Research strategy

From Hearts of Iron 3 Wiki
Jump to navigation Jump to search

This is a strategy guide for generic advice on how to best achieve research goals. To better understand research mechanics, read the Research article in the Reference section. Research is driven by assignment of Leadership points.

Knowledge Levels

Knowledge Decay

All Knowledge levels decrease by 2.5% every month. Therefore, in order to maintain a Knowledge level of at least 20 (the maximum bonus), you will need to increase that Knowledge level by about 0.5 every month. For Practical levels this can be done by building units and engaging in combat. For Theoretical levels, this is one tech every two months; which means you will need two tech slots or so depending on your research modifiers.

Practical Knowledge

Practical knowledge is an abstract concept, relating to how much experience a nation has in the particular field. It aggregates such factors as engineering experience from the past, on-going construction, and the like. Practical knowledge is used in the research of several technologies, but is affected only by producing certain units. Like all other fields, the various practical knowledge fields are capped for assisting in research. They do not, however, have a cap for production.

For a table listing what units provide which practical knowledge, please see Practical Knowledge.

What To Research

  • Concentration: It is best to concentrate on one area of tech at a time, as this lets you take the most advantage of Knowledge levels. Depending on how much leadership and IC you have, pick a limited selection of units and focus on those techs. Also, unless your country is tiny or you are in immediate mortal danger, education and the computing techs will be the most important.
  • Research Efficiency: Research efficiency techs (Census Tabulation Machine and the Computing Machine techs) were finally fixed in patch 1.4 and are worth researching.
  • Combat Events: Several techs affect the change of getting combat events. One would expect the attacker to have the advantage overall (otherwise, there would have to be some other strong reason for attacking), but worse modifiers (since terrain, digging in, etc. tends to benefit the defender). By this reckoning, Encirclement (Integrated Support Doctrine) may be the best attacker event, if the attacker has the units to fill in the additional combat width. Otherwise, Assault (Mass Assault) is probably best. On the defending side, Tactical Withdrawal (Elastic Defense) will probably do the most to help a defender hold out longer, followed by Ambush (Guerilla Warfare).

What Not To Research

Obviously you want to focus on techs that benefit you the most and skip the ones that bring you little or no benefit. However, there are some cases where having a tech at all is a bad thing. How? Apart from doctrines, all unit upgrade techs increase the IC cost, time, and supply and fuel consumption of the units they affect. Furthermore, some upgrades penalize certain areas of units. For most components of naval units this is not such a big deal, since if you don't want a certain upgrade level you can build a ship without it. However, other units upgrade automatically, and you have no control over this barring disabling upgrading completely. Some techs you may consider skipping, even if you use the associated units, are:

  • Tank Engine: If you plan to use a type of armor in the same divisions as slower units, you may consider skipping engine techs for that type of armor.
  • Tank Armor: On the other hand, if speed is your goal for a type of armor, you may consider skipping armor techs for that type of armor.
  • Defensiveness: More generally, if you don't plan to defend with a particular type of unit, you might skip its Defensiveness upgrades.
  • Toughness: Likewise, if you don't plan to be on the attack with a particular type of unit, you might skip its Toughness upgrades.
  • Weak Upgrades: Some upgrades may not be worth the price. For example, is it worth researching Rocket Carriage Sights for a paltry 0.2 points of Toughness? Militia Guns for half that amount?
  • The pure theory techs are not affected by knowledge levels. Their only effect is to increase a Theoretical Knowledge level. At most one Knowledge level offers a 10% decrease in research time for its component in research. Given that most techs are only 30% based on theory, you would probably need 20-30 techs to break even... but by the time you researched 20 of them, either the knowledge gain from the pure theory would have decayed, or you would have maxed out your bonus anyway. Therefore, researching pure theory will probably never give you a direct advantage in research throughput. Therefore, the only time to research pure theory is if you are trying to push a tech series (e.g. nuclear techs) as far ahead of the historical year as possible. This is not generally very practical, however, due to the severe penalty for researching ahead.

Note that there are three techs, Supply Transportation, Supply Organization, and Civil Defense, that appear on the theory folder but are not theories per se. Unlike the other techs in this folder they do provide benefits other than increasing a Knowledge level. In addition they are one of the few ways to improve early game land combat experience for the majority of countries.

Historical Dates

Jumping the Gun

The days needed to finish a tech is recalculated every day. Therefore, you can "jump the gun" a little—if a tech takes 100 days to complete before the penalty for researching ahead of time, and you start it 99 days before the historical year, you will still finish after the 100th day. In other words, the penalty for researching ahead of the historical year is applied only at the end of the research. As soon as January 1 hits, the penalty for researching ahead of time is effectively retroactively removed.

As such, you can start on a new year's techs a few months (depending on other research modifiers) ahead of time without penalty.

This also means that techs which take an estimated time of more than a year as per the tooltip will finish earlier than expected.

Also, note that, for techs with multiple levels, any extra days of research beyond that needed to complete the tech are credited toward the next levels. Therefore, if such a tech is being researched ahead of historical date and is scheduled to complete in November or December, it may be worthwhile to stop the tech just before it finishes and continue it again on January 1 of the next year—you may receive a sizeable number of days credited to the next level.

Meta-Research Techs

Obviously attractive are the techs that increase Leadership or research speed. The question is, how far to research them ahead of time? It is actually surprisingly easy to get a rule of thumb for this. It boils down to this question: Assuming you want to research this tech at all, how far ahead of the historical year should you research it?

No matter how far ahead you research a tech, the cost of researching an additional year ahead is 100% of the research days it takes to research the entire tech without the time penalty.

The benefit of researching an additional year ahead is one extra year of having that tech, minus the extra time it takes to research the tech an additional year ahead of time. Given the way modifiers are calculated, you can time your research to start 2 years ahead with a -200% modifier but complete 1 Jan when the Ahead of Historic Year modifier changes to only -100% and only have to pay for the 1 year penalty (162.5 days for Mechanical Computing Machine, 175 days for Electronic Computing Machine, and 187.5 days for Education) and then have the entire year of benefit. Any positive modifiers (Knowledge modifier or already completed Tech research modifiers) will lessen this penalty. Completing after 1 Jan will lessen that benefit accordingly.

Now consider the three techs in question:

  • Mechanical Computing Machine: This gives a 2% bonus to research per level; this means that every year you research ahead of time gives you about 4.05 days of research, per point of research you have. Therefore, if you are devoting more than 40 or so leadership points to research, you should always research it as far ahead of time as possible. This is unlikely.
  • Electronic Computing Machine: Here the bonus is much larger: 5%. Therefore, if you are devoting about 18 or more leadership points to research, you should always research it as far ahead of time as possible. This is possible, but only if you have lots of leadership.
  • Education: Also 5%, but this affects all leadership! Therefore, if you have about 21 or more leadership points period, you should always research it as far ahead of time as possible.

Again, these assume no other modifiers except the Ahead of Historic Year modifier. If you have nice research modifiers and/or Knowledge levels (or plan on acquiring such), lean further toward researching ahead of time; if they are poor, lean toward not researching ahead of time. If you are a major or regional power, once major land combat gets rolling you will probably have plenty of Land Combat Experience; once it reaches the maximum bonus of 39.5% at level 20, the break-even point will drop all the way to 9.02 leadership points. Likewise, if you can maintain a high Electronic Engineering Theory, the break-even point for Electronic Computing Machine can drop as far as 8.17 Leadership devoted to research (unlikely you can keep it this far, though—more likely you'll be able to maintain a level of about 8, for a break-even point of about 13 Leadership). If the game end year is coming up, you may be better off not researching it at all.

Although these researches may pay off in the end, the time this will take increases with how far ahead you research. If you only have enough leadership/research to reach the break-even point for researching ahead of time, the tech will pay for itself after a number of years equal to the number of years you researched ahead plus one (for the base cost), counting from when you began researching it. Of course, if you have more leadership/research than the break-even point, it will pay for itself faster.

Then, there is the question of whether you should research the tech at all—unless you've modded the game, it will end for sure by the beginning of 1948. (If the game never ended, it would be worth it to research every level of these techs, although not early if your leadership/research is not high enough.) To put things into perspective, you need the following base leadership-years remaining before the end of the game by the time the tech finishes to break even (with no modifiers):

  • Mechanical Computing Machine: 22.3 leadership-years on research
  • Electronic Computing Machine: 9.59 leadership-years on research
  • Education: 10.27 leadership-years total

If you're a small country, it may not be worth researching the 1946 computing machine techs at all if the game ends in 1948. On the other hand, if you're a large country with lots of conquests, you might be able to get enough benefit from rushing techs even past the game end date to make it worth it.