Version 0.17.60

[Adamo]

Doesn't sound like that at all and other people have confused that before

He said "turrets", though.

[Serenity]

He said "turrets", though.

So? He also said that they removed the damage bonus and that you can't select the oil to use anymore. Neither of which isn't true. There are other people who read "removed flamethrower ammo" and thought that referred to turrets.

[BlueTemplar]

I hate these changes about flame turrets only useing crude oil.
This was super unnessecary. The damage bonus was not big anyway so why did you remove it?

That wasn't changed. Only the handheld flamethrower. Turrets don't use ammunition canisters

He was specifically talking about the turrets, not the handheld flamethrower. Turrets now use crude oil since all we have is crude oil until blue science.

Doesn't sound like that at all and other people have confused that before

Only way to deal with those people, as they keep showing up :


----

There was a time long ago I found people here that refuse to use bots. Now it's such an important part of the game. What seems weird to me is all this talk of objecting to pushing bots back like they were in the perfect position before... to be honest in my recent play-throughs of 0.17.x I tech'd into bots way before I had the factory to make use of them. If anything I'm ok for them to come later but it would be nice if they started with a couple of speed boosts already applied.

Are you sure that was about conbots, rather than logibots (with requester chests) ?

[meganothing]

I do not contest the general usefulness of statistics, but i would like to highlight some stuff that in this context makes some of the applications you propose difficult.

The approach from general statistic is providing different information than if you have the equivalent of cookie that would give metadata, such as, when this achievement is reached, the game was purchased for this amount time, the player launched the game XX amount of sessions, of average lengh XXX, the average number of player in the game ?, the countries from which player came from ? the region inside the country where the player came from ? the neighbourhood ?

The gradation might seems irrationnal but the deeper you go the more valuable is your information. On a different aspect, you would need to know which youtubers the person follows, that can give good information no ? , also how old they are ?, what are their academical background ?, how they sit when they play ?

Why should someone need the age of new players if he is NOT interested in any correlation with their age? If we needed exact numbers, sure, steam would practically have to be Big Brother to every player of Factorio. But who needs exact numbers when Wube just needs to know if there is a big trend or no trend?

The only data I used in my example calculation was accumulated number of sales and accumulated number of achievements. You probably would need day by day exact numbers to get meaningful results, but that's it. No privacy-relevant data and no spying on your computer, unless you consider the recording of achievements as spying. No meta-data, no data we didn't expect them to give to developers already anyway.

The concern it raises is willing to rely on statistics at all cost for 100% safe decisions measured and stuff will lead to those need of more precise information on what people do with the things they buy.The different method you can come up to extract meaning from datas are less efficient if the datas are not registered with metadatas that will allow to extract the meaning later on, those implies sending infomation regularly to valve ?

Are we still talking about statistics? There is no 100% in statistics. A statistician can't even tell you the time without specifying an error margin .

Wube doesn't need 100% safety. If they get told there is a trend, or there is no trend, with a probability of 0.9 (i.e. 90% in layman terms), that would surely be good enough for them. It wouldn't be good enough in a life-or-death situation, but for the decision to change oil or not? Definitely.

I'd rather find another solution that doesn't involve Wube needing Valve to do whatever thing over the internet on my computer as a normal developement procedures something like human being expressing themselves debating listening sharing stuff would better suit my little preferences in the matter. and I think you will learn more doing that from what new player experience when they play the game.

Why would it need Valve to do anything they didn't do already and have done for years? You might like to read their privacy statement. It should list all data they collect from you. It should also tell you what data they provide to other parties. I expect they can only forward anonymised and accumulated data, otherwise they could get into a lot of trouble in the EU (and even in some states in the US as well).

[Antaios]

What i dislike is the way there were described as "needed for massive thing " and in a way the counterproductive way to present bots is that it gives incentive to seek for routine sort of if you aim at utilizing them at their fullest potential ( more abuse ), the blueprints system is genious don't get me wrong the bots too, though the interaction of them and human nature and whatever you want to call it, from experience, the multiplayer base when just random people would meet up and discuss to organize, was different before than when 1 or 2 person that launched/owned the server paste their BP all along and newbs just act as some sorts of super robots , that replace the gosth and handcraft things before the real robots.

a)
how other people play and have fun with bots should be of no concern to you
b)
If you play that way with bots, then say it's not fun, that's your own fault. Similar for other people/servers.

Setting up the oil refining has always been my least favorite part of the game. The thing that I did to enjoy the rest of it.

That being said, I was quite nervous about such a big change to the game, and downloaded the new release with trepidation. The old way was what I was used to. I had invested many hours into creating a blueprint that would allow me to set up the refining as quickly as possible so that I could get robots up and running to finish the job.

Up until this release, oil refining felt like "make-work" to me. It feels complicated not because it serves the game, but because it just "has" to be complicated. Part of the reason why it felt this way was the dearth of products that the oil is used in. Heavy oil is used in flamethrowers and can be used in the flamethrower turrets and fuel blocks. Light oil is the same. I do not cast aspersions or otherwise on these game design decisions. I merely explain how I felt about it, and what I did to work around the problem.

This is scapegoating at it's finest.
Tell me, what's your opinion on smelting? It's overall functionally the same, only mind numbingly less interesting, and yet most players have no issue with it.

Petroleum Gas is where it was at. All the important products came from this. I generally set up the initial refining, got bots researched, then advanced oil so that I could set up my 10:1:7 ratio cracking and then get back to the game at hand. The greater efficiency of converting light oil into fuel blocks as apposed to petroleum gas into fuel blocks was irrelevant against the ability to set-it and forget-it. It was not *that* much less efficient, even if you consider the electricity needed to power the structures.

A lot of experienced players 'tedium' with oil processing is their insistence on trying to get it self sufficient as fast as possible. It's like saying you're going to solar to make power self sufficient, then complaining that laying hundreds of solar panels is tedious.

17.60 frees the game up for more player choices. One *can* power the majority of their oil industry off of Refineries refining oil into petroleum gas with a token group of refineries to provide heavy oil (or you can just use Coal Liquefaction if you choose to do so instead). It takes a lot of refineries to do it, though. I was falling behind with fifteen of them running full bore. 45 PG per structure is just not a lot. Modules will help this. I have not added them to the structures yet.

I question this, as it removes player choices relating to the use of light or heavy oil before blue science.

[Optera]

I'm firmly against the over reliance on petroleum products and dumbing down of basic oil processing introduced with 0.17.60.

• Basic oil processing has been dumbed down for beginners, so they struggle later on with cracking from adv. processing.
  It breaks ratio perfect builds ralying on basic + adv + liquefaction. doesn't affect me per se since I prefer circuit controlled cracking, still sad there's one less way to play.
• Rocket fuel now takes solid fuel AND light oil.
  The most efficient rocket fuel chain already makes solid fuel from light oil, now it needs light oil again. Its an 10% reduction in efficiency again for no other reason than to lock it behind adv processing. you could get the same effect with removing solid fuel from heavy and pet.
• Bots have been moved way too far into progression.
  I usually rush into adv processing anyway, but very annoying never the less.
• The only change this "fix" introduced I like is chemical science requiring sulfur instead of solid fuel.
  That change would be nicer if sulfur was made either directly from raw oil or from light + heavy oil instead of the overly used pet.

My solution to 0.17.60 is using Pre 0-17-60 Oil and SulfurFromOils.

[BattleFluffy]

I've thought about this a bit more now. Although I previously said I didn't like the oil change, I am starting to see some upsides that I did not see previously.

One common feature of what I said before was that "AOP isn't useful anymore".
However, AOP is still useful in midgame when it is first unlocked. Long term, AOP or CL needs to be used to make the rocket fuel and lubricant for a megabase - there is considerable debate about which of these options is best.

Previously, BOP was a crappier version of AOP. There was no reason to use it in ultra late game. There were only 2 potentially useful recipes: AOP and CL.
With the oil change, all 3 oil recipes are potentially useful in ultra late game.
That is a good thing.

With that said...

I still think that new players are going to have a confusing time with the "can't mix fluids" message when trying to switch from BOP to AOP with a single long pipe connecting all the refinery's inputs.
I believe the best solution to this problem is simply a more informative error message about what's wrong. To be exact about it, I think whenever the player tries to switch any machine to any recipe that would cause a fluid box conflict, a different error message should be shown, as this is basically a different situation than when they are laying pipes.

[mmmPI]

Why should someone need the age of new players if he is NOT interested in any correlation with their age? If we needed exact numbers, sure, steam would practically have to be Big Brother to every player of Factorio. But who needs exact numbers when Wube just needs to know if there is a big trend or no trend?

If you don't want to understand the general ideas, i can explain why each point is questionnable. The age means something as such: If younger player i average finish the game faster, and suddenly one youtuber whose public is mostly that age launches rocket. Does the trend in come in from the new update from Wube or from many players following the pace of one persone ?

The big trend you can see is absolutely irrelevant if you don't know why it is there. And you have only those highly unprecise allegation like " yeah before the price increase there was a spike " waow, this is so precise i'd bet the future of my company on it for every major update.

The only data I used in my example calculation was accumulated number of sales and accumulated number of achievements. You probably would need day by day exact numbers to get meaningful results, but that's it. No privacy-relevant data and no spying on your computer, unless you consider the recording of achievements as spying. No meta-data, no data we didn't expect them to give to developers already anyway.

The only data you use are general statitstic from aggregated number, exactly as the total % of who voted this and who voted that. Now yes if you want to know why they voted this or that , you would need to be a big brother. As such now the data are pretty useless to detect how much time woud a new player need to launch rocket , ( and even more if oil is easier to get a grasp on ). This you can't tell from the data you propose to use. Now if you take in account the fact that some players let their server running, and some don't , you'd need to know in which proportion in order to account for the real playtime that is needed. Hence you'd need to spy on the person to know if it actively playing or just cooking watching trains. Or just have so much imprecision piled on top on each other that it wouldn't be meaningful at all.

Are we still talking about statistics? There is no 100% in statistics. A statistician can't even tell you the time without specifying an error margin .
Wube doesn't need 100% safety. If they get told there is a trend, or there is no trend, with a probability of 0.9 (i.e. 90% in layman terms), that would surely be good enough for them. It wouldn't be good enough in a life-or-death situation, but for the decision to change oil or not? Definitely.

Yeah sure let's just use no other method than"i think it's true", that's a good way to run a business with 30 or so employee and millions of customer. I specifically mentionned in my post than you need to evaluate your assumption first. You need a way to produce reliable result a few time before you are pretty sure you are not forgetting meaningfull data ( proven by a few correct prediction ). How do you know you have 0.9 ? and not 0.2 ? Well you don't.
EDIT: clarification : you have no way to corelate the spike to a meaningfull event and then evaluate your margin , it could be 0.9 or 0.2, sure the spike is there , but there 20% chance it comes from your update 20% chance it is linked to holiday in that region in the world 20% chance it comes from one mod that was very used that broke in an update and many old player were using only this one and so forth.

I expect they can only forward anonymised and accumulated data, otherwise they could get into a lot of trouble in the EU (and even in some states in the US as well).

Maybe this should hint you at WHY it is not possible to go for your approach with such confidence ? the paradox is as such : If you only have anonymised and accumulated data , you can't extract the meaning of individual behavior, like from the % of voter you can't extract their revenue/age/etc. If you have data that allow you to do what you describe, then it's probably forbidden by law in regions that are significant enough in size to make the whole approach problematic.

[mmmPI]

What i dislike is the way there were described as "needed for massive thing " and in a way the counterproductive way to present bots is that it gives incentive to seek for routine sort of if you aim at utilizing them at their fullest potential ( more abuse ), the blueprints system is genious don't get me wrong the bots too, though the interaction of them and human nature and whatever you want to call it, from experience, the multiplayer base when just random people would meet up and discuss to organize, was different before than when 1 or 2 person that launched/owned the server paste their BP all along and newbs just act as some sorts of super robots , that replace the gosth and handcraft things before the real robots.

a)
how other people play and have fun with bots should be of no concern to you
b)
If you play that way with bots, then say it's not fun, that's your own fault. Similar for other people/servers.

a) why shouldn't I be concerned ? If many people can't handle oil, there is a change to it, i'm concerned. If many people don't use the pickaxe or find it annoying, it gets removed for me too. If i have the impression that one usage is an abuse of what the devs wanted to do. I do feel concerned enough too voice the fact that i think it's a problem. What if devs decide that bots needs a nerf because they consider the way some people use it an abuse ?

Also i feel the complaint from delaying them comes as exagerated, so i try to understand why it is such a big deal for others, because i play with other people and i care what they think/may think about the game i offered them for example . (you told me that was a fun game about robots really it gets boring fast ).

b) I'm not playing like that because i find it boring, if other people like it it's fine for me, now i voice my opinion for what i think would be better for the game. I do not ask for what i would particularly prefer for me ( what should be vanilla VS what i use now that i know the game well), even i advocate for something that i would not be using myself, but i think it would make the game itself better as interactive fun teaching software . I think such software would be better as it was pointed out in the thread "If players learn by themselves guided by the system". It is a concern to me this idea of "teaching the wrong way to make solid fuel", it would be self inflicted for robots => I agree with that unless the games forces them to your throat as soon as you start.

Even though i do wish the new players were introduced to conbots faster as i really think it's important in the fun of factorio, i am concerned that the devs are shy implementing that, maybe because it would look like validating a "wrong" behavior.

[huancz]

I already voiced why I don't like the oil change in some way in the other thread, but it was burried in the comment and unclear...

Of course there is the short term pain of broken base, realism down the drain etc... But there are games in development that regularly wipe all saves (I got Factorio precisely because I refunded one such game and had free $19.99 in store credit. Not because of the wipes, but they took the game in completely unacceptable direction - so far from the original promise that they had to remove it from sale on GOG). Here most bases could be repaired with reasonable effort. I already got over it, wanted to try few mods anyway so I just started a new map. Realism was already broken in favour of gameplay in several areas (consuming PG to obtain sulphur included).

I feel that worst was the way it was communicated. What I'm missing is some reason, or data. In all that 70+ pages of comments (I read most of them, but not 100%) I failed to find official reason WHY do devs even think that something as radical as this must be done, and on such short notice. Do they receive 1 forum complaint per six month "my oil is not working, why??? I'm refunding right now!!!@1". One every week? Every 10 minutes on discord? Is it about the achievement ratios? Based on the outcome, frame for this discussion was from the start "Oil is too hard and that is given, you may have a chance to influence what solution we choose if your argument is really good. BTW the fix we already cooked is going live 3 days from now, get ready for significant rebuild". Coming up with good argument is pretty hard if you don't know what precisely is the problem.

Doing something just for the feeling that you are doing something is the best way to make matters much much worse than the original issue. Those mock reasons I gave are AFAIK just speculations from the community. No word of what they are trying to accomplish, aside from "make the mid-game experience smoother". I have no idea how this smoothness was/will be measured.

I don't think they did this just for the fun of it - the fact that they managed to get this game to the state where we can be passionate about it is proof enough that they are competent. They just didn't feel the need to convince us with data that the problem is as dire as the solution was. *THIS* is what made it unnecessarily hard to swallow. For me at least.

[BlueTemplar]

[...]
I feel that worst was the way it was communicated. What I'm missing is some reason, or data. In all that 70+ pages of comments (I read most of them, but not 100%) I failed to find official reason WHY do devs even think that something as radical as this must be done, and on such short notice.
[...]

Didn't these cover it ?

OK, I know nothing about developpement - I play, write & teach music - but I know for sure when I'm facing someone passionate about his work, and what I read here makes me a little bit mad. Everyone complaining about being ignored should read those 4 posts :

viewtopic.php?f=38&t=73684&p=445621#p445621
viewtopic.php?f=38&t=73684&p=445733#p445733
viewtopic.php?f=38&t=73684&p=445794#p445794
viewtopic.php?f=38&t=73684&p=445843#p445843

and look at the date : 3 of them were written on a sunday. Did you read them ? Did you test the Oil changes mod V453000 made to allow us to test these changes ? From what I know, the oil changes were delayed to allow the team to look further on this, but it wasn't a guarantee that your solution got picked.

Requiring more consideration from Factorio devs is basically asking for a personal Factorio 24/24 hotline.

----

Remember that in 0.17 there's a huge new change that is impacting how we build our factories :
the new ghost item / blueprint tools !

Now, like conbots, they are not intended for new players, and so are hidden by default... maybe for good reason ?

[BattleFluffy]

Some are saying the oil change simply pushes the multiple-fluids puzzle further down the tech tree, meaning the player may need to reformat their refinery or replace it completely.

I personally do not see anything wrong with this. I think it's natural that later science packs should be harder than early ones. Up until now Chemical Science has been the hardest one to make. This change makes Chemical Science easier and the later sciences harder. That just makes good game sense to me, like the first level in a game should be easy, and the last levels should be much harder...
I also don't see an issue with reformatting things either. To me that's just the nature of the game. We already reformat our smelters for electric furnace, this doesn't seem particularly different.


Seems I am gradually moving into the "for the oil changes" camp again..

[meganothing]

If you don't want to understand the general ideas, i can explain why each point is questionnable. The age means something as such: If younger player i average finish the game faster, and suddenly one youtuber whose public is mostly that age launches rocket. Does the trend in come in from the new update from Wube or from many players following the pace of one persone ?

The big trend you can see is absolutely irrelevant if you don't know why it is there. And you have only those highly unprecise allegation like " yeah before the price increase there was a spike " waow, this is so precise i'd bet the future of my company on it for every major update.

"Bet the future of my company" in relation to the oil change is hyperbole. We are speaking about a relatively small balancing change. Only the Factorio forum is not used to this and thinks this is a world-changing event because it is used to an exemplary restraint by Wube which I haven't seen anywhere else in EA.

Yes, the spike is not precise. And the resulting spike of rocket achievements is even less precise and almost certainly smeared over a larger time frame. But if you have multiple of these events and enough new players you can calculate with sufficient precision to make predictions that are not too far off the truth.

if you have 8 sale spikes and in one or two of those the data is very different because of a youtuber or some promotion, it is detectable that those two should not be counted. If youtubers have such an influence on rocket achievements there would also be spikes independant of sales and you would probably see that youtube spikes have a different sharper form and could discriminate them from normal sale spikes.

Sure, you could hypothesize that the majority of those 8 spikes are all tainted by youtubers, but first of all, the probability of that should be very low and the statistician can at least see that the data is unusable. There won't be a wrong result in this case, just the result that the data isn't usable/conclusive.(in that case Wube would have to rely only on test plays and the rather unreliable general feedback of users complaining about oil)

The only data you use are general statitstic from aggregated number, exactly as the total % of who voted this and who voted that. Now yes if you want to know why they voted this or that , you would need to be a big brother. As such now the data are pretty useless to detect how much time woud a new player need to launch rocket , ( and even more if oil is easier to get a grasp on ). This you can't tell from the data you propose to use. Now if you take in account the fact that some players let their server running, and some don't , you'd need to know in which proportion in order to account for the real playtime that is needed. Hence you'd need to spy on the person to know if it actively playing or just cooking watching trains. Or just have so much imprecision piled on top on each other that it wouldn't be meaningful at all.

It doesn't matter if users need more or less time. They definitely will be smeared over a long time. Does the term "bell curve" ring a bell? While they are smeared, it is very probable that you can find an intervall of say 20 days where the majority (60, 70, 80%, who knows) of new users have either reached rocket or given up. And this time is stable enough (even with changes like the oil) that you can still detect that smeared spike. If not, Wube has to employ other means, like test users.

About letting the server running. How many new players will really start the game on constantly running servers? First of all this is something Wube can find out as well, secondly it probably is a small minority. Even if it were half the new players that would just mean two smeared spikes.

Yeah sure let's just use no other method than"i think it's true", that's a good way to run a business with 30 or so employee and millions of customer. I specifically mentionned in my post than you need to evaluate your assumption first. You need a way to produce reliable result a few time before you are pretty sure you are not forgetting meaningfull data ( proven by a few correct prediction ). How do you know you have 0.9 ? and not 0.2 ? Well you don't.
EDIT: clarification : you have no way to corelate the spike to a meaningfull event and then evaluate your margin , it could be 0.9 or 0.2, sure the spike is there , but there 20% chance it comes from your update 20% chance it is linked to holiday in that region in the world 20% chance it comes from one mod that was very used that broke in an update and many old player were using only this one and so forth.

Statistics works with multiples, the more the merrier. In the last 3 years (in which Factorio was already much of what it is today) there were about 5 steam sales per year creating 15 events you can compare and evaluate. You can for example hold 3 of those events back, make a hypothesis and check that hypothesis with the 3 events held back. Or wait for the next steam sale this year. We naturally both have no idea if there are big enough spikes to make a predicition, only Wube does. But if for example a sales spike is followed by an achievement spike in 15 of those 15 events you can calculate the chance it comes from the sale and it likely will be above 0.9. If only 5 of those events produce a distinct spike or they vanish in a sea of spikes you can already, without any math, see that the data is not good enough to give you such information.

Holidays are open knowledge, you can see their effect or non-effect in a region every year. You can compare the steam sale near the holiday with a steam sale without that holiday or a different holiday and observe the effect in that region. You can compare with the worldwide sales and substract the effect of that holiday in that region, even if it were a substantial influence which I highly doubt.

A million dollar bussiness or a backstreet barber shop makes no difference, except the million dollar bussiness likely has enough customers so that statistics really works. Again, Wube might not have the numbers to make a prediction with 0.9 probability, but they likely have the numbers to say whether their confidence is 0.9 or 0.2. Error margin prediction is one of the key ingredients in statistics, well researched and not voodoo science. Even if Wube can't do it there are people or companies for hire who can do it.

Maybe this should hint you at WHY it is not possible to go for your approach with such confidence ? the paradox is as such : If you only have anonymised and accumulated data , you can't extract the meaning of individual behavior, like from the % of voter you can't extract their revenue/age/etc. If you have data that allow you to do what you describe, then it's probably forbidden by law in regions that are significant enough in size to make the whole approach problematic.

Statistic works because if in 2017 lets say 30% of new users were teenagers then in 2018 probably again around 30% of new users were teenagers, plus minus 1 to 5%(?). And that difference is small enough that it doesn't matter in comparing the years and you can even calculate the maximum error it generates. Even if teenager new users were playing the game completely different. Even if some youtuber once generates a spike that needs to be substrracted again.

I think it's time to agree to disagree.

[netmand]

There was a time long ago I found people here that refuse to use bots. Now it's such an important part of the game. What seems weird to me is all this talk of objecting to pushing bots back like they were in the perfect position before... to be honest in my recent play-throughs of 0.17.x I tech'd into bots way before I had the factory to make use of them. If anything I'm ok for them to come later but it would be nice if they started with a couple of speed boosts already applied.

Are you sure that was about conbots, rather than logibots (with requester chests) ?

I'm not sure what that was about since I wasn't one of them. But it was far back enough to when you tech'd into bots you got both, and there was no separation of the chests. I think the point was to stick to belting so maybe they were only talking about logistical bots... and still using con bots? yeah ok that seems stranger for a stigma to have but whatever. I just found it odd, people here are sometimes odd, including me. e.g. I don't believe in using mods in these experimental releases.

[BlueTemplar]

I prefer to avoid logibots too, as I don't really like how they change logistics to be about recharging (etc.) rather than layout...
Heck, I'd rather use logicarts instead !

[Theikkru]

Some are saying the oil change simply pushes the multiple-fluids puzzle further down the tech tree, meaning the player may need to reformat their refinery or replace it completely.
[...]

This is considered a problem because the game teaches players to do things one way (1-in-1-out), but then turns around and tells them they did it wrong, and forces them to redo things another way (multiple outputs) later. See:

[...]

I believe it is very common in Factorio to have to rebuild things, because the player did not expect some things to be needed later, or unlocked a more efficient way to do something (better belts, beacons, logistic robots), or the amount of for example electronic or advanced circuits they will need.
However, I tend to agree that having to rebuild because the game suddenly gives you a new mandatory recipe could potentially feel more arbitrary and forced, especially newer players tend to build just s few refineries so the issue does not have to be huge, and setting up a separate refinery instead of altering the original one also has a quite a bit of value as an opportunity to do it again and better.

While this may be true purely statistically, I think it is very important that the rebuilding be entirely at the discretion of the player. Belts, furnaces, beacons, and robots are all things that a player may decide to upgrade, rebuild, or replace, but always entirely of his/her own volition, because he/she concludes that the benefits of the rebuild or upgrade outweigh the effort lost in demolishing the old setup and the time expended reconstructing the new. Should a player conclude the opposite, the rebuild does not take place, and some other alternative is chosen, with player and game no worse for wear.
Future demand is not entirely unexpected either; a player can always ask his/herself how he/she would expand his/her production to accommodate higher demand whenever designing a new line, and leave space for expansion accordingly. In normal cases, so future-proofing affords the player some degree of insurance against having to tear down his/her hard work in the face of increasing demand. Conversely, the player who does not plan ahead thus has no one but him/herself to blame for the predicament, in a classic case of reap what you sow.

In this new revision, however, it is not very reasonable to expect a (new) player to deduce from existing information (see next point) that the refinery will require additional connections later, and ergo it cannot be expected that the unknowing player would account for it somehow. Thus, upon encountering advanced processing, the hapless newcomer is essentially commanded by the game mechanics to demolish the old refinery (and all the effort that went into it) in order to accommodate the new. (I will also note here that keeping the old refinery and constructing a new one elsewhere is not a satisfactory option, due to the large disparity in output between the recipes; players would still feel pressured to decommission the old to save on crude.) While this may not faze those who just haphazardly cobbled together a refinery or two without so much as a moment's deliberation, it will come as a rude slap in the face to those who spent time meticulously arranging their construct to nestle neatly between its neighbors, or those who carefully aligned it to be expandable in the future. In effect, the game thus punishes players proportionally to how much effort they invested in their basic refinery setup (and worse, all without construction bots to ameliorate the sting). This is a terrible lesson to teach.

If, however, a light oil output were included, players would have to account for it in initial design, and the astute could infer from it the presence of a 3rd output as well. Even in the case where no planning ahead is applied, the assembly would only require routing of one additional pipe on each side for advanced processing; a far more feasible prospect in most situations.
[...]

[netmand]

I think you're missing the context ... We've all been waiting for a solution ...

Yes I missed the context of your conversation I quoted before. Apologies.
No, I am not waiting for a solution.

I play, I report bugs. I soak information about the game from places like this.

[Yijare]

The removal of picks meant mods couldn't take advantage of them being a thing anymore iirc.

And? Mods only expanded where teh steelpick ended. And yes, i rather have some omnious Diamond-Pick than no pick at all.
Removing the Pick was a bad decision.

These new changes just got more exciting for me. Come on guys, a recipe that only outputs petroleum? How could you not like that? This change potentially enhances the game even after Advanced oil processing tech has been earned.

Plain ol' NOPE.

A way I've made supplying large amounts of metals was to make them at the mining site, so my trains are delivering 4k iron plate instead of 2k iron ore. Of course this sort of thing was too complex for me for oil outposts because there are too many materials involved, but no longer! why ship crude when you can ship petroleum: [...]

And you are hit later on with the large brick that adv. oil processing is now. Basic giving magically only the most advanced stuff is plain ol stupid.

BUT shipping 25,000 petroleum from the oil outpost by tanker can contribute to make 2,500 plastic bars AND I don't have to deal with the logistics of the other materials (Water in, Light and Heavy Oil out). Of course at the outpost I'm using Basic oil processing but that doesn't matter since I'm still filling the tanker.

I'd rather fill an 2-8 train with crude and supply my refineries with that than magically have trillions of barrels of heavily refinded petchochem right of the get go.

I dunno guys can we get out of some of our tunnel vision we have in proclaiming this change a total one way or the other? Can you come up with a way to make this change work for you?

You have NOT read anything in this thread or FFF304 OR FFF305, have you?

[...] imagine someone knowing that earth is round, seeing people tell "For now, just think that Earth is flat". It's a non-sense not to tell them "Earth is like a ball, but physics is for later, you're too young".

Can we at least agree that Nauvis is flat ?

DEUS VULT! \[T]/

[netmand]

Plain ol' NOPE.

It's a shame that many people here feel that way. I've pain-stakingly read most of the three threads you mentioned, and it astonishes me that there are so many people out there playing the game would hold onto an old version of an experimental game. I'll acknowledge that there's been a fair amount of constructive discussion but most of it is based of the fact that people are holding onto an old experience, guessing what new player should be, and feel they're an authority of game progression merely because they've spent x hundreds/thousands of hours playing the game, discussing the game, and maybe even promoting the game.

I spent weeks avoiding to write this very response to all the hub-bub. It's not my intention to imply that we should all just blindly accept changes, especially when they solicit our opinions. I think it's ok to voice our discontent but there's been a edge of aggressiveness and unproductive denial throughout these threads that makes me feel uncomfortable about how our community is helping guide the game's development.

I just got through actual game-play of 0.17.60 into blue tech last night. I must say I find the oil refinery change a good one. I'll admit I was missing my construction bots when building out my base defenses but it wasn't so bad that I was behind the evolution curve. My weapons tech in relation to the biters that show up is just about right, mind you there's still a lot of improvement that could be had in my game-play. But it's still a fun game to play even starting over on a new map. There's still the move into Advanced oil processing and Coal liquefaction comparison that my game needs to make, but to me this change so far is a very viable.

[mmmPI]

I do not contest the general usefulness of statistics, but i would like to highlight some stuff that in this context makes some of the applications you propose difficult.

I should have said impossible to be clearer.

Yes, the spike is not precise ( assuming there is one in the first place). And the resulting spike of rocket achievements is even less precise and almost certainly smeared over a larger time frame.
if you have 8 sale spikes and in one or two of those the data is very different because of a youtuber or some promotion, it is detectable that those two should not be counted.

You'd need to know stuff you don't to associate unless you track stuff you can't + Are you sure steam sales boost factorio sells ? what about the % that is bought on website ?on Gog ? those little % don't count i guess right ? 80% of 80% of 80% is only +/- 50% .

80% player on steam ? yeah the majority, 80% of them play to launch a rocket and not just fight aliens because they bought the wrong game ? yeah the majority, in those 80% don't let their game run when they go to sleep to continue the next day ? that's only 50% of your player left.

If youtubers have such an influence on rocket achievements there would also be spikes independant of sales and you would probably see that youtube spikes have a different sharper form and could discriminate them from normal sale spikes.

Sure, you could hypothesize that the majority of those 8 spikes are all tainted by youtubers, but first of all, the probability of that should be very low and the statistician can at least see that the data is unusable. ( it's not because he sees it that the other person trust him lol)

There won't be a wrong result in this case, just the result that the data isn't usable/conclusive.(in that case Wube would have to rely only on test plays and the rather unreliable general feedback of users complaining about oil)

Same idea you have no way to know such impact unless you ask specifically the users to know if they were watching a youtuber and detect that no other thing that this can make a spike such as the one observed if there is such a spike , you don't need to hypothesize something like that, you need to have a way to know what can impact and how it impact the achievement for that 10%-50% of player ? this is way too imprecise to be reliable.

It doesn't matter if users need more or less time. They definitely will be smeared over a long time. While they are smeared, it is very probable that you can find an intervall of say 20 days where the majority (60, 70, 80%, who knows) of new users have either reached rocket or given up. And this time is stable enough (even with changes like the oil) that you can still detect that smeared spike. If not, Wube has to employ other means, like test users.

About letting the server running. How many new players will really start the game on constantly running servers? First of all this is something Wube can find out as well, No. there's lan play/hamachi play from Gog yet other minorities secondly it probably is a small minority yeah the 80% of 80% of 80% thing, if you just arbitrarily decide that this is minority and shouldn't be accounted for you're making mistake. Even if it were half the new players that would just mean two smeared spikes. That's an hypothethis.

Yeah maybe it's 7 month that will be impractical to use, also the lengh of an average game could evolve with the update, so if you test 2 differents updates in 2 month time, and you want to measure if the average for a new player to get the rocket is higher or lower than 7 month as it was before then it's completly unusable.

Statistic works because if in 2017 lets say 30% of new users were teenagers then in 2018 probably again around 30% of new users were teenagers,plus minus 1 to 5%(?).

No, this is an example of common mistake. If a very popular youtuber amongst the youger audience massively attract teenager this particular year ?

=> oh but we detect this ???? => REALLY without knowing their age and who's their favourite youtuber and without asking them why they got to the rocket ? this is not statistic this is magic !

This is saying like well you know five years ago people voted that, so it will be the same this time. minus 1-5%. ( as if event don't matter. ) At the end the % of voter changes, and you can't use statistic to explain why if you don't ask the people to try and discover a pattern into why they did change their mind since five years ago but you'd need to ask them stuff that are not aggregated anonymised data, all the contrary, you'd need precise qualified data on each individual if you want to extract a pattern that could explain the change.

Now imagine you'd ask to new player "why are you on average younger than the player last year ?" I need to know because i am not sure the game was made harder over the time or if it's just that there are more casual player the closer we are from version 1. This is another thing statistic can't do.

I think it's time to agree to disagree.

I do not say it's impossible to use statistics, just saying that some of the stuff you say is not correct, i think you undersestimate the difficulty of what you say, even if some of what you argue make sense, like it answers the point, but it doesn't do it while being true.

The best metaphora i can come up with is this one : If i put ice cube in my drink, i can study a sample of the composition, and with math sort out the % of water in the whole drink then say how much ice was added. But i can't tell you if there was 1 or 2 or 3 ice cube, this information is lost. I could tell you that if the ice cube were from different color but same concentration of color in the ice. The information would be preserved from a sample of the drink you could say there this % of red and this % of blue and this % of yellow, hence there were at least 3 ice cube. ( maybe there was a bigger red ice cube maybe there was 2 red ice cube ) . Some of the information you cannot recover from aggregated data, no matter how hard you try. ( Maybe there was 100000 small red ice cube if you only have the drink no matter what you theorize on the average ice cube it's irrelevant.

So i think i understand how you start your reasoning, but i think you are mistaken and not seeing it ,hence i disagree for the last time after that PM me if you want to because this feel likes flooding off topic. The only link is that for me it stresses the importance of human beings expressing themselves with words for feedback:)