What's your research strategy?

[Ray F]

So, under this scheme, the optimal strategy for pure output of RPs (disregarding tactics, in other words), would be to equalize all fields, and always make sure that every turn your research output is never decreasing from the previous turn. Pretty simple to manage, eh?

What if we just said that, in addition to the research matching bonus for going up to 1/6th, there is an additional 10% interest on the cumulative spending?

This would retain the "focus vs balanced" decision created by the 1/6th change, but add a "now vs later" decision created by compounding interest.

These two gameplay decisions are not connected so there is no reason that the two mechanics need to be connected. Perhaps we should rename them for clarity.

For example, we could call 50% bonus for up to 1/6th of total empire spending a "private sector" bonus. This is analogous to government investment in a field spurring additional private-sector spending (up to a limit, of course).

The 10% on all accumulated spending would be "interest" because it compounds.

Then, in the alpha, playtesting could then help find better values for the 50% and 10%, if those are too powerful (or weak).

[Psillycyber]

So, under this scheme, the optimal strategy for pure output of RPs (disregarding tactics, in other words), would be to equalize all fields, and always make sure that every turn your research output is never decreasing from the previous turn. Pretty simple to manage, eh?

What if we just said that, in addition to the research matching bonus for going up to 1/6th, there is an additional 10% interest on the cumulative spending?

This would retain the "focus vs balanced" decision created by the 1/6th change, but add a "now vs later" decision created by compounding interest.

These two gameplay decisions are not connected so there is no reason that the two mechanics need to be connected. Perhaps we should rename them for clarity.

For example, we could call 50% bonus for up to 1/6th of total empire spending a "private sector" bonus. This is analogous to government investment in a field spurring additional private-sector spending (up to a limit, of course).

The 10% on all accumulated spending would be "interest" because it compounds.

Then, in the alpha, playtesting could then help find better values for the 50% and 10%, if those are too powerful (or weak).

Yes, except Ref's objection to making the 10% interest contingent on cumulative spending, rather than recent spending, is that it encourages the player to seed his techs and then turn off research, which also, by the way, is more micro-intensive than being encouraged to keep tech spending level or gradually increasing, as my most recent proposal incentivizes.

I like the idea of nailing down the exact numerical values in alpha. But it would be good to have the overall framework nailed down ahead of time, so I'm glad we are all getting a chance to hash this out.

[HansLemurson]

Until I read this thread, the way that I had always thought MoO1's research system worked was that any major funding changes took a turn to take effect.
If you were putting 100BC/turn into a category, this gives you 100RP.
If you increased funding to 250BC/turn, then you would still get 100RP for that turn.
On the next turn if your funding stayed at 250BC, you would get 250RP.

Essentially this means you'd get penalized whenever there was a major change in funding. If your RP dropped very low for a turn before being restored to normal then you'd essentially lose a turn of science-money.

There are clearly some flaws in this idea, which is probably why it's not the actual mechanic.

[HoursWithoutTime]

This thread totally confused me. But that happens a lot. I'm not the sharpest tool in the shed...

I'm trying to learn a new programming language and the idea of using it to create a gui app, along with writing to an Excel file would be good practice and how long could it take to gather up the tech research numbers you guys are talking about and try to sort it out. A couple of hours, right? Haha, more like 10.

Anyway, I started a game as the Alkaris and used oreo to give Altair 200 factories and 100 pop. Then I maxed out tech in everything but Planetology (to avoid the Planetology bonus). The app I wrote checks the save1.gam timestamp every 2 seconds and grabs the research levels when it sees a change. Then it writes each of the 6 tech categories to an Excel spreadsheet.

I ran a few different tests. The Alkaris were generating 139 research points per turn and that number wasn't changing because factories and pop were maxxed and I was building anything.

Test 1: 46 clicks to computers & 4 clicks to construction for 10 turns:
Then 4 clicks to computers and 46 clicks to construction for 10 turns:
Then 1 click to computers and 1 click to construction and 48 clicks to force fields for 10 turns:
It doesn't look like I can post attachments so copy and pasting the data will have to do - copy and pasting into a spreadsheet will give you a better view

Year Computers Construction Force Fields Total Diff
2300 0 0 0 0
2301 114 9 0 123 123
2302 258 21 0 279 156
2303 423 35 0 458 179
2304 613 51 0 664 206
2305 831 69 0 900 236
2306 1082 90 0 1172 272
2307 1371 114 0 1485 313
2308 1703 142 0 1845 360
2309 2084 174 0 2258 413
2310 2465 207 0 2672 414

2311 2498 365 0 2863 191
2312 2531 546 0 3077 214
2313 2564 754 0 3318 241
2314 2597 994 0 3591 273
2315 2630 1270 0 3900 309
2316 2663 1587 0 4250 350
2317 2696 1952 0 4648 398
2318 2729 2333 0 5062 414
2319 2762 2714 0 5476 414
2320 2795 3095 0 5890 414

2321 2801 3101 119 6021 131
2322 2807 3107 269 6183 162
2323 2813 3113 442 6368 185
2324 2819 3119 641 6579 211
2325 2825 3125 870 6820 241
2326 2831 3131 1133 7095 275
2327 2837 3137 1435 7409 314
2328 2843 3143 1783 7769 360
2329 2849 3149 2182 8180 411
2330 2855 3155 2581 8591 411

[HoursWithoutTime]

Test 2: 10 clicks to everything but planetology for 30 turns:

Year Computers Construction Force Fields Propulsion Weapons Total Diff
2300 0 0 0 0 0 0
2301 24 24 24 24 24 120 120
2302 54 54 54 54 54 270 150
2303 89 89 89 89 89 445 175
2304 129 129 129 129 129 645 200
2305 175 175 175 175 175 875 230
2306 228 228 228 228 228 1140 265
2307 289 289 289 289 289 1445 305
2308 359 359 359 359 359 1795 350
2309 439 439 439 439 439 2195 400
2310 520 520 520 520 520 2600 405

2311 601 601 601 601 601 3005 405
2312 682 682 682 682 682 3410 405
2313 763 763 763 763 763 3815 405
2314 844 844 844 844 844 4220 405
2315 925 925 925 925 925 4625 405
2316 1006 1006 1006 1006 1006 5030 405
2317 1087 1087 1087 1087 1087 5435 405
2318 1168 1168 1168 1168 1168 5840 405
2319 1249 1249 1249 1249 1249 6245 405
2320 1330 1330 1330 1330 1330 6650 405

2321 1411 1411 1411 1411 1411 7055 405
2322 1492 1492 1492 1492 1492 7460 405
2323 1573 1573 1573 1573 1573 7865 405
2324 1654 1654 1654 1654 1654 8270 405
2325 1735 1735 1735 1735 1735 8675 405
2326 1816 1816 1816 1816 1816 9080 405
2327 1897 1897 1897 1897 1897 9485 405
2328 1978 1978 1978 1978 1978 9890 405
2329 2059 2059 2059 2059 2059 10295 405
2330 2140 2140 2140 2140 2140 10700 405

[HoursWithoutTime]

Test 3: All to computers for 30 turns

Year Computers Total Diff
2300 0 0
2301 125 125 125
2302 282 282 157
2303 463 463 181
2304 671 671 208
2305 910 910 239
2306 1185 1185 275
2307 1501 1501 316
2308 1865 1865 364
2309 2282 2282 417
2310 2699 2699 417
2311 3116 3116 417
2312 3533 3533 417
2313 3950 3950 417
2314 4367 4367 417
2315 4784 4784 417
2316 5201 5201 417
2317 5618 5618 417
2318 6035 6035 417
2319 6452 6452 417
2320 6869 6869 417
2321 7286 7286 417
2322 7703 7703 417
2323 8120 8120 417
2324 8537 8537 417
2325 8954 8954 417
2326 9371 9371 417
2327 9788 9788 417
2328 10205 10205 417
2329 10622 10622 417
2330 11039 11039 417

[HoursWithoutTime]

Final overall totals:

Test 1 - splitting research between computers and construction and finally with force fields: 8591 RPs in 30 years
Test 2 - splitting research evenly among all the categories except planetology: 10700 RPs in 30 years
Test 3 - putting the whole shebang into computers: 11039 RPs in 30 years

Conclusions? Anyone? Personally I always thought the strategy guide might be wrong about how splitting research evenly would produce the best results, although that's what I tend to do through the middle game. It looks like actively moving research between the categories can backfire - that might just be the way I did the test though.

[RFS-81]

Maybe I'm dense, but I'm not sure I understand what you're trying to do and what you're confused about...

The RP that get added to a tech consist of your spending this turn + a bonus. The bonus is either 15% of the RP that are already invested in the tech, or twice your spending this turn, whichever is lower. A lot of discussion in this thread was about how this system should be changed in a spiritual successor of MOO, maybe that's what confused you?

The optimal research strategy is to put some large lump of cash into a field (the precise amount depends on how much efficiency you want to sacrifice for speed), and then invest just enough to reap the full bonus. If you're not using kyrub's unofficial patch, that's pretty hard to do in practice though.

Your Test 3 really surprised me. Based on the math, slow-and-steady spending should be more efficient than crash-funding, and the seed-and-trickle method described above should be more efficient still. I think what's going on here is due to rounding errors: Your total RP after the first turn are 120 in Test 2 and 125 in Test 3, so you lose 5 points in Test 2 to rounding. Thanks to the power of compound interest (and probably some more rounding errors in later turns) that snowballs to the large difference you see in the final outcome. I think that rounding errors matter less if you have a large empire, so dumping everything into one field would probably not be a good idea in practice. Maybe that's something you'd want to test?

EDIT: I've been thinking some more...If you're putting the research output of a single colony towards end-of-the-ladder techs, that should always qualify as slow-and-steady no matter how you split it up, so it's no surprise that rounding errors start to dominate.

Also: Why did you pick that particular strategy for Test 1?

[HoursWithoutTime]

The reason for testing this goes back a few years to when I read in the OSG that splitting research points evenly among the categories would give the best bang for the buck. But I've always known that that can't be true - X % of one big pile of cash would yield the same return as the same amount of cash divided up into smaller piles, each collecting the same interest rate. Another reason is the 7.5% thing. I frequently give 1 click to each of 5 techs (so I don't lose research points), and give the remaining 45 to a tech I desperately need. Have I been missing out on bonuses in my who-knows-how-many times playing this game? So I tested.

Yeah, I noticed the round-off errors too and I suspect that's the difference between dividing evenly and putting everything into a single tech. But it's still worth noting, putting everything into a single tech doesn't hurt.

What I was trying to do in test 1 was test the 7.5% rule. 10 turns with one tech getting just over 7.5% with another getting the rest. And then reverse it for 10 years and see who's ahead. And then try reducing those 2 techs down to a single click and see how they do. 50 clicks to play with in the tech screen and giving one category 46 and another 4 should have meant that they were getting 92% and 8% respectively. But my test was flawed - thanks to round-off errors, it was more like 92.7% and 7.3%. Bummer. So here's test 1 again. This time 45 clicks to computers for 10 turns and 5 to construction and then 5 and 45 for the next 10. The percents now are 91% and 9% so this should be a better test for the 7.5% rule. What I find interesting, is how after 20 years, the category that started off with less (construction) is ahead of the category that started with almost all the research points. The biggest takeaway for me in all this is splitting tech evenly is a good strategy as is putting everything into one category. Moving tech dollars around might be a good idea for people with lots of grey cells. For me, probably not so much.



Test 1 redux: 45 clicks (91%) to computers & 5 clicks (9%) to construction for 10 turns:
Then 5 clicks (9%) to computers and 45 clicks (91%) to construction for 10 turns:
Then 1 click to computers and 1 click to construction and 48 clicks to force fields for 10 turns:

Incidentally, if it matters I'm using the original Moo purchased in the 90s with the 1.3 patch.

Year Computers Construction Force Fields Total Diff
2300 0 0 0 0
2301 112 11 0 123 123
2302 253 25 0 278 155
2303 415 41 0 456 178
2304 602 60 0 662 206
2305 817 82 0 899 237
2306 1064 107 0 1171 272
2307 1348 136 0 1484 313
2308 1675 169 0 1844 360
2309 2050 207 0 2257 413
2310 2425 246 0 2671 414

2311 2464 407 0 2871 200
2312 2503 593 0 3096 225
2313 2542 806 0 3348 252
2314 2581 1051 0 3632 284
2315 2620 1333 0 3953 321
2316 2659 1657 0 4316 363
2317 2698 2030 0 4728 412
2318 2737 2405 0 5142 414
2319 2776 2780 0 5556 414
2320 2815 3155 0 5970 414

2321 2821 3161 119 6101 131
2322 2827 3167 269 6263 162
2323 2833 3173 442 6448 185
2324 2839 3179 641 6659 211
2325 2845 3185 870 6900 241
2326 2851 3191 1133 7175 275
2327 2857 3197 1435 7489 314
2328 2863 3203 1783 7849 360
2329 2869 3209 2182 8260 411
2330 2875 3215 2581 8671 411

[RFS-81]

The reason for testing this goes back a few years to when I read in the OSG that splitting research points evenly among the categories would give the best bang for the buck.  But I've always known that that can't be true - X % of one big pile of cash would yield the same return as the same amount of cash divided up into smaller piles, each collecting the same interest rate.  Another reason is the 7.5% thing.  I frequently give 1 click to each of 5 techs (so I don't lose research points), and give the remaining 45 to a tech I desperately need.   Have I been missing out on bonuses in my who-knows-how-many times playing this game?

What you're missing is that the bonus for a field is reset to 0 whenever you complete the tech that you're currently researching in that field. For example, if you have a tech that costs 120 RP, and you invest 60 RP over 2 turns, you get it very quickly, but you hardly get a bonus. If you spend less per turn, you will have to wait longer, but you save a larger number of RP that can be invested elsewhere. It's a trade-off between speed and efficiency.

If the cost of a tech was infinite, then you would be right. If you only have a single colony, then the cost of end-of-the-ladder techs is quite close to infinite.