amstrad GX4000 was basically a combination of already existing modes and palette selector definition. Almost anyone could have added the mode, except the Multipaint source is so convoluted only I know my way around it.

However, the VIC-20 modes, despite seemingly based on similar concepts as other modes, still required quite a lot of tweaking. So, who knows.

I probably need to look into Processing 4, just to keep up with the official developments.

Multipaint web site:

Sunday 5 February 2023

Jill Lepore: IF THEN

I was lured in by the initial impression this book would have stories of old-timey punch card computers and their operation. Not really. This is more of a political story revolving around computers and their application in behavioral science-led questions. And all the better for that I think.

If Then: How One Data Company Invented The Future, from the historian Jill Lepore, weaves together politics, university life, zeitgeist and popular culture, in an attempt to explain how computer prediction and data analysis came to impact on politics and public perception in the 1960s.

The prologue and backdrop is the famous event of televising the 1952 presidential election, doubled with a televised UNIVAC prediction of the outcome. For some inventive minds, the event suggested that instead of using computer to report vote results faster, a computer might influence who could get elected.


Through the era of Dwight Eisenhower, John Kennedy, Lyndon Jonhson, Vietnam, Martin Luther King, emerges an unlikely narrative of the Simulmatics Corporation, a company that used behavioral science-based computer modelling to predict people's actions. A "people machine". 

The Kennedy election in 1960 was followed by a bold claim from Simulmatics that their reports, based on number crunching and 480 voter types, resulted in policy suggestions that landed Kennedy the presidency. Through mist of assertions, denials, counter-denials, fiction and non-fiction, it is probably impossible to say what the real impact was.

Among the cast of characters, the author details the involvement of Eugene Burdick, now a largely forgotten Hollywood-celebrity author of political thrillers with a sci-fi bent. His book The 480, was a fictional account of Simulmatics, and not from a neutral perspective either. 

Although the claimed influence resulted in a sort of a scandal in itself, the novel probably ignited more debate and public opinion about computers than anything else. The fear of computers predicting what people will do and government reacting to it before people themselves knew what they were about to do, seemed to become real rather soon. Whatever the truth, the idea of a candidate led by a computer entered into the popular culture big time, reducing voters into punch-card caricatures with no agency of their own.

Simulmatics banged their drum with the Kennedy report, also basking in the (somewhat indirect) infamy of The 480, but didn't have it easy with subsequent clients. Politicians and the administration were understandably not too keen to rely on the company, and other industries didn't have that great data or didn't want to give it away. 

Rather quickly, advertisement companies with long history and large databases could do their own computing, and Simulmatics was left in the sidelines in the field they helped create.

Vietnam war provided a life extension for Simulmatics, as they set up shop in Saigon and attempted to collect data for predicting insurgencies and find opportunities for propaganda and soft warfare. Simulmatics was hardly alone in this, but at least in this telling it looks theirs was the most incompetent and shoddy research setup, an exemplary of how not to conduct social research in another country.

In meantime, administration had changed and Nixon hardly had use for an outside predictor company, as voter segmenting and balancing segment responses to policy changes with estimated votes, were already old hat. (It's unclear whether Simulmatics offered to politics anything new in the first place, except more speed and "real time".)

The book then continues exploring the latter days of Simulmatics, its entanglement to the civil right questions and the black vote issue, a supposed linchpin of the Simulmatics report for Kennedy. Back in USA, Simulmatics engaged in predicting not rural insurgencies in another country, but riots in the urban areas. In fact it seems "predicting riots" was something of an ongoing sub-genre in the area of behavioral science applications.

The author is then able to weave and knot the story of the corporation to the moment when the two political parties began to polarize in the way we now recognize. This is perhaps more zeitgeist represented by the story of Simulmatics, rather than Simulmatics or computers being a direct root cause for it.

What strikes as both funny and terrifying is the unprincipled nature, even corruptness of it all. In absence of data, Simulmatics would cobble up some, patch it with educated guesses, then present results to clients. Again, this speaks likely more about the time than about Simulmatics or the field of computer-based behavioral science. But I can’t avoid having flashbacks to university innovation and research activity.

Of the central cast, to me the only relatively familiar figure is Ithiel de Sola Pool, whose name became tarnished through the Simulmatics/Vietnam association. MIT students protested vocally against defense funding of research projects. The opposition was even extended to the openly presented and rather innocent Arpanet, the consequences of which hardly anyone understood at the time.

Here these two developments at least touch each other in history. Simulmatics, representing the more graspable idea of government control over people through mainframes, and "reducing people into punchcards", was still a name brand at these debates. It was soon forgotten from history whereas the developments that grew out of Arpanet grew into global importance.

Later, Pool was able to predict in great accuracy the consequences of a world-wide internet: in principle anyone can have personalized newfeeds and therefore live in reality-bubbles of their own, or get behind fringe ideas by networking in ways that were impossible before.

Recounting the legacy of Simulmatics, Lepore appears to liken Media Lab as a continuation of the practice of milking government money for grand promises of technical revolutions, a "Simulmatics scheme" of sorts. Not sure if I got this correctly, or perhaps this is a case of Harvard bashing MIT or something.

Looking Back

The book makes the always gleeful and enjoyable demonstration that a technology we thought was new, was actually used a long time ago and largely abandoned. Perhaps even for moral and ethical reasons. And then as time passed it came to be done anyway, by the Cambridge Analyticas and the Facebooks of our age.

This requires the insight that the mainframe-as-predictor, and the large-scale internet practices of user data manipulation can be juxtaposed, and to make say something about each other. What could have been a historical footnote becomes more important through the lens of the present.

Apart from mentioning IBM computer models and bit about FORTRAN and punch card tech, the hardware is not meticulously explored. The algorithms are simplified into kind of IF-THEN evaluations (hence the title), so as not to confuse the more casual reader. Then again, the US election system is not explained in any great accuracy either, so the text is able to cast an overview of a web of events without boring anyone greatly.

From today’s perspective, Simulmatics was a fledgling attempt at best, and would seem naive if it wasn't acted out against the backdrop of presidential campaigns, civil right questions, the Vietnam war, and a world on brink of nuclear destruction. Certainly the story is a precursor to the data analytics practices later embraced by any large-scale commercial or political internet operation.

It's also darkly funny, this story of the would-be Merlins in Kennedy's collapsed Camelot. The writing is infectious, making me feel like I'm suddenly an expert in areas I had no knowledge of prior to reading this book. Perhaps the author also borrowed a leaf from Eugene Burdick's supposedly lurid and unsubtle prose, as if to show the pen is mightier than the punch-card after all.