Current state of merchant agents?

[Mark_Everson]

Well, I think that specific idea would take up too much memory. Along the silk road, say, you'd probably have scores of merchants. And polling all the merchants every time a unit moves would indeed eat too many clocks. I'm sure we can come up with a reasonable compromise. Each 30-sq area of the map would keep track of which merchants going through it or something to reduce the combinatorial or search explosion.

BTW Java just isn't all That slow. Exectution speed is probably within a factor of two of C or C++ for most real tasks. Graphics is what eats up Most of the time in cutting edge games, and we will not be having marching units or other BS like that

[roquijad]

Mark:
I'm not sure what idea you say is too expensive in memory... storing merchants routs in map squares?

[roquijad]

What about considering cities in-the-route (cities where the cargo passes through in its way to the destination city)? I think cities (or provinces if you prefer) of this type can get a benefit from this passing routes (services paid by the merchants to people at the city). I guess a lot of cities had flourished in this way, specially those with ports receiving cargo from abroad and being the gate to inland.
What do you think?

Rodrigo

[Mark_Everson]

Yes. The merchants clearly need to know their routes. But giving Every square a pointer to Every merchant that goes through it could be substantial. Maybe its also manageable, depends on the number of merchants. I am most worried about modern times when they will go all over the place! But we'll figure out a decent way to do it no problem.

[shimmin]

I'm a little bit confused.

Every square doesn't need a references to every merchant. Every square just needs a list of references to those merchants that use it. For most squares, that list will be empty.

[Mark_Everson]

Shimmin:

That's what I said. A square would need a pointer to every merchant that goes through it.

Quick estimate... Average trade path length = 50 squares (320x200 map though it might change) 1000 merchants in the late game.
Giving each square a poiter requires 50K pointers. That's a lot of memory for something that isn't core to the game.

[Lord God Jinnai]

What do u do about modern times when u merchants can go everywhere just about and u also have coorperations to worry about (any the should be put in there someway of interaction, thats for the econ model).

[alms66]

In reviewing the Merchant Model, I have come up with a proposal to modify the Merchant agents themselves in a small but meaningful way. By giving the merchant agents a lifetime similar to characters, adding different kinds of merchant agents and a maximum number of trade routes, we can diversify the trade market within Clash. The numbers in the chart below need to be revised for the game. I just put some numbers in, Mark or Shimmin (if he’s still with us) can probably better fill in the numbers.

Merchant Chart

“Unit” /Max#Routes /MaxCash /Tech /Lifetime /Cost
Merchant /Base /Base /Currency /Base /Base
Trading Comp. /2x /3x /Banking /1.5x /3x
Corp. Merchant /4x /6x /Corporation /2x /5x

[Stuff2]

I think that one merchant should be able to trade many different goods at the same time.
If there is a trade-route between two cities/regions it should involve all goods that are profitable in the same route. I think this would save some memory too. Instead of having 25 'terrain improvements'as trade routes u have one trade route which is belonging to xxx (merchant number 3 from venedig).

[roquijad]

What ever happened to this thread?
I remember it was a fun one!
Is shimmin still with us?

[Mark_Everson]

TK:

I see what you're saying, but at least as I'd envisioned it the player wouldn't know anything about individual merchants/corporations anyway... Only when they got to the point of being represented by a character would they be more noticeable than overall trade. However, what you suggest would be easy to do if it turns out there is some real advantage to it...

Stuff2:

A merchant/Corp. can certainly have more than one route. What you suggest seems kind of artificial in that a second good from a source city may well be more in need elsewhere. And so why should it go to a particular city, as opposed to the merchant who bids highest for it regardless of the city of origin of the merchant? I don't understand what you mean about ' terrain improvements ' as trade routes either... I think merchants will really work best if each merchant may have several routes, and each route stands on its own.

Rodrigo:

I wrote shimmin several months ago, and he was too busy to do anything further. I assume that's still the case... Merchants won't be implemented until at least demo 6 anyway, so we can keep discussing things at our leisure for the moment.

[shimmin]

Mark's right about my busy-ness. Actually, I just completed a senior design project yesterday, and look forward to having time to contribute again starting next week or so.

Which threads would be most fruitful for me to read to catch up on the last six months of development?

--RS

[Mark_Everson]

Hi shimmin:

Good to have you back! Actually, I don't think anything much that was done previously has been invalidated. So other than this thread, the best place to look is the main economics thread. Other than that we have been working on the demo 5 specifications for several areas... but since we weren't thinking we would get to merchants in demo 5 there isn't much there you need to be aware of I think. The demo 5 econ thread probably would be a good thing for you to look over anyway, but as I said I don't think it will be essential.

[Mark_Everson]

Well, obviously shimmin is long gone...

But the good news is I have a crude version of the merchants up, and they should make it into Demo 5.1 unless I screw up majorly.

The merchants I have now make point-to-point trades. They are designed to move goods mostly between cities that have sources of different Specials. It also occurred to me that in addition to the long-range merchants, there maybe should be local ones. The local ones would just try to reasonably distribute goods that are available somewhere within a province throughout the province as practical. Between the two types I think we can simulate much of the trade benefits that happen in real societies.

Any opinions?

[Yoav Sissman]

I thought that all the goods produced in a province will be available in any part of it, so local marchents won't be needed.

Are you saying that production and demand will be handled in the map square level? That seems overly complicated.