Why has there never been a Traveller computer game?

[MrMorden]

I would give my left nut for a really detailed Traveller computer game. Kind of a combination space simulator and RPG. Use an area the size of the Marches so there's plenty of room to roam around.

Make it a game like Morrowind - completely open-ended, but with a main plot you can get involved with at any point if you so choose. If not, you are free to trade, explore, get in fights, whatever.

Ideally, the game would have a cooperative multiplayer mode, so you could have a crew on a Free Trader or Scout/Courier. Space combat could be handled based on your crew slot: pillots fly the ship, gunners shoot, engineers fix damage, etc. On the ground it would be a cooperative first-person game.

Of course, you'd have to have a lot of random adventures on the various worlds, some involving roleplaying and talking to NPCs, and some involving good old action.

The technology is finally here for a full-featured Traveller computer game...no we only have to convince someone to make it.

(BTW, did I read something about a company purchasing the rights from Marc Miller in 1999 to make a Traveller game?)

 

[R_Kane]

There have been 2 to my knowledge...both MegaTraveller...back in the late 80s or early 90s

Somewhere here I've seen a link to a site that has them up as free downloadable programs. Maybe someone could repost that link or direct to the thread that has it.

 

[theSea]

You can find them at Freelance Traveller. Most of the way down the page.

--michael

 

[robject]

About 5 (!) years ago I built a CGI Traveller trade simulator, where players could log in and ply the lanes of the Spinward Marches, trading cargo and ferrying passengers. It was very primitive, but loads of fun just due to the 'feeling' that one was actually 'in' the Spinward Marches, interacting with it on a gut level instead of looking at it like it was a museum piece. It was a sandbox to play in.

I've been trying to write an improved version ever since. Hard to do when it's just a hobby and one is deluded into thinking it ought to be done 'correctly'. Ah well.

 

[kaladorn]

Let me preface this by saying I'm only offering my own personal opinion based on past experiences...

As someone whose title used to read "Senior Network Architect" for a company whose focus was massively multi-user on-line RPGs (for a bit of a niche market, admittedly), and as someone who has designed scaleable client-server architectures and worked on a cross between an First Person Shooter and an RPG, and worked on many large software projects, let me just offer these insights:

1. The hardware requirements to do this on a large scale are non-trivial
2. The management requirements for maintaining the hardware and software, to have this done well and to be worthwhile, are non-trivial
3. The amount of money involved in computer game development may well be over $1 million per (useful and interesting) game produced, in some cases more than that
4. The amount of time and effort that goes into such products far exceeds the amount that $1 million would suggest, as most of the developers and testers get into it and contribute lots of extra time

This is, to do it right and give it any kind of decent look and feel and to provide and kind of RPG aspects, a very expensive and significant undertaking.

AI is .... nasty. Our AI programmers used to do some pretty funky things, and even then the AI were hard to get right. Scaleability of code is a huge challenge. Platform portability and least common denominator design for graphics is a challenge. The massive amounts of work by animators, artists, and modellers required to do something like this anything near to justice would be just staggering.

 

[EvilDrGanymede]

The old 8-bit 'Elite' game (and its sequels) comes pretty close in feel....

 

[Elliot]

Did the Elite guys play Traveller - I always thought it was just too close to the game to be coincidence (beam and pulse lasers, free trading)!

Does the fact that D20 rule have now been put into computer game format (Neverwinter Nights, Temple of Elemental Evil) make it any easier to do a T20 game?

 

[Hecateus]

I for one would be willing to pay monthly fee for a Good Travelleresque Role Playing Game. I am not so sure there are any out there.

p.s. I would settle for good text MUD.

 

[MrMorden]

Originally posted by kaladorn:
Let me preface this by saying I'm only offering my own personal opinion based on past experiences...

As someone whose title used to read "Senior Network Architect" for a company whose focus was massively multi-user on-line RPGs (for a bit of a niche market, admittedly), and as someone who has designed scaleable client-server architectures and worked on a cross between an First Person Shooter and an RPG, and worked on many large software projects, let me just offer these insights:

1. The hardware requirements to do this on a large scale are non-trivial
2. The management requirements for maintaining the hardware and software, to have this done well and to be worthwhile, are non-trivial
3. The amount of money involved in computer game development may well be over $1 million per (useful and interesting) game produced, in some cases more than that
4. The amount of time and effort that goes into such products far exceeds the amount that $1 million would suggest, as most of the developers and testers get into it and contribute lots of extra time

This is, to do it right and give it any kind of decent look and feel and to provide and kind of RPG aspects, a very expensive and significant undertaking.

AI is .... nasty. Our AI programmers used to do some pretty funky things, and even then the AI were hard to get right. Scaleability of code is a huge challenge. Platform portability and least common denominator design for graphics is a challenge. The massive amounts of work by animators, artists, and modellers required to do something like this anything near to justice would be just staggering.

I understand all of these issues. I especially have an understanding of the AI issues, since I have a Master's in AI specializing in game theory, simulations, and emergent behavior.

That said, the games industry is larger than the movie industry by far. Most games take significant resources to develop...I don't think any of this should stand in the way.

If Everquest II can be done, so can Traveller.

BTW, it does not have to be a massively-multiplayer game. Just a 4 person networked game with all other characters being AI could work. Again, Morrowind is a good model for how the AI and character behavior could work. Also, you don't have to model the whole Marches; one subsector can provide a lot of adventure. Model the starports in detail, and then most of the rest of planets could be wilderness areas.

The more combat oriented the game is, the easier the AI is to do. Tactical thinking is pretty easy to model (mainly because it's such a limited domain), while realist human interaction is very hard.

 

[robject]

Originally posted by kaladorn:
[snip]... to do it right and give it any kind of decent look and feel and to provide and kind of RPG aspects, a very expensive and significant undertaking.

Kaladorn's right.

On the other hand, the requirements to do something limited, small-scale, and graphically primitive are reachable, and the main requirement is time and some thought. A rough simulation is do-able. A true RPG game is hard.

 

[EvilDrGanymede]

Originally posted by Elliot:
Did the Elite guys play Traveller - I always thought it was just too close to the game to be coincidence (beam and pulse lasers, free trading)!

IIRC, I think at least one of them did, yes.

 

[kaladorn]

Mr. Morden makes good points. (BTW, Condolences on the Master's in AI... OUCH). One of my buddies is just getting his convocation for Comp Sci PhD today. (Mostly neural net work)

The more you limit the problem domain, the more manageable it is. Of course, the more limited something is, sometimes the less interesting. I recall playing the old Trav games and constantly wanting to do stuff the software prohibited.

One last comment, I don't think that tactical combat is anywhere near as simple as you think, if you want a *very realistic* model of human behaviour. As a game-designer (not just of the computer variety), I will say that the two single hardest 'game' topics to get right (and I have rarely seen it done right) on the skirmish level are command & control and human morale. Morale is far easier to get right at higher levels of abstraction than it is on an individual basis. There is an ungodly amount of psychology involved here, and some physiologogical realities, and it does not all add up neatly. So, you can do a 'good enough' version reasonably easily, as you might suggest. To do a 'very good' version, you've got a problem one or two orders of magnitude harder.

 

[vutpakdi]

Originally posted by Hecateus:
I for one would be willing to pay monthly fee for a Good Travelleresque Role Playing Game. I am not so sure there are any out there.

p.s. I would settle for good text MUD.

Perhaps the online GURPs project will eventually have a GURPS Traveller game?

Ron

 

[MrMorden]

Originally posted by kaladorn:
Mr. Morden makes good points. (BTW, Condolences on the Master's in AI... OUCH). One of my buddies is just getting his convocation for Comp Sci PhD today. (Mostly neural net work)

The more you limit the problem domain, the more manageable it is. Of course, the more limited something is, sometimes the less interesting. I recall playing the old Trav games and constantly wanting to do stuff the software prohibited.

One last comment, I don't think that tactical combat is anywhere near as simple as you think, if you want a *very realistic* model of human behaviour. As a game-designer (not just of the computer variety), I will say that the two single hardest 'game' topics to get right (and I have rarely seen it done right) on the skirmish level are command & control and human morale. Morale is far easier to get right at higher levels of abstraction than it is on an individual basis. There is an ungodly amount of psychology involved here, and some physiologogical realities, and it does not all add up neatly. So, you can do a 'good enough' version reasonably easily, as you might suggest. To do a 'very good' version, you've got a problem one or two orders of magnitude harder.

Congrats to your buddy...my work was in Genetic Algorithms (GAs), which is a type of heuristic search (all problems of intelligence can be reduced to problems of search).

For my thesis I did a 2D simulation of a fighter jet (modelled on flight data from the F/A-18 Hornet) evading a high-speed air-to-air missile. I applied a GA-based classifier system (a type of expert system) to the controls of the aircraft, and over time it got about 40% better at evading the missile than if it just maneuvered randomly. I could have gotten much better results, but I didn't have the time to refine things (and I needed to graduate some time!).

I think the main problem in game AI is that most games use canned tactical scripts (which are very simple to make), which are little more than a bunch of "If-Then" statements. Games would benefit a lot by starting with scripts, but then adding some heuristic method to modify the scripts to adapt to player behavior. You'd end up with Ai doing some startling things, like sneaking around behind the player, etc.

Given the budgets of major software titles these days, and the talent many game companies have, I just don't see the AI as the stumling block here. Probably making a large, in-depth Universe is much harder, just because of the sheer volume of information that must be developed.

But even that is not a huge issue. A clever programmer can make some tools that will generate most of that for him/her. All it takes is some planning and forethought.

Andy

 

[kaladorn]

Mr. Morden (Andy?),

My buddy is working on things like UGVs and the like in a US Think Tank that does next-gen what-if scenarios for military applications. His thesis for his Masters in Comp Sci was on pattern recognition (recognizing various vehicle types at differing resolutions using neural nets).

Training AIs is always one of the tough parts. As you point out, it takes time and to get it better, you need to invest that time. Talib also clued me in to the fact that the order you present data to a neural net system will influence the way the net builds up its weightings. And there are, I believe, training dead ends sometimes. Or at least unpredicted responses.

When we were doing the MMORPG AI, it did use I think GAs or neural nets. It was trained. But it did sometimes develop quirks.

My comment regarding doing the morale and command and control justice was not, in case you were curious, an indictment of the programmers. It *is* possible to create programming that either follows defineable rules or learns from datasets.

However, in the context of single figure morale, the shortcoming is not in the programming. It is in our understanding of those rules and our lack of those datasets. There is a lot of work done on this area, but there are also a lot of contradictions in it. Also, much of the work applies at the company or platoon level. Get down to the individual figure, and many more factors and a wider range of responses can occur. And we're only now coming to understand (slowly) the psychological mechanisms that cause them.

So, since we imperfectly understand the situation and have very little data that is directly useful in this instance, it is effectively impossible to get a working computer model.

That is was my point.

 

[MrMorden]

Sure I hear ya. I wasn't actually arguing with you, I was just discussing the issues involved. No harm, no foul!

 

[robject]

Pardon me while I design out loud; typing this in helps me think.

Currently, I have a model of the TU in the database, plus I have code that lets me materialize bits of the model to client code. That will turn into the actual "Model" object -- the clearinghouse for all data access.

I also have the beginnings of a 'View': a little application that materializes the objects in a solar system and paints them on a panel, updating their positions when necessary. It gets data directly from the Model object; thus there is a specific contract between the two.

Now, each View ought to spawn off a Controller, which is implemented as a listener of the View. It receives control-type events from the panel owned by the View, and generates the appropriate Model Events, which in turn manipulate the Model.

Meanwhile, the Views poll the Model (or they receive a notification) for updated data, and reflect that data using their display logic.

At least, it's a good prototype.

 

[blades]

Well, speaking of someone who's not only played, but owns, both MegaTraveller games. there's some problems with them for modern computers. First off, they are called MegaTraveller and MegaTraveller 2: the Zhodoni Conspiracy. The huge problem thta I've not been abel to overcome, even with Moslo and the like, is the fact that these games run off system clock. Not a problem if you've got an old 8086 or 286 kickign about... so basically, if you'ev got an old Dos box sitting about you're doing good, the older the better. they originally cam eon 5.25" floppies, the first one has no install program, it fits on a a single 3.5" disk and is playable OFF THE DISK! *girn* Good old games. The graphics are certianly dated and blocky, but the game play is cool. The Character creation i nthe first one is fairly simplistic, but *much* more detailed in the second one. Erm...don't really have much mroe to say abotu them other than to wish for a computer to play them on...

 

[Kevinyon]

I enjoyed both of the megatraveller computer games. They were by Paragon, weren't they? We used to use the charector generator out of the second one for creating charectors for games.
I remember seeing a computer magazine ad for Megatraveller 3, but never saw the actual game.
The last Elite that I saw had a very good traveller feel.

 

[Pwyll]

Elite was awesome!