Topic: main roads becoming normal again

teppo wrote:

This is the first time I feel the temptation..

I felt it too

Top

Quote

king_of_nowhere wrote:

In later game, when producing pack animals is not an issue, I would just like more roads to stay promoted. So I definitely would add options to manually choose to promote roads, or to manually deactivate the demoting mechanism

I'm not convinced that underutilized roads should stay promoted for ever in later game. But if that feature should get included in the new algorithm, it would be as easy to approximate as skipping the animal-maintenance cost whenever the empire had animals in stock. Similarly, promoting roads could become easier in later game, by simply making the animal cost inversely proportional to the total number of animals in the empire (so all the player would have to do would be to produce more animals). And those are two more examples of how the wallet model can simplify the automation of things which previously were too hard to be thought feasible in ways other than manual options.

Top

Quote

king_of_nowhere wrote:

Your hint is wrong. The correct one starts with "B" and ends with "rain". That's just it. You can't use the fact that the AI sucks at managing roads as an argument. Heck, half the maps out there, the AI can't make a full economy on them because they are small. Doesn't make them bad maps.

I didn't say that maps are bad, I said that they are special. And some of your own maps are both special and funny, they are not bad. If you feel threatened by my humor, I could consider limiting it. The quality of our conversation is certainly no fun and I'm not sure what's the reason behind that.

As about your argument on my humor, the AI sucks, yes. And that AI is coded by the same people who coded Widelands. Would you be surprised to know that some of the rest of their code sucks as well? Of course coding the AI is more difficult than coding flag animations, so don't be as quick to judge those people as you are with my ideas. But this is not my point. Actually, some other aspects of Widelands are remarkably well-coded. Something else is wrong.

The best models, those that don't suck, are usually AI-friendly models. Now read that inversely: If a model is not AI-friendly, chances are that it sucks. And the flags model sucks so badly, that it takes super-computers (like human brains) to find ways of applying it in small maps. By the way, chess is played in a small map. Do you know why the AI in chess can today win the best human players? If you answered "modern processing power", you are almost wrong. The primary reason is because programmers have finally managed to model chess' complexity in a very computer-friendly way, which remains strong even in low-end computers. Therefore, for the fact that Widelands is no fan to play against the AI, you may blame the AI, or you may blame the programmers of the AI, still the real guilty one is the model. A bad model can make it hard for the programmers to code a good AI for it, thus forcing them to code a bad AI. Programming is all about effectively modeling various problems, so as to give good solutions. Coding is needed only to express those models in a form understandable by computers.

Settlers was never about flags, it has always been about ...settlements (or empires). Flags were an ambitious effort to model the roads of those settlements. A model with some merits and also some weaknesses. One of its weaknesses is that it's not realistic. Another is that it's not AI-friendly. Another is that it's not always human-friendly, as at times it is just painful. And this pain is not because of its complexity, but because of its badness. The pain is not part of the game's design, nobody tried to insert it in the game, it is there only as a weakness of the model. The fact that there are people who find that weakness thrilling and enjoy the resulting pain, is nothing new, as experiencing pain is a whole category of challenge on its own (and there is no shortage of painful games). And yet Settlers is fun to play for many other more average people, cause its strengths overshadow its weaknesses. But Widelands can do better than that, if ever decides to.

king_of_nowhere wrote:

bad argument there. If simpler was better, then there would be only game, and it would be "press X to win". No needless complication to distract those without the time to master it.

People are here generally because they appreciate complexity. Remove complecity, and the game suck. Or, at least, you're catering to a different audience.

Another knee-jerk reaction... If you had any idea how real-life big transportation companies and similar ones deal with real-life roads, you would know that Widelands' complexity, the one you consider yourself a master of, is nothing more than a toy compared to the real-life road-arrangement challenges which professional experts in that field face, trying to save real fuel, real time, and real money, while at the same time respecting various regulations and not risking lawsuits. And the primary tool in their job is computer simulators. Widelands could take some clues from that field and become an overall better game. If you prefer for it to remain primarily a flag-arrangement toy, good with you. But don't deceive yourself that this is a pillar of complexity for particularly smart or time-investing people.

I don't remember Widelands or Settlers being promoted as games of challenging complexity. There are other more complicated games out there. But if you genuinely appreciate complexity, there is a much better game: Widelands' codebase. If you play with it (even its first level), you'll take a taste of why complexity-hungry people tend to become programmers. And since I happen to be a programmer myself, you are the one who addresses the wrong audience about complexity. By the way, Widelands has many other merits than its complexity in comparison to other games, and other people may be here for those merits. Some of those merits are served by complexity, some others are better served by simplicity. In any case, complexity is not on central stage. It can be a means, but it wasn't meant to be an end.

Moreover, complexity comes in more that one flavor. While some players may exhaust their energy in squeezing little advantages out of exploiting complicated edge cases in the weaknesses of the model, other players may leave the model in auto-pilot mode and focus instead in the so-called "emerging complexity" (an often abused term) of effectively combining various strategies in a length of time. Weak models tend to favor the former type of complexity, while strong models tend to favor the latter one.

king_of_nowhere wrote:

So, roads with flags stay the way they are. they are a big part of the game's flavor. Yes, of course this is nothing more than a matter of tastes, but if we change something so fundamental, we may as well play a different game.

Of course this is a matter of taste, but it's not the matter I had in my mind at all. I would be crazy to expect any fundamental change to take place within the next years, even if Widelands' development had a fast pace. And I cannot understand why anyone would consider and treat a single person like me as being a real danger of drastically changing the direction of an open-source project. But even if sometime the flags model was replaced by a better one, that wouldn't be a big deal nor a different game. Only some old players would have to invest some more time to master the new model. Some of them would even find that the new model actually fits better their taste, but only after tasting it for themselves. Their reactions to tiny changes is a proof of that.

My whole point in mentioning flags was to demonstrate how the superiority of the wallet model can get confirmed by its AI-friendliness, as well as by its potential in facilitating even fundamental changes. And thus to persuade some more people and hopefully implement a solution for the problems you wrote yourself in the opening post. And so to satisfy your taste as well.

That was a long post, and long posts have more chances of containing wrong or bad things. But can you find anything positive in it? This is the complicated challenge. Can you find any interesting idea in it? Any insight on what makes a game successful and a successful game even more successful? Anything with the potential of applying to Widelands? Anything that you could borrow to form it into an idea of your own? Or are we wasting each-other's time here? There will always be differences among us, including big ones. But there will never be any benefit in focusing on them, except if honestly trying to bridge them or combine them into something better. Let's try making this process more positive for everyone involved.

Top

Quote

teppo wrote:

Long time ago, it was unsurprising to see that argument here.

I'm that special case of player who is old to Settlers but new to Widelands. I have expectedly lost some episodes. Still the frequency of an empty argument doesn't make it any less empty.

teppo wrote:

The correct way to read the sentence is that changing the foundations requires a strong case.

Whatever the foundations may be, I'm not targeting them. I'm merely demonstrating the power of the wallet model.

teppo wrote:

I asked for concrete examples, got none. Lay out those better ideas. Nagging instead is antonym of victory.

Actually you made fun of my idea, so I followed along. Don't hold your breath for the things you ask for. Unless you realize that we are on the same camp, fighting is a fruitless thing.

Top

Quote

Another thing that we'll need to consider coding-wise is that flags aren't only used to store wares and mark junctions. You can also use them to define the layout for your road, if you click bit by bit rather than using the auto-layout. The new model would also need a replacement for that function.

Some of the resistance comes only not from player's point of view, but also because not having flags will be a big engine change, and there are other engine changes that we' like to have first. So, overthrowing the flags model will not happen any time soon in any case, unless you care enough about it to implement it yourself.

I think we should keep the wallet model discussion separate from the flag discussion, because the two don't depend on each other - we could keep the flags and have a wallet model at the same time, for example.

The wallet model does sound interesting, but I don't think that we have anybody available at the moment to implement such a thing. So, it will take a while. Just to give you an idea of what our current contributors' focus is at:

• SirVer - Market & Trading, Graphics backend
• Tibor - AI
• Notabilis - Networking
• Hessenfarmer - Empire campaigns
• Nordfriese - Frisian tribe & campaigns
• Kaputtnik, trimardio - Website
• hjd - Build system
• Tino - Windows packaging
• Myself - internationalization, UI, code cleanups project management

Plus some more people who help with testing and occasional bits and bobs, like teppo and Hasi50.

As you can see, no economy/road system expert on the team at the moment. SirVer would be capable, but he doesn't have any time these days

Busy indexing nil values

Top

Quote