Beiträge von Oakville

I don't understand how fewer players should cause prices to rise. If the demand is based on the number of players, wouldn't prices drop as the fraction of active players rises? If demand is static, wouldn't they drop when the number of players was large? In neither case does a small number of players cause the prices to rise. Maybe the game is doing something weird that I don't know about?

I can tell you what happened. After about the first month, I started doing
a significant amount of QL research. I don't know if you've noticed, but QL research is not
counted in the ROI score, but it IS counted in the company value for
lowering your ROI score. Just why this is deemed a desirable
mechanic, I can't explain. Punishing new players who don't have a
chance to win anyhow for doing the research they so desperately need
in order to have any chance just seems to me to be heaping insult on
top of injury. For the past week or so, I've been making about 7M/WU
with about 7M/WU of research and an ROI of near zero. After about 6
weeks, when Auto tree prices were well into collapse, I looked around
and saw 100 QL goods selling for 3-4x what I could get for mine, and
decided to toss in the towel and focus exclusively on research and
the Manufacturer score. It really isn't worth even trying for the
overall score for about a year or so, until you can build up your QL
levels, and I intend to make no attempt whatsoever at any of the
medals in the next round.

I would expect that people feel
strongly about keeping their QL levels, and I don't think it's really
necessary to get rid of them, just so long as there is some change so
that other players have a chance. What I would do is make research
for the current round be 1 RL/1WU, with only the 500k/building
level/WU cost. Buildings stay working on the same thing until
changed or finished. The starting research level is based on the
total research spent on the good, and does not change from the
current formula of Cost = 8x10^3x(QL)^3/3+4x10^6*(QL)*Lab Building
Level, so most of the RL levels gained in the current round will be
lost at the game reset.

I would also make the desired QL for
factories selectable, multiply production cost by (1+desired QL/100),
and have demand be 1 consumer at each desired QL who chooses the good
with the cheapest value of an effective price of price (for
QL>=Desired QL) or price + normal price * 4 * ((Desired
QL-QL)/100) (for QLto normal price/effective price for a price below the normal price,
and (normal price/effective price) ^ 0.5 for a price above it. This
turns the job of selecting a QL to sell into a rock/paper/scissors
game, since if everyone sells at the highest QL, they're throwing
money away on production costs and getting nothing for it.

Re:#1,#10: I'm not completely decided
on this, but I tend to think that the challenge of finding a new area
where you can still make profits after the auto tree fully saturates
would make the game more interesting, and that a 3-month game is a
good idea. The problem currently is that even when the auto tree is
saturated so that the QL 0 people make nothing, the QL 100 people are
still pulling in money hand over fist , so the auto tree never
saturates for them, making the same thing work every game. This
would no longer be the case after a major change to the QL system. I
do think it's good that the price for some goods drops to the point
that it forces people to move into a different tree.

Re#7: Without a change to the QL
situation, this enhances the current reliance on QLs to win.

Re#6: If they're limited to 20, I'll
just have bigger offers. In any case, I think having the market
become saturated is an intended game mechanic. The game would feel
much different if the price actually adjusted fast enough to balance
the market. I've already posted a bit about the January rush which I
think is undesirable.

Re#8: As I understand it, the way to
get a high vendor score under the current system is to have one store
of each type, plus a ton of auto stores, so the change hasn't really fixed
the problem it was intended to solve. I'm actually kind of
interested to see what would happen under the scoring system of the
previous round, plus quick QL gains for the current round only.
Everyone would pile into autos, and with the top players no longer
insulated from price fluctuations, it would turn into a red-ink
bloodbath, and then they'd have to move a lot of production out in
order to actually make money.

Numbering by the proposed change leaves the impression that our suggestions are rather scattershot and bouncing all over the place. Numbering them by what I feel are the most important problems to address does, I think, already show some widespread agreement. As I see it, the most important problems to address are:

#1: Assuming that people play well, only way to win the game is to have 100 quality in the auto tree and to go heavily into autos, to deliberately suppress one's earnings in the midgame, and to prefrom only minimal quality research since, for some inexplicable reason, money spent on quality research does not appear to count towards ROI. New players have no chance for their first year or so. This is boring and needs to be changed. Suggestions to correct this problem include my suggestions #1,2,3,4, One-Stroke's suggestions #1,2,3,4 from the "No price changes thread" and Barmcake's suggestions #2,3,4,5,7.

2. It would be good to increase player interaction. Suggestions along this line include my suggestions #6,7,8,9, One-Stroke's 3rd and 5th suggestions from his list in the "Game Theory" thread, and Vana'Diel's suggestions from that thread.

3. How trade is handled when the price is fixed far from equilibrium needs to be changed. Suggestions for this include my suggestions #5,6, and Barmcake's suggestion #6.

It would be too handy. I agree that it would be incredibly useful, but, well, don't we need SOME difficulty in the game? I'd actually got the other way and make switching products incur 4 WU downtime, and then add a multiplicative economies of scale to production of (1+N*0.02) based on the number of factories producing the same thing. That way it would make some sense to specialize in one component, so that style of play would work too.
Factories don't actually cost that much. The ROI hit from tearing them down isn't so big if you only do a few per day. The real hit is the downtime of building them back up to 10.

5. The current January classified rush is not fun and it needs to disappear. When goods are scarce, the current game provides a huge bonus to the first person to get their classified in after the date becomes January. I'd allow players to set prices outside of the currently allowable price range. The prices actually paid would be moved until they were in the allowable range, but if the normal price moved in the meantime, it would be possible to still have them at the maximum price without entering a new classified or offer every January.

6. In order to facilitate player interaction, I would have trade prices move at about the same speed as now, but in the correct direction. Average trade prices is NOT a reliable indicator of which direction the price needs to move. In the previous round, I had cases where I had classified out for days at the maximum price, and the price wouldn't move. Then someone would come along and sell it at the maximum price, and I'd buy it up in about 30 minutes. The next year, the price would move DOWN.

All of the top players have learned that, if you want to buy a good which is underpriced, the way to do it is to sell 10 of it at the maximum possible price. Someone else will buy it and the price will move up, eventually encouraging new supply to move in. This is highly counterintuitive and is not useful to new players. Prices should move up if there are classifieds up at above the current price which sit unfilled for six hours or more. They should move down if there are offers below the current price which sit unfilled for six hours or more. That's it. The average trade price has nothing to do with where the market price should be.

7. Goods are reliably available only in 2 quality levels, QL 0 and QL 100. Due to QL rounding, it is not reasonably practical to regularly buy goods at any other quality level. In order to facilitate player interaction, I would round the QL down only when removing goods from the warehouse. Otherwise, the QL would be kept as a float.

8. In order to facilitate player interaction, I would drop the sale cost to a flat 3%.

9. In order to facilitate player interaction, I would make all goods required to produce a good have a RL at least 1 level lower, and if possible widely separated. Yes, this means that the only shop available at RL 0 would be the grocery.
To Dave:
Current year Normal Price = (Wiki listed Normal Price)*100%/(SuDe % market satisfaction)
For supply levels over the normal demand, the current demand function already closely models this, except that instead of using the SuDe % market satisfaction, which is really the product stock level, it uses the actual market satisfaction, which the game refuses to divulge to us. This is why the auto tree revenue has been nearly constant. The problem with the dominance of the auto tree is merely that it's priced so high that even at these reduced prices, it's usually still the most profitable tree, and also that there aren't that many people at 100 QL, so they see less competition.

For supply levels below the normal demand, this is an awful function, because it allows the first person into a tree to capture the entire available revenue of the tree by merely selling one of it, which is now worth 100x as much. (Wiki listed Normal Price)*100%/(real % market satisfaction)^0.5 would work nicely, though.

I suppose I could get behind the idea of an economy with 100 consumers with QL 0 and 1 each at each other QL value X with a hard limit that they don't buy below their quality below their QL value and a demand of 100/((price/(1+X/100))^2. It would give returns outsized compared to the number of factories, so people would do it, but the returns would still be small compared to the total available revenue at saturated supply, so it wouldn't unbalance the game

This isn't the right normalization function. Since the production costs are the same as the manufacturing score, this function would leave autos as still by far the most profitable industry, albeit by smaller margin. Assuming that you want to have profits be the same for each product, the function you would want is something like:

Price = Price of components + production cost per unit + C * F(RL) / Production rate .

The constant C is something in the range of 100k for the more profitable products, and F(RL) should be something like 1+RL/50 or 1+RL/25. With your factor of 2.5/2 = 1.25 there would be very little reason to research the higher level products, and nobody would do it unless they were going for score instead of just trying to make money and do product research.

With no change to the current demand function, your price function would lead people to spread out over the available products, which would all sell for near maximum price, ending most player interaction. I do think something of a partial move in this direction would be an excellent idea, but truly equalizing prices across goods would get rid of most of the interesting parts of the game, such as deciding which goods in the store to produce and which to leave alone.

The current demand function doesn't feel quite right to me either. Sometimes I can't compete with the other players... literally. Looking at City Cars, for example, I can sell 20 per turn, at 253,000. If I drop the price, at 135,000 I can sell 40... at no change in revenue. Or I can sell 50 for 100k... at no change in revenue. But these aren't coming out of other players' sales. The total sales in the SDE graph goes up by this amount. So far as I can tell, I'm selling cars into my own little world, with no interaction with other players. Shouldn't it feel like, well, like I'm competing with the other players? The fact that the demand function is an undocumented black box doesn't help, but I've been pinging it long enought that I have some idea of what it's like by now.

It feels clunky, too. When I'm the only one selling an item, I usually have one store which is selling a few items less, often selling at half of what the others are. This appears to be to due rounding. And it's always the same store. When the prices of my stores are all the same, it's always the same store always gets its turn to sell last, which smells of sloppy programming. I can understand that Stdok wants to have an method which is deterministic, computationally quick and also has no dependence on previous events, but, still...

It's not as bad as some here are suggesting, though. It seems to basically work, except for the fact that the demand levels are set a slightly too high for many of them to saturate. The "Supply" number from the "SuDe" statistics is NOT the number of products sold. It's the stock in the product warehouses as a percentage of the base demand prior to the sales step. Even though Motor-Cars are showing "Supply" as 1250% of base demand, the real number being sold is somewhere around 300-400%. If you check the shop statistics you can see that, despite the ever-increasing sales, the auto shop revenue (mislabeled "volume") has been on a downward trend for quite some time.

At QL 0, the auto tree isn't the most profitable one at the current time. In fact, it isn't even anywhere close. I'm just hanging around because at this point, the manufacturing score (which everyone knows is nothing but a competition to see who can build the most cars) is the only one I have a chance at, so I figure I should hang tough and get the gold medal for that one. (I never had a chance at the overall from the start, without QLs) I'm not quite sure why so many other people keep piling into auto tree as if it still made money.

Anyhow here are my proposed changes:

1. I do like One Stroke's idea of adjusting the prices slightly each round, but I think it would eventually lead to all goods being close to equally profitable again, with the same bad effects. Perhaps we should record the fraction of the normal demand which has been sold throughout the game, drop the prices of goods in the top half by 2%, drop prices of the rest by 2%, and then multiply by 0.8+0.4*Rnd(0), and finally multiply all prices by the same factor which depends on the profits of the previous turn, in order to insure that prices don't drift to high or low. This way there would always be goods which were better than others, but which ones they were would slowly change.

2. What if you could change the QL of your manufacturing plants, and the cost to run your manufacturing plants was multiplied by (1+Desired QL/100), or even (1+Desired QL/50), since manufacturing costs are always so small? And what if the demand function was such that you had 1 consumer for each QL who always chose the shop with the lowest price if it had an equal or higher QL, but the lowest Price + Normal price * (QL difference/50) if it had a lower QL? Setting the QL you're selling at would be a rock, papers, scissors problem, and you'd have to worry about which QL you should be selling, instead of always picking the highest one possible.

3. In order to give players some price guidance, I think that we should go to a fixed price-width window for demand, instead of the current one where the window is a set by the number of stores. If the lowest price is P, and your price is Q, your chances of getting the sale would be 1- (Q-P)*10/Normal Price. This is hard to implement without either a random function call or a price memory, in which prices remember previous sales. I'm not sure it wouldn't be better to have store prices relatively stable over the short term anyhow (in the same season). It would be more consistent with how the rest of the game works. I will write pseudo-code for this if you're actually interested.

4. The ROI component of scoring needs to disappear. In the real world, there's always something else useful you can do with your money, so it's a useful indicator. In this game, it's not true, and the ROI gives you a heavy penalty for having made money earlier. As One Stroke has pointed out, the only way to win is to deliberately sabotage your returns in the middle game, which is not fun. Everyone with a high score will also have a high vendor score and manufacturing score, but since the ROI is the change of the log of the company value, their ROI changes will not be high, and this the ranking of this component consequently has a heavily outsized effect compared to the other two. Furthermore, the game handles stock and outstanding requests for goods as if they were worth virtually nothing, causing a high ROI volatility for anyone interacting with other players, making the game less interactive for anyone seriously going for the high score.

The entire daily ranking scoring system serves no useful function, and I would get rid of it. It was intended as a method to balance the importance of the different types of scores, but it ignores the fact that the top players have significant score differences only in the ROI score. There are other was to balance the importance of scores, however.

The current event system is not fun. It usually involves deliberately throwing away cash just to get score points, which most players don't want to do. I'd replace it with the following system:

Randomly, every 1-12 months, a new random product is chosen for an event, which runs concurrently with other events. The event lasts 4 years, or until 4 years of normal demand has been filled. Until this happens, all shops selling the good sell 4x as much as normal at the maximum sale price, and are credited with event points equivalent to maximum price they could have sold their goods for.

The overall high score would be the company value times the new event score. This would force people trying to win the overall score category to spread out into other branches and have many goods available. It would also throw some random price changes into the game to make things interesting.

.

It's a mistranslation as well. The "Supply" for final goods is really stock as a percentage of normal demand. Similarly, the "Sales volume" in our new shop list is really sales revenue (volume would be the number of goods).