PA Economy Factsheet (for new & improving players)

Going to revive this thread briefly, if only to say that I'm going to add the following to the front of the thread at some point later today:

Section 3: Investment and the cost of build power

In the long-term, economy management is about planning a set of investments, and balancing what you need to spend your resources on now against what you'll need to spend them on in the future. You will obviously need to invest your resources in units at some point, but many of your investments will be related directly to your economy. You'll either:

1. be spending metal on your ability to spend future metal production in useful ways.
or
2. be spending metal on things that will increase your metal production.

Point 2 is simple: you want more metal available to spend in future because metal is what you need to get pretty much anything done in the game. You should always be planning how you will increase your metal production in some way, and 80-90% of the time you should actually be building more metal production, whether that's through expanding T1 mexes across the map or through adding T2 mexes in your core territories.

Point 1 is more complex. Your ability to spend metal is your build power, your ability to get things done in the game. While you don't usually want to have vastly more build power than you have metal production - build power costs metal to produce like everything else, and surplus build power is wasted - it is much, much better to have too much build power than too little. More importantly, build power costs energy to run. So with every investment in build power you also have to make a corresponding investment in energy production.

This energy cost is in some ways a 'hidden cost' of building more engineers and factories. But once you understand that when you buy an engineer or factory you also have to buy the energy production to run it, you're well on your way to understanding how to invest effectively in build power.

Some investments are better than others, though. For their cost, T2 power plants are the most productive means of generating energy, giving an impressive 1.852 energy output per unit metal cost, compared to 1.33333 for the T1 generator and 1.451 for the solar array. Solar arrays might have something going for them in that they can't be taken out by ground units, but in terms of raw power the T2 generator is the way to go.

Likewise, as shown in the spreadsheet above, whilst assisting T1 factories is always a waste of build power due to the short production times of units compared to the roll-off times of the factories, assisting T2 factories with T1 bot fabbers is often far more efficient than building new T2 factories due to the high cost of both the factories and the units being produced. In fact there is a magic number of assisting T1 bots for each type of T2 factory, which represents the peak value-for money of investing in assisting bots versus building new factories. So:

T2 bot factories should be assisted by 23 T1 fabrication bots
T2 vehicle factories should be assisted by 19 T1 fabrication bots
T2 air factories should be assisted by 16 T1 fabrication bots
T2 naval factories should be assisted by 53(!) T1 fabrication bots (if you can build it close enough to the shore) -
OR 21 T1 fabrication subs (the most cost-effective naval build unit)

These numbers assume, of course, that the assisting bots are working perfectly (which they don't, always) and that you can actually fit them around the factory in a way that doesn't impede the roll-off of units. They also assume that you're powering everything with T2 generators, and that you're building combat units rather than the much cheaper engineers. Still, these numbers represent your best plan of investment in build power if you want the ability to produce T2 units, especially when you consider that T1 bot fabbers are more flexible than factories and can be dragged off the line to spam up turrets to counter imminent threats.

EDIT: Nope, this makes the front post too big. Well, damn. I'll just point people from the front post to this one.

That's really useful, surprised no one has worked that out and started using it. I certainly will be now!

This is mostly down to Googlefrog: he did the metal cost/unit build power calculations, and I've just extrapolated on them a little and tried to interpret them in terms of an overall economic strategy.

I think the reason people haven't yet picked up on this is that assisting only makes sense in T2; in T1 assisting factories is a terrible idea, so when you make the tech switch actually your entire economic modus operandi changes.

Can you explain how you got the number of engineers to be used? It seems too high.

The numbers are based on the spreadsheet calculations. Basically, for any combination of a factory + (n) assisting bots, you calculate the total metal cost of buying both the factory and assisting bots AND buying the proportion of a power generator you need to power their combined nanolathes for the rest of the game.

You then work out, for each number of assisting bots, how much of a factory's build power will be wasted due to sitting idle during roll-off time (this changes with the cost of each unit, but not by that much: it only throws the final calculation out by at most 1-2 assisting bots). By factoring this waste into the calculation you can then work out the total cost, in metal, for each effective unit of build power you get, for each number of assisting engineers.

Then you just find the number of assisting engineers for which that cost is at its lowest; that gives you the ratio of factory:assisting bots for which you spend the least total metal per unit of build power. This does only give a theoretical optimum, however: in actual fact all it means is that the closer to the optimum number you get, the more build power you're getting for the total infrastructure spend.

The number does seem high, but when you think that a T2 vehicle factory costs somewhere in the region of 30 T1 bot engineers, whilst having the build power of at most 4, it should be clear that the weaker energy efficiency and wasted build time due to roll-off when assisting with T1 bots is only going to go so far to compensate.

Useful study thanks

When doing this I was a bit surprised at how good assisting is. With a longer term strategic view it is probably good to do some assisting for T1 factories because then you'll have lots of constructors to put on T2 factories.

You've missed an important point here. The factory and constructors do not drain energy when they idle during the roll-off time. The spreadsheet takes this into account by multiplying the energy drain by the proportion of time which they are active to give average energy drain. This means you need enough storage as buffer and to make sure your factories don't synchronize their production. The cost of a constructor is partially in base cost and partially in the cost of energy generation required to run it. This effect means that a higher energy drain and lower base cost is better for constructors when assisting.

Hmm yes I have. It is an important step, but it's also one that's rather weakly defined in the current set of calculations. It doesn't, however, really affect the overall logic of them.

For simplicity's sake, the spreadsheet really defines a minimum spend on energy production, and doesn't account for the fact that, for example:

1. It's impossible to buy a fraction of a T2 power generator; you can use part of its production for something else, but that's not always going to be the case.

2. It's possible to predict the cost of the storage needed to smooth out a single factory's production for a given combination of factory, assisting bots, and units produced.

Since these details are quite frankly an arse to add into the calculations, and we both understand that with the idle energy spikes accounted for the spreadsheet gives a best-case scenario, whilst without they provide a worst-case scenario, I think they're rightly omitted.

I agree that we don't need those extra factors in the calculations. When you have a lot of infrastructure the effect of being unable to buy partial energy generators will average out. Also you will not need much storage because your factories can have their downtime offset.

When you have a small enough amount of resources to make the effect of discrete purchases the analysis in the spreadsheet is the wrong way to approach the problem. You're better off thinking in terms of build orders instead of efficiency when the number of things you can afford to build is low. Cheap things start paying for themselves faster and this is most important in small economy cases.

So there is not much overlap between the usefulness of the analysis in the spreadsheet and when it would be useful to know how much storage you need. By the time the analysis becomes useful your factories will not require you to have more storage than normal.

Hmm I see your point. That said, the small-economy factor is actually fairly prevalent in the game: your economy is small at the start of the game, and that's when you need the most aggressive optimisation because of the broadly exponential growth rate of the PA economy. That's why I used a discrete model rather than continuous model for the factory comparison sheet. But these things definitely will average out over time, and certainly when T2 kicks in the assumptions of your analysis will be justified.

I've been thinking about trying to put together a build-order calculator on and off these past weeks as a means of sidestepping that small-economy factor in this kind of analysis. But that would be a fairly complex spreadsheet to say the least! And I'm not sure what it would offer over just testing build orders in the game; perhaps it would tell us how many doxes we could afford to buy as early scouts/harassers, and when to build that early air factory?

Uh... Can you add a column to the spreadsheet which does that calculation?

This is a very informative thread. I had no idea there were energy efficiency concerns until reading the OP.

Balancing Discussion:
Submarine Fabber: I think it's odd how the submarine fabber is more efficient. This seems backwards, because once Uber adds the submerge the submersible fabber will have a stealth advantage.

Vehicle Fabber: Looking at the stats, vehicle fabbers are the same thing as bot fabbers but with less health and higher price tag. Would be nice if they fabbed 15 metal per second and had 225 health. Down side could be a movement of 7. Right now they are the same speed as bots which is 12. Would fit similarly if the Adv. Vehicle Fabber had 45 metal per second and a movement of 7 as well.

General T2: I think it would be nice if T2 fabbers were -25% more energy efficient than T1. This would make them one step closer to commanders and factories than their T2 counter parts when energy becomes scarce.

General Factory: I believe the Uber discussion with the community was that Factories would spend metal faster than Fabbers to reduce the advantage of assisting. 20 assistors per factory is a traffic issue IMO. I think we could easily double the metal spent by each factory and the energy cost of doing so in the next patch without any big concern about negative side affects.

@OP added a tab to the spread sheet with double factory production numbers. Hope you don't mind.

Air Factory: I understand why air fabbers are less efficient than their slower land based counterparts; however, the Air Factory has no mobility advantage. Therefore, looking at the numbers the build rate should be buffed to 24 at minimum, which would still be relatively weak due to the higher initial factory cost but beat fabrication bots.

Fabrication assisting has advantages that never get old:
1. You probably have some extra fabbers in your base when not building defenses that are 'free' work.
2. You probably can squeeze more production into a smaller area with assisting providing defensive advantages.
3. You can prioritize your expenditures and accelerate specific fabrications.

Submarine fabber has only a limited number of buildings it can build - it builds hardly anything compared to the fab ship.

They need to be at the same speed as bots. As it is, everyone goes bots because they cost less metal. Faster speed is more important than faster build speed

Except if you double factory energy cost, it slows down the beginning of the game (the amount of infrastructure before first factory)

Absolutely agreed

Good information. Maybe this is a reasonable compromise.

This is an area of ambiguity. The fact that the bot fabbers are better than vehicle fabbers at everything is clear (cheaper and tougher with no cons). The speed reduction I discussed was primarily just to differentiate the two fabbers more, becuase they are similar.

Scathis could just as easily increase the speed of the vehicle fabbers to that of the skitter and they could be weaker and faster than bot fabbers.

Unit balance does have a desire/mandate to differentiate roles.

I do think economic efficiency would beat movement speed in assisting and defenses and I think vehicles are currently more defensive. Maybe discuss further?

I see what you are saying. The commander energy production could be adjusted. Keep in mind if your commander is assisting your factory you are already quadrupling energy cost and production speed. Also, I think Quitch has discussed simple solutions to the early game lul.

That's true, he did.

Erm.. . This is a hot topic in the forums again... You're right.

Take the pelter for example. If you have a bot 3000 units away from an air fabber, you are better off moving the bot to build the pelter, rather than getting the air fabber to build a pelter at its own location, because either way it will be completed at the same time.

That is the major argument against fast units with less fab rate and less fab efficiency

As someone who never played the previous Annihilation games and just dipped his toes into the beta for PA, I found this thread extremely helpful.

Thanks!

That's all assuming only one air fabber. That's a strawman.

Unit Speed less important than economic efficiency? I was comparing apples with apples.

This deserves a sticky.

This absolutely does deserve a sticky, it deserves moving into general discussion also.