Current state of merchant agents?

[shimmin]

Looking at the web page, I see the concept of computer-controlled merchants that handle much of the trade-route / commerce type stuff.

This is a neat idea.

To what extent has this concept been fleshed? At the moment, it appears you know what you want them to do but not how they are going to accomplish that. Is this an accurate assessment, and if it is, is this an area where ideas would be useful?

[Mark_Everson]

Hi shimmin:

Discussions and thoughts toward implementation would indeed be useful. Merchants are not far beyond what's already in the documentation. I don't know exactly what you've seen... but a search on "merchant" here at the forum should give you it all. (look in the 'Best Way...Suggestions Implemented' thread for how). Why don't you start with your thoughts right here, and we can link it to one of the econ threads.

Beyond whats in the docs the only idea I've had lately is that merchants might have contracts with cities that are entered into the turn before. I thought this concept would smooth things out a bit. Anyway, fire away, and I'll respond when I can. Thoughts on how they'd interface with the general econ framework would be valuable too.

Mark

[This message has been edited by Mark_Everson (edited October 13, 1999).]

[Yakopepper]

Im sorry for asking this but Ive gotten tired of looking, but where is the thread on the merchant agent? Ive got a few ideas but I perfer not to repeat anything and look foolish doing it?

Sir Mark Everson, one other thing, I havent heard anything about Wonders. Is there a reason? Or am I just blind?

Daniel
[This message has been edited by Yakopepper (edited October 14, 1999).]
[This message has been edited by Yakopepper (edited October 14, 1999).]

[Mark_Everson]

Daniel:

This Is the merchant thread . Pretty sad ain't it? The discussion on merchants that exists is in the Miscellaneous section of the Economic Model. If you open the link and scroll down a bit, you should get what little I have written up.

Wonders are planned to go in at least for scenario design. Probably real-world type wonders will be in the standard version of Clash also. I personally don't like silly ones like Leo's Workshop. If its a big issue for you and you want to push your case or make suggestions, just start a thread called wonders. Doing a search on "wonder" will probably net you all the little snippets of conversation on the topic that we've had from time to time.

[This message has been edited by Mark_Everson (edited December 07, 1999).]

[Yakopepper]

Thanks Sir Mark Everson.

Wondrers arent really a big issue for me, but for the sake of realism, they are essential. At least I mean the real 7 Wonders of the world. Not the "Sid Meier Wonders."

Since this is the Merchant thread, Ill stick to it. It's nice to not have to worry about trade, even though it extremely important in history.

Id like there to be a way to do what the nazis did to the allies in WWII. Try to isolate Britain by cutting off supplies via sinking merchant ships. If done during gameplay(if possible) would be an act of war and done only during war. Since you're the brains behind the game, maybe you could figure out howd this work.

I dont know how much of the trade realm the computer will control, but it should be optionized. Im just hoping the AI will be good enough to trust. Which, BTW, I havnt found yet in a game.

Thanks
Daniel

[Kull]

One other feature of the merchants/trade feature are links to intelligence gathering and tariffs on trade. The Diplomacy model contains some general thoughts on this, but the details are still sadly lacking (looking at self in mirror!)
http://apolyton.net/forums/Forum21/HTML/000106.html

[shimmin]

I am drawing a merchant-agent proposad based on the following guidelines. Expect to see it sometime this weekend.

1) Since every is computer-controlled and therefore eats up clock cycles, what they do should be simple.
2) This goes doubly so, since for the most part, the player does not directly see their mechanics, only their effects.
3) The vast majority of merchants will be computer controlled (player influenceable, but not controllable). If an urgent need arises, however, players should be able to foot a mildly hefty bill to create merchants they can control directly.
4) Despite being simple, the merchant model should be general enough to accomodate features such as:

trade routes and triangular trade
effects of technology and infrastructure
gov't influence -- taxes, tariffs, blockades, embargoes, bans
criminal elements -- piracy, smuggling, black markets

5) The merchants should also have a general enough to be used as quartermasters, distributors, and generally any activity that involves getting goods from where they are to where they should be.

These are the values I'm trying to incorporate, and what I gathered from the web page and the overall philosophy I'm getting from the Clash staff. Is there a task the merchants need to perform that I'm missing? Something you feel is unnecessary?

[Mark_Everson]

shimmin:

Looks Great. I can't think of anything really important to add. The only minor role you missed is merchants as bankers, esp lenders to governments before modern times.

Please just post the first draft when you have one, so that if I've not mentioned something important that we catch it b4 you waste a lot of effort.

Lookin' Good,

Mark

[Lord God Jinnai]

I know this was discussed earlier but merchants should be carriers of diseases and also could sell information on other countries.

Also there should be a type of goods production where certain places produce certain items and may need others (import and export)

[Mark_Everson]

LGJ:

You need to read the economy model . The rare goods you talk about that involve merchants are called specials. We'll probably have about 20 of them varying from tin in ancient times to oil, aluminum, and uranium in modern times.

[Lord God Jinnai]

I meant more for luxuries and finished goods. Though those are important also.

[Mark_Everson]

There are luxuries and finished goods among them. The list in detail is TBD. But I think there are interesting tie-ins between the Tech model, finished-goods specials, and merchants that we should look into. For example the first civ in an area that gets to metallurgy 15% might result in one province specializing in metalwares. In game terms That province would automatically get the ability to produce the Special metalwares. The merchants based there would now obtain an advantage because their home base has something desirable to export.

This connection would benefit the merchant model, by making the player able to deliver to his merchants a competitive advantage. It would also benefit the Tech model since there would be some interest in achieving lots of little goals connected to specials. You'd get a message something like: "because of our technological progress, Paris has developed as a new center for producing metalwares".

What do you guys think?

[Yakopepper]

I dont think it should always be the first civ to develop metallurgy should earn a providence specializing in metalwares. It doesnt follow history. Maybe have each of the first 3 civs HAVE a chance to develope the specialty, but only one would come out with it.

Daniel

[shimmin]

A basic merchant model.

At the broadest level, merchants have four attributes.
1. An owner
2. A cultural profile
3. An amount of cash
4. A list of trade routes

The "owner" is simply whoever gets the merchant's profits (or absorbs their losses). This will almost always be the entity that created the merchant in the first place.

The cultural profile determines the merchant's stomach for risk and creativity, as well as their likelihood of aiding tech diffusion.

The cash is the merchant's liquid assets.

The trade routes are the most complex part. They are how the merchant makes money.

I. Merchant creation
Purchasing goods in one locale, paying for their transportation over long distances, and then hoping they can be sold at a profit requires a fair amount of capital. In short, someone must be able to gather money together and be willing to take a risk with it.

The people of a province can take the initiative to build a merchant just like they could build a schoolhouse or improve their agricultural capacity. In this case, the province's populace is effectively the owner of the merchant, and the merchant has that province's cultural profile The effectiveness of these merchants is based on the province's PCI and technology. Using the current proposed tech tree:
1. Currency: trade becomes possible. Wealthy individuals may risk their own assets. Populace can build merchants; these start out with half the province's PCI in cash (times some multiple).
2. Banking: the idea of loaning money. People willing to take risk obtain loans to start merchanting. Poulace-built merchants start with the province's PCI in cash (times the multiple).
3. Corporation: large numbers of small investments can be focused into colossal undertakings. Populace-bulit merchants start with twixe the PCI in cash (again, times the multiple).

Player-built merchants:
The player can foot the bill for merchant creation, with whatever penalties are normally assessed on the purchase price for player rather than populace building of infrastructure. The player's treasury is now the owner of the merchant. These merchants start with no cash. Instead, the player directly transfers money from the treasury to the new merchant for starting capital.

The cultural profile for player-created merchants is that of whatever province the player creates the merchant in.

(Under economic systems that completely ban private enterprise, this will be the only way to make new merchants. Such merchants still don't require extensive player control after the initial creation. In contrast to populace-created merchants, however, they will obey direct orders from the player.)

II. Trade routes

Each time the agent gets to think, it identifies a special available in great excess in its home province, or one not available but useful in its home province. If it fails to find one here, it considers other provinces in its civ before giving up.

Once a province and a good are chosen, it generates a list of provinces (based on the knowlege of its home civ) where the good is expensive (if dealing with a special found in the home province) or available (if demanded in the home province). It will set up a trial trade route path to as many of these as is clock-feasible. (Two or three will probably be OK, if you use something like A*. A* is especially nice since it can automatically empoy differing cost functions for each step.) The cost functions will depend at the very least on terrain and transportation technologies, but I really don't want to draw up a list until you decide whether you like this method or want something more abstract that doesn't use specific trade paths. This method gives the trade route a specific path, which gives the player huge numbers of ways to affect trade without controlling it, such as:
a) security -- when traveling far from garrisoned cities, or through civs that have poor law and order, the merchant must hire an armed escort to protect the goods to protect from highwaymen, pirates, etc.
b) infrastructure -- no nonsense of roads producing trade. Here, roads are good b/c merchants use them (they decrease their cost function), and so it is less expensive for them to get specials in and out of your cities.
c) law -- charge a tarriff, and you get a cut of the trade going through your civ into your treasury, increasing the merchant's cost function. Of course, merchants will try to go around you, but sometimes, they can't. On the other hand, you could subsidize a trade, or more severely, ban it. (This won't make the trade impossible, it will just add a heft cost function to carrying it through your civ, increasing with the effectiveness of your law enforcement and decreasing with the corruption of your officials.)
d. warfare -- set up a siege or blockade, and the merchants' cost function goes up quite a bit due to the need for smuggling it through.

After finding a path, the merchant calculates two numbers
1) the unit startup cost of the route -- based on the type(s) of transport (physical and social) needed to be paid as one-time costs.
2) the unit proifitability of the route (price difference between the supplier and customer provinces, plus ongoing security wages and bribes), divided by the amount of time the trip take. If there exists a commodity that can be traded in the return direction, its price difference will be added to the forwards price difference in this calculation.

This would be the point to try to find a triangular trade route, if no return commodity exists and clock cycles permit.

Based on the start-up costs, the profitablity (essentially a pay-back rate), and its cultural profile, the merchant picks a trade route (if any) to buy into. It pays the startup costs (times 1.5 or so -- see below), and thereafter, gets the pay-back rate paid into its account. I don't know how often you want to reinspect the path to take advantage of new infrastructure or check for interruption by warfare. Changes in commodity prices should be checked every turn simply because that pays into the merchant's account and because it's easy.

III. Merchants and their bank accounts

The merchant's cultural preferences dictate what they do with the money that accumulates in their account -- how much they reinvest and how much goes to the owner. Reinvestment can take one of two forms. Either they can expand an existing trade (by paying more startup cost to increase the capacity of their route, increasing their profit by a similar factor). Small increases should have some discount over large ones. Let's say that if the merchant wants to increase a route's capacity by more than 50% in one turn they pay the same 1.5 penalty they did on route startup. Otherwise, they can save up money, and then scout out a second route to invest in.

IV. What's missing
This is rough right now. The trade route thing is the biggest question I have right now -- do you want the additional flavor that comes with the trade routes have set paths, or do you want abstract routes? Quartermasters are an easy modification -- rather than deal in specials, they deal in food and resources, and their "home province" is an army (also, they _must_ find a route -- their "cultural profile" essentially says pay as little as possible, but pay what you have to.)

Essentially, my question at this point is: considering that this is largely a computer-controlled activity, is the flavor that comes from trade routes having exact paths worth the complexity?

[Mark_Everson]

Hi shimmin:

Looks Reeel good to me.

I personally like using an actual route rather than a more abstract system. Its especially important during warfare when supplies of strategic specials might be needed by one side. Since we already have the Map AI required for the military stuff, IMO the extra clocks needed for the merchants shouldn't be too bad. Also one can always change the "hunting time interval" for new routes if its a problem.

One other possibility for the tech tie-in is to have a base tech Commerce (LGJ are you listening in?) and have a merchant's efficiency track that. That'd make it more fluid than the big jumps you have.

Great Work,

Mark

[Lord God Jinnai]

I was using the slightly broader term economics to cover this. Like mechanics and architecture commerce is simply and branch of this that could be considered too detailed. Also the tech "Theory of Economics" should also increase the capasity of merchants alot and allow for more than triangular trade routes if their's a lot of profit to be made that couldn't otherwise.

[shimmin]

I had a list of possible modes of transport and a few features thereof in the reply box, but the browser froze when I hit submit.

Anyway, rather than reproduce the list, let me just report the two main aspects in which I'm not entirely satisfied with it.

1) At several points, it features "quantum leaps" in transport ability at certain technologies, some of which really were historical. For the most part, I'd like to use the basic techs to make some of the progress more gradual. Problem is, from reading the tech threads, I can't get any consensus about how these work, or even what they are! I see a percentage system that several weren't entirely satisfied with, but I also see a few alternate ideas that met with some approval, but nothing definite. Have any decisions been made here?

2) Today, "merchants" have become more of a shipping industry than the ones who actually take the risks of trade upon themselves -- they're hired by stuff's owner simply to move it rather than take any interest in how the stuff sells once it gets there. Is this something we want to address?

[Mark_Everson]

1) Remember the Leaps look immediate looking back, but before you can get that railroad bonus you need to build a Lot of infrastructure. I would prefer to have the merchants use the best infra available when they're pathfinding, and then have mods, as you suggest, for Commerce level and Land or Sea Transport level. I'm not sure I've talked LGJ into these, but you can go to the tech 2 thread and put in a bid for them if you agree with me . As to percents vs levels etc. why don't you figure what you really need and then post your opinion on tech 2... IMO what you're doing can work with either case anyway with only small modification.

2) Personally I don't really want to address that distinction, but since you're writing it up, propose whatever you think is best and then we'll talk it over if there's disagreement...

Cya,

Mark

[Lord God Jinnai]

Merchants didn't travel to unknown regions, with a few exceptions (Marco Polo and Christopher Columbus are 2 famous ones). In any event these people had reasons for taking the risks, profit from a direct source instead of going through the middleman (arabs or venitians example). This means as things such as silk pass from place to place it becomes more expensive and the profit is less.

Also on a related note there should be about 50 common items throughout time periods and about 20 rare items that change through time and may become common in later periods or obsolete. This would greatly enhance the merchant aspect since a country producing silk in medivial times could make a fortune, but in modern times, though they could still profit, it would be far less.

[shimmin]

Another frozen browser. Yech.

Anyway, here's my "transportation through the ages" list. I've based it on LGJ's tech tree, although in browsing the tech tree to make this list, I have a couple comments that I'll put in the tech thread.

General format:
Transport (required tech). Speed (arbitrary) / Capacity (arbitrary) / Range (squares between resupplies) (% speed lost if moving beyond range -- foraging penalty). Terrain restrictions.

General notes: if a foraging penalty isn't listed, the transport can't forage, and so cannot move beyond its range. Many sea / air units don't have terrain restrictions per se, but do require certain infrastructure in the provinces they service.

LAND
Porter (none). 1/1/3 (20%). Any land.
Pack animal (Domestication). 1/8/3 (50%). Any land.
Courier (Animal riding). 2/2/4 (50%). Any land.
Cart (Animal-drawn carts). 1/20/4 (50%). Clear land or roads.
Wagon (Wagons). 1/100/5 (50%). Clear land or roads.
Steam rail (Railroad). 10/1000/10. Rails.
Deisel rail (Combustion engine). 16/2000/20. Rails.
Truck (Automobile). 16/750/10. Roads.
Solar truck (Solar power). 10/750/none. Roads.
Solar rail (Solar power). 10/2000/none. Rails.
High-speed rail (Superconductor). 70/1000/15. Rails.

SEA
Raft (Rafts). 2/20/none. Rivers, downstream.
Riverboat (Riverboats). 2/100/none. Rivers.
Galley (Seafaring). 2/500/6 (70%). Rivers, coastal waters.
Cog (Sails). 2/2000/10 (70%). River, ocean.
Caravel (Navigation). 3/3000/80 (70%). Ocean.
Galleon (Clocks). 4/10.000/120 (70%). Ocean, needs port facilities.
Packet ship (Factory production). 5/40.000/180 (70%). Ocean, needs port facilities.
Clipper (Factory production). 7/20.000/180 (70%). Ocean, needs port facilities.
Steamer (Steamships). 6/200.000/180. Ocean, needs port facilities.
Freighter (Steel ships) 10/1.000.000/250. Ocean, needs port facilities.
Hydrofoil (Jets). 16/20.000/10. Ocean.
Nuclear freighter (Fission power). 12/1.000.000/none. Ocean, needs port facilities.
Solar freighter (Solar power). 8/1.000.000/none. Ocean, needs port facilities.

AIR
Dirigible (Pressure containment). 16/3000/200. Needs landing facilities.
Air transport (Flight). 70/100/20. Needs landing facilities.
Jet transport (Jets). 180/5000/120. Needs landing facilities.
Solar airship (Solar power). 12/3000/none. Needs landing facilities.
Hypersonic transport (Space shuttle). 2000/3000/300. Needs landing facilities.

For these, I was thinking about aircraft more or less equivalent to the Zeppelin / DC-3 / Boeing 747 / (no example) / NASP (a temporarily suspended NASA project, designed to do New York to Tokyo in 30 minutes, flights leaving twice daily), respectively.

[Mark_Everson]

LGJ:

Agree about merchants generally not being explorers. Where was that said?

50 common / 20 rare strikes me as too many. At that number you're starting to simulate a Real economy. I think it will take too many clock cycles that the player will derive no benfit from. However, we can just try out your idea in playtesting and see.

shimmin:

Great list! I haven't looked it over in detail, but it looks quite good on cursory inspection. I assume the quality of the infrastructure should affect the speed and/or range as the model gets more developed... and the basic tech level involved when we figure that out ;-) Also culture should be in there in some way. FE when moving through a very corrupt country you'll need to pay every local bureaucrat...

This is gonna be a fun model!

[alms66]

Would it be possible to actually show the merchant walk/sail/fly along the trade route(without actually showing the route on the map, too ugly!) This would allow pirating by other civs(and possibly start a war, although the "pirate" should have an option to stay concealed, maybe?)
Also if we did this merchants could be the "suppliers" to an army, if the merchant was destroyed so too would be the supply line!!! Can we say "Starvation?"

[rentgen]

One comment about pirating. I was really annoyed by this feature in Civ:CTP because you had to assign the routes over and over again.

I think it would be useful to provide possibility to accompany the merchants with military units. In fact, this is the only way it works in war times! Nobody sends suppliances on their own they are always under protection of some military units (convoy ships etc)

-- Anton

[Mark_Everson]

Toubabo_Koomi:

On showing merchants walking on the map: I don't think we want to do that. It would be Really busy. There are going to be merchants going all over the place. I don't know what shimmin thinks, but my general take is that we would allow the player to call up a map with merchant routes marked on it, but generally merchant routes wouldn't show at all. I had envisioned not actually moving the merchants square-by-square. They would just calculate their best route, then be assumed to be somewhere on it. Remember, they're not individual merchants, more like guilds or trading companies. So they wouldn't necessarily have a specific location on the map anyway.

Great idea on using the merchants as quartermasters for the army. In fact it's such a great idea, we've already thought of it

Piracy will certainly be possible. Increasing intelligence about the exact route, and the merchant's timing (figured abstractly) will make success in piracy more likely.


rentgen:

So in CTP you need to re-assign merchant routes every time they are attacked? Ugh. In Clash, the merchants will figure their own routes. This will include what they know about hostile activities going from war, to piracy, to corrupt bureaucrats that demand large bribes. If the player owns the merchant, I guess we would allow the player to specify a general route. But hopefully that won't be necessary except for micromanagement freaks...

We hadn't gotten that far, but you are right, you need to be able to escort supplies. I don't think it will be a big problem to put that in.

[roquijad]

This thread looks like forgotten. Will anybody see the following post?

About blockades and piracy:
Suppose you have a merchant trading from city A to B. You have 2 options:
a) The route from A to B was defined and stored in memory when the trade was started by the merchant finding the cheapest route and avoiding known enemy presence.
b) The route is computed looking for the cheapest route and avoiding enemy presence every N game turns.

Every turn of play the computer checks enemy presence in every map square that the route uses. (If the route was just created, no enemies will be there since the route was in such way created. But, as game advances, troops may move into map squares used by the route) If enemy is there, a probability is calculated for cargo to be intecepted mod by enemy tech and strenght. If cargo is intercepted, merchant loses some $ and re-evaluates trade between A and B considering new routes (in fact merchant does what he did in the first place). If danger is high enough, merchant will prefer trade between some other cities.
Now consider you control the above enemy civ. The question is how you make your troops intercept the civ cargos. You should be able to ask the computer to tell you what's the best route form A to B using a dialogue box. The computer shows you this on the map for a few seconds and then you send your troops there. Notice that this best route is not necessarily the one used, for computer uses your map knowledge and not you're enemy's which was used to create the real route. Maybe, given some inteligence reports, you could be able to know the precise routes.
Pirates may do the same as you do trying to intercept routes.
A civ can protect a route by two ways. It moves their troops to destroy enemy presence or it can assign troops to the route, in which case the troops no longer stays in a particular map square nor can be used normaly as any other task force. The probability of enemies intercepting the cargo should be mod by this guard.
The biggest problem I foresee here is that the computer has to check for enemy presence every game turn which is expensive.

A comment on options a) and b) above:
Using a) routes can't adjust according to terrain development and the merchant uses always the same route regardless of tech advance. But it saves a lot of computer resorces since you only have to store the trading cities.
Using b) can make merchant sensitive to road/railroad/etc developments, but it uses more computer resources since best routes need to be found, for every existent route, every N turns.

Rodrigo

[Lord God Jinnai]

This has some good ideas, though haven't read too much, just skimmed. I perfer option b. It shouldn't take up too much more since new merchants will continuously be created and old ones die without heirs for a dynasty or aren't part of orgainiszation. I'd like to have further discussion put int the Character/Dynasty topic (If its alright with Mark) since merchants and pirates will be taken up via me if that's alright since it will make coding much easier imo.

[shimmin]

Rodrigo

Regarding computational expense of checking for enemy presence every turn.

There's a cheaper way of doing it. Consider the trade route like just another terrain feature. When units move, they have to check the terrain to see if it has any effects on their movement (a land unit can't move across the ocean, roads speed movement, etc.). The reverse (and equally inexpensive) process is checking the terrain to see if the unit has any effect on it.

Enemy presence checks need only be made when military force moves into a previously unoccupied square.

[Mark_Everson]

Rodrigo:

Merchants will regularly examine potential new trade routes, and better routes for existing trades.

LGJ:

IMO merchants belong as a sub-class of the economic model. Besides, Shimmin is working in this area, he just had a busy patch of real life. Pirates would be a toss-up between Military and Characters. If you want to make specific proposals about Pirates, be my guest .

Shimmin:

Good to see you again

[roquijad]

shimmin: so that means you have to store in a terrain element the routes passing trhough it?... A military unit moves into an unocuppied square, the computer see a list of routes passing through it (previously recorded when the merchat first started the trade) and then calcs a % of intercepting some of them.... is that what you mean? If it is, I think is fine if having stored that info attached to map squares isn't expensive...

Rodrigo

[shimmin]

Rodrigo

That's what I had in mind. I'm not an expert in coding, but I suspect that when dealing with the inherent slowness of Java (an especially considering the quality of AI people are envisioning), using more memory to gain speed will often be a trade worth making.

Paul, other map people -- would the map's code permit the storing of trade routes as some sort of terrain enhancement? (Could trade routes inherit much of the same data structure of roads or somesuch?)