Ratjoy.com » Forums » EoS General Discussion » I'm done
You need to register or login to post a reply.
I'm done
|
Scott (Admin) RJ: Ratan Joyce CO: Ratan Joyce Post Rating: 0 + / - Total Posts: 1175 Karma: 5083 Joined: Jan 13, 2012 |
Posted on May 15, 2012 @BBPlease see my PM. @Nathan Thanks, but the hardware is already in place. (No donation needed) capitalism-online.com is running on a private server, and will have no problem handling the data if I just move the game over without any changes. I just want to solve the software part because I know once we have 10k+ users, and a problem kicks in, it'll be impossible to deal with. The script since the quality merge has simplified into: a single, fast query to pull out all saleable store-item pairs from the every company's warehouse, and sell them. The main concern is each tick an enormous number of lines gets generated, and yet they must be kept if I were to generate the detailed statistics. (Revenue per store, per item, a pie chart with sales % from market leaders for each product, ...) A simple calculation: 500 active players * 2 companies per player * 10 stores per company * 20 items per store * (1 quality per item) * 96 ticks per day * 100 bytes per row = 1.9 GB per day Replace the 500 players with 10000 players, increase the companies from 2 to 4, the items from 20 to 40, and the size can grow exponentially. Those still playing this game daily AND saying this game won't grow after the change are just not realizing the facts, the game is in Beta and I've not even started marketing the game yet. The game went from 200 to 500 daily players in 3 days when 2 active players posted about the game on 2 different forums - and that's what worried me sick. After the shelves change I'd use other scripts to "summarize" the tables into daily values on each tick, and then wipe out the sales log each tick, and the 8 shelves system will simplify the summary part a lot by linking the summary to "store_id", "shelf_id" (8) pairs instead of "store_id", "product_id" (300+) pairs. |
|
Bob Malone RJ: Bob Malone CO: Malone Post Rating: 0 + / - Total Posts: 341 Karma: 191 Joined: Apr 17, 2012 |
Posted on May 15, 2012 I surrender ;) ... I completely understand your technical concern, but I have the feeling that you focused only on database aspect and not on gameplay impact. But let's see the impact on the game after it's implemented :) I guess this server is also there for that purpose. |
|
Andrew Naples RJ: Clemen Salad CO: Clemen Post Rating: 0 + / - Total Posts: 214 Karma: 89 Joined: Apr 26, 2012 |
Posted on May 15, 2012 I understand the concern too but I fear the cost of this solution will be far too high.
|
|
Jack Park RJ: Jack Park CO: Cicero Post Rating: 0 + / - Total Posts: 19 Karma: 26 Joined: Apr 15, 2012 |
Posted on May 15, 2012 "I understand the concern too but I fear the cost of this solution will be far too high."This is exactly what I think after having read your posts Scott. I apoligize for calling you incompetent, as now I can see where you are coming from and from a purely programming point of veiw it is a very rational decision. I just feel like you did not adaquetly consider gameplay in your decision making proccess. I'm quite sure there are many other ways to reduce data generation, like increasing amount of time per tick. I mean really how much of a diffrence would it make to have there be one tick an hour instead of 4? That would then give you 4 times the amount of time to porccess each tick. Although I can still see who there would be a lot more data left over to proccess if proccessing ability didn't keep up with the number of players. |
|
Jayle Trigger RJ: Berry Punch CO: Applejack Post Rating: 36 + / - Total Posts: 93 Karma: 85 Joined: Apr 9, 2012 |
Posted on May 15, 2012 Does this change really ruin the gameplay? Only if you convinced yourself that it will.1. Players could still run multiple companies that feeds itself materials for retail. 2. Players could dedicate more plots of land to stores, to sell every possible product. Instead of just having 1 or 2 store plots to sell everything, you now need more plots to sell in the same numbers you were previously. Oh no... I'm less efficient now! Save me Capitalism! 3. Players could now more easily get into niche markets that few others will hit because of the difficulty/process to make the products. You know those 'easy' 1-2 factories to make products some of you are complaining about? Yea... they are going to be the first to overflow the demand as everyone and their mother gets in on it, exactly for that reason. Also, those single high quality, high margin items... a good number of players will go after that stuff too, probably overflowing the demand fairly quickly too... (Honestly, Scott could make hitting the demand less forgiving.... hit 100% and not 1 more of that item will sell. Anyone else expanding into the retail business for demand capped items should drop per tic sales for everyone else... trying to sell the same thing... encouraging people to directly compete and diversify... Forgive my ignorance if it is already like that.) |
|
Lorenzo Boccaccia RJ: Mon Opoli CO: Mon Opoli Post Rating: 0 + / - Total Posts: 80 Karma: 71 Joined: Apr 10, 2012 |
Posted on May 15, 2012 think about properly implementing the retail market: no more sales than the market size and the market size that dramatically shrink when average price is over x times the wholesale value..now you have a fair mechanism, not gamey and player driven, by which over specialization is self defeating. it still doesn't address the aspect that a b2b only company may be a not viable option, but it's a start |
|
Jack Park RJ: Jack Park CO: Cicero Post Rating: 0 + / - Total Posts: 19 Karma: 26 Joined: Apr 15, 2012 |
Posted on May 15, 2012 Yeah but that wouldn't happen in a day, you wouldn't be needing to constantly re-adjust production of goods. If a good is over produced by only some 5% then it might not even be worht changing it. It owuld take a very long time for that to happen to any one good and by the it did happen you would likely be half way to setting up the production for the replacment good. |
|
Scott (Admin) RJ: Ratan Joyce CO: Ratan Joyce Post Rating: 0 + / - Total Posts: 1175 Karma: 5083 Joined: Jan 13, 2012 |
Posted on May 15, 2012 Thanks Jayle & Lorenzo, making the demand less forgiving is an easy tweak that can be done (and most likely will be done) whenever it is needed.
|
|
Andrew Naples RJ: Clemen Salad CO: Clemen Post Rating: 0 + / - Total Posts: 214 Karma: 89 Joined: Apr 26, 2012 |
Posted on May 15, 2012 Jayle no one is this aggressively stupid. 1. How does having multiple companies change a damn thing? There is still only one high margin item. Plus how is that better than now? 2. is just wrong. read the updates and misunderstands the issue. Nice snark though. 3. is barely coherent and doesn't exist as the game is now As far as a demand type change, it needs to be fleshed out. Itd be jarring to all of a sudden sell 0 items overnight. |
|
Nathan Dilday RJ: ndkid CO: ndkid Post Rating: 0 + / - Total Posts: 47 Karma: 46 Joined: Apr 17, 2012 |
Posted on May 15, 2012 I don't mind the idea of any given store having only 8 items in stock. The portion I'm less comfortable with is that selling 2 items will be half as effective as selling 1 item. I think the proponents of selling a diverse selection being something that increases, rather than decreases, sales, have a point. And it doesn't sound like providing some small multiplier effect on sales for having multiple things for sale would greatly affect performance.For the technical aspects, I would note that you're effectively cutting the average store size from 20 items to 8 (but, with the current mechanics, probably even fewer); a similar 60% reduction could likely be had going down the path of making all of a given company's land shared (which could probably be implemented now by just having every company start with zero plots open and a certain number of plot openings across three types are free), which seems to be a more positive change than the shelving one. (And, again, it seems the shelving change would likely be more positive if it didn't seem to encourage trying to optimize sale of a single product over sale of 8. Unless booms and busts are happening so frequently that demand can't be predicted with much certainty tick to tick, it certainly seems to be the case that focusing on the single best profit-margin item is best.) (The CS part of my brain also feels that a scaling solution which just changes one value from 20 to 8 isn't actually solving the scaling, it's just letting you double the population before it crops up again.) |
|
Lorenzo Boccaccia RJ: Mon Opoli CO: Mon Opoli Post Rating: 0 + / - Total Posts: 80 Karma: 71 Joined: Apr 10, 2012 |
Posted on May 15, 2012 after reading some more of the old post, I'd say that all pieces will come together when the game gets stable and there is time to add more products and more intermediate steps.once all products have their chains properly implemented (like the car/jet company, requiring a ten or so factory to build stuff end to end) the problem of viability of the b2b market will magically solve itself. Think having more packaging (cans, bottles, boxes, wraps) and intermediate steps (smelting, binding, processing, grinding, packing, forging) for everything (printers sold in boxes and requiring plastic to be molded properly...) and you'll lengthen the production just enough to prevent independence of anyone. |
|
Bob Malone RJ: Bob Malone CO: Malone Post Rating: 0 + / - Total Posts: 341 Karma: 191 Joined: Apr 17, 2012 |
Posted on May 15, 2012 (The CS part of my brain also feels that a scaling solution which just changes one value from 20 to 8 isn't actually solving the scaling, it's just letting you double the population before it crops up again.) Indeed.... Let's be even more conservativ : - only 2 ticks per hour - all store of same type considered as only one - 8 shelves per store 50 000 active players * 2 companies per player * 2 stores per company * 8 items per store * (1 quality per item) * 48 ticks per day * 100 bytes per row = 7.6 GB per day And 50000 users are not that big for such a ( good ) game. |
|
Scott (Admin) RJ: Ratan Joyce CO: Ratan Joyce Post Rating: 0 + / - Total Posts: 1175 Karma: 5083 Joined: Jan 13, 2012 |
Posted on May 15, 2012 You're right Bob, I realized the same thing some time ago, that is when I decided to add a "summarizing" mechanism.The other reason for the shelves is to reduce micromanagement, but still require manual input and NOT a single button that does the purchase/pricing. Also, the new server has 2x 240 GB SSDs, I'm sure it can handle 100 GB without any issue. |
|
Michael Tsui RJ: Reisen CO: Reisen Udongein Inaba Post Rating: 0 + / - Total Posts: 45 Karma: 30 Joined: Apr 26, 2012 |
Posted on May 15, 2012 Scott, sufficient to say, this alone is the reason to delay implementation for at least a week and deliberate better with the players.Once the game gone gold it would be much, much worse if PR continues in a way like this. |
|
Harold Holt RJ: Harold Holt Post Rating: 0 + / - Total Posts: 28 Karma: 48 Joined: Apr 5, 2012 |
Posted on May 16, 2012 Each store is calculated independently in the sales tick? Not each store type? I assumed you would take the total of the store sizes after adjusting for marketing. Wouldn't that reduce a lot of the load?
|
You need to register or login to post a reply.