What is efficiency?

[Chiki]

I didn't want to put this in the tier list thread because efficiency is a concept that's far wider than mere tier list philosophy. Efficiency concerns the playthroughs that we all do; for example, drafts, playlogs and so on. And I am sure many efficiency players will want to know what the hell efficiency is before even doing an efficiency playthrough, so I believe that this warrants its own thread.

Efficiency isn't really a well defined term on this forum--it's some vague idea about "brisk" play (but not playing as fast as possible) and some idea about reliability, but neither of these terms are well defined. Why is this such a big deal, however? Why is it important that we define our terms well? My view is that it's impossible to have a proper debate at all without first clarifying on the terms and definitions that are used in a debate.

Consider two religious people debating about the existence of God against two atheists. One person is debating about God in the sense that God can literally violate the laws of logic and nature, as he is omnipotent and can do anything. The other person is debating about God in the sense that God cannot violate the laws of logic, and is omnipotent within the laws of logic. So the first person takes a more literal sense of God while the second one takes a more metaphorical sense. The two atheists have the exactly same views with the exact same definitions, so there is no disagreement between them.

How would this debate work out? Well, the two people would have completely different arguments against the atheists who claim that the problem of evil prevents God's existence. Perhaps the first one would argue that God can literally prevent the existence of evil and would choose not to. The second one may argue that God cannot intervene, as it would violate the laws of logic for him to do so. The two don't have any common ground despite arguing for the same concept! They would be debating against themselves rather than against the atheists. A proper debate would be impossible.

This confusion that arises from people have different ideas about concepts can happen for every single concept that can have varying meanings. And yes, it can apply to efficiency--because people have no idea on how fast to go, or how reliably they should play.

Now one may say, "We don't have any problem debating at all! There's no confusion and you're the only one who doesn't understand what we mean by efficiency!" Let's see what issues there are in debates today. For example, the Awakening tier list thread:

It's interesting now that Paralogue 4's included, the assumed number of levels gained in chapter 9 conveniently goes from two to one (with "if at all", too).

Well, despite the conveniency, 1 level is kind to Nowi in Chapter 9 considering the layout. We're not assuming babying here.

Two users have a different idea about how quickly we should play, and about how much we should punish Nowi for not being able to gain levels without spending turns. This is just a quick example: there was a lot of confusion in the tier list thread about how to tier Nowi, because a lot of different people had different ideas about what efficiency even is.

If you still aren't convinced about how important having precise definitions about terms is in debates: look under the spoiler tag for some guides on debating and the importance of definitions:

http://atheism.about.com/od/logicalarguments/a/definitions.htm

Why is discussing the nature of definitions important? If two people involved in a debate are using entirely different types of definitions for the same term, they may only end up talking past each other.

http://flynn.debating.net/austasia.htm

Definitions are an important aspect of a debate. They play a large role in determining whether a debate is focused and enjoyable, or waywardly meandering and unclear.

I hope you are convinced that it is indeed to make precise definitions on the terms we use in debating. Now we have to agree on what efficiency even is before we try to apply it to tier list threads and playthroughs--and this will start a lot of meaningless arguments if we can't even agree upon a definition for efficiency. What is efficiency?

As far as I know, though I haven't been around much to witness tier listing history, efficiency came about as a smart way to measure the usefulness of a unit. Going faster is good. Being reliable is good. Overall, it was a good development for debating in general. It changed the focus from debating units based on growths to a much more rational way of tiering. But the question is: how fast should we go, and what kind of reliability should we aim for?

We have two solutions:

1. Since we consider going fast to be good, why not go as fast as possible?

2. We can establish an arbitrary, widely-agreed upon limit for efficiency. For example, we can agree upon a certain number of turns per chapter, and a certain reliability for every action, or the set of actions.

This is left up to the reader to decide.

I personally prefer a mixture of 1 and 2. I want to go as fast as possible, since going fast is good and it really allows us to see the difference between units as much as possible. Units like the Avatar, Morgan and Sumia shine in FE13 whereas units like Sully really drop in usefulness im comparison if we don't assume going as fast as possible. But I also want to establish an arbitrary, widely-agreed upon limit for reliability. For example, a set of actions over, say, 60% reliability is acceptable while we go as quickly as possible. A 4 turn clear, the lowest possible clear while keeping reliability above a certain point, with a 61% chance is preferable over a 5 turn clear with a 90% chance. That is my view.

Now this raises an issue many have pointed out. "We don't want to go as fast as possible, because then the Avatar would go into top tier and everyone else would go into bottom tier." They argue that going as fast as possible ruins too many characters to allow us to rank units.

But going as fast as possible is just a playstyle and nothing more. There are infinitely many contexts in which you can play as fast as possible. You can play as fast as possible by using units like Sully instead of Sumia, and so on. So it does not prevent us from ranking these units highly--in fact, Sully is still ranked top tier in my tier list despite my rarely using her at all in my LTC playthrough. Every unit can be ranked just as fairly as in other tier lists.

Another counterargument to my view is that it isn't representative of how people play in general. People rarely play as fast as possible, so it isn't really an accurate way to tier units.

There are two replies I have to this. One reply I have to this is that efficiency playthroughs are even worse for this, because everyone has a personal idea of what efficiency is and that no idea of efficiency is the same, so everyone does their own playthrough but no one plays as efficiently as each other. So vague tier lists are even worse for tiering units, as they represent one person's definition of efficiency, and not a widely-agreed one.

My second reply to this is that it doesn't really matter. Tier lists were never really meant to be used by most players anyway. After all, normal playthroughs can make anyone useful--Virion and Donnel can become fantastic units with turtling, and ranking units will not matter at all.

I tried to keep my elitist language to a minimum here, so I hope that you can try to consider my point of view without immediately ignoring it as nothing more than the ravings of an "elitist."

[Paper Jam]

I would think that the ranking system of FE6 and 7, or something like it, would be a good way of defining efficiency: a (possibly arbitrary) par for each chapter's turn count, a threshold of experience to be gained in each chapter, and a threshold of funds or assets to be obtained in each chapter.

It wouldn't work so well in games like FE8 or 13, where skirmishes and xenologues make experience and funds much less of an issue than they are in other games, though.

Edited June 8, 2013 by Paper Jam

[RJWalker]

Don't compare anything to religious debates. That's just stupid.

[Chiki]

You can just change my example to anything else, such as a debate on politics, or whatever. Any concept will allow you to grasp the point of that example.

[dondon151]

the answer to this question is about as concrete as that to "what is warmth?"

Edited June 8, 2013 by dondon151

[Chiki]

the answer to this question is about as concrete as that to "what is warmth?"

1. The quality, state, or sensation of being warm; moderate and comfortable heat.

[Lord Raven]

Efficiency: the quality, state, or sensation of being efficient

You're really using "warm" in a definition of "warmth"? Then what does it mean to be warm?

[Redwall]

I think I have figured out a way of rigorously accounting for the chance of death. What do you think?

[dondon151]

obvious non-answer

pffft you really missed the point, didn't you

[Chiki]

Efficiency: the quality, state, or sensation of being efficient

You're really using "warm" in a definition of "warmth"? Then what does it mean to be warm?

I got the definition from Dictionary.com. All dictionaries have the same definition of warmth. So why can't we do the same?

Why can't we agree upon a definition of efficiency to use in all our playthroughs and tier lists, the same way that these dictionaries did? That is the purpose of this thread.

Paper Jam had the best answer in this thread so far.

I think I have figured out a way of rigorously accounting for the chance of death. What do you think?

I like it, but we have to agree on what efficiency is before even considering the details of reliability.

[Redwall]

The metric I describe is exactly what I think efficiency should be. It simultaneously accounts for chance of death, turncounts, and possibilities of slightly higher clears if the fastest clears fail due to RNG screwage.

[dondon151]

Why can't we agree upon a definition of efficiency to use in all our playthroughs and tier lists, the same way that these dictionaries did? That is the purpose of this thread.

you really missed the point

take a look at the definition of "warm" for a second.

1. having or giving out a moderate degree of heat, as perceived by the senses: a warm bath.

2. of or at a moderately high temperature; characterized by comparatively high temperature: a warm oven; a warm climate; a warm summer.

(3. a country in fire emblem: awakening)

definition 1 says "as perceived by the senses." definition 2 says "comparatively high temperature." according to the dictionary, there is no concrete definition of warmth! yet we all know what it feels like to be warm. at the same ambient temperature, with the same amount of clothing, some people will feel warmer than others.

so similarly we all know what it feels like to be efficient. trying for a concrete definition is far too limiting. if you are satisfied with the definition of "warmth" in the dictionary, you must also be satisfied with this definition of "efficiency:"

1. the state or quality of being efficient; competency in performance.

which is stating basically nothing of value.

Edited June 8, 2013 by dondon151

[Chiki]

you really missed the point

take a look at the definition of "warm" for a second.

1. having or giving out a moderate degree of heat, as perceived by the senses: a warm bath.

2. of or at a moderately high temperature; characterized by comparatively high temperature: a warm oven; a warm climate; a warm summer.

(3. a country in fire emblem: awakening)

definition 1 says "as perceived by the senses." definition 2 says "comparatively high temperature." according to the dictionary, there is no concrete definition of warmth! yet we all know what it feels like to be warm. at the same ambient temperature, with the same amount of clothing, some people will feel warmer than others.

Then let's agree and "vote" upon an arbitrary limit at which we generally do feel warm.

The metric I describe is exactly what I think efficiency should be. It simultaneously accounts for chance of death, turncounts, and possibilities of slightly higher clears if the fastest clears fail due to RNG screwage.

I'd expand your definition to not just a game over but other important things. Sumia dying in Chapter 3 is a big no, but that doesn't seem to be covered.

[Constable Reggie]

Not sure why you're using my quote as an example. I pointed out that you slyly changed the levels Nowi got in chapter 9 from 2 to 1 after you included paralogue 4. I was calling out your stupidly blatant sandbagging, not making a point about efficiency or some shit. You yourself changed the number of levels Nowi got, not me.

The irony, man. It's just baffling

Edited June 8, 2013 by Constable Reggie

[Redwall]

I'd expand your definition to not just a game over but other important things. Sumia dying in Chapter 3 is a big no, but that doesn't seem to be covered.

Already included, if you think about it.

Suppose Sumia saves 100 turns post-Chapter 3. Suppose I execute a plan for C3 that has two outcomes, each occurring at 50% probability: I complete the level in three turns with no Sumia death (50%), or I complete the level in three turns with a Sumia death (also 50%; this of course results in the loss of 100 turns post-Chapter 3).

The expected mean TC in Chapter 3 alone is three turns. The expected mean TC for the entire game is cleanly accounted for, and penalizes the player accordingly via expected turns.

Edited June 8, 2013 by Redwall

[dondon151]

Then let's agree and "vote" upon an arbitrary limit at which we generally do feel warm.

quite preposterous. i feel warm in short sleeves at 60 degrees F ambient temp. my grandmother doesn't feel warm unless she's wearing long sleeves at 80 degrees F ambient temp.

"efficiency" is one of those words whose purpose is to create an abstract definition. everyone has an idea of what it means. for me, "efficiency" is equivalent to LTC. for someone like lumi, "efficiency" is not as equivalent, but it's cutting close. it is not bad for words like these to exist. our perception on the world is not based solely on concrete examples.

Edited June 8, 2013 by dondon151

[Chiki]

quite preposterous. i feel warm in short sleeves at 60 degrees F ambient temp. my grandmother doesn't feel warm unless she's wearing long sleeves at 80 degrees F ambient temp.

"efficiency" is one of those words whose purpose is to create an abstract definition. everyone has an idea of what it means. for me, "efficiency" is equivalent to LTC. for someone like lumi, "efficiency" is not as equivalent, but it's cutting close. it is not bad for words like these to exist.

That's where democracy comes in--your grandmother's opinion would not be important in this community of teenagers and young adults. Most people here will have a similar idea of what warmth is.

The reason it's bad it's because it's just impossible to have debates if everyone has their own idea of what efficiency is. For example, LTC efficiency people can't argue against SDS efficiency people because they believe that babying Nowi has no cost, whereas the LTC efficiency people do because their idea of efficiency is different. There is no way they can debate.

The only way we can have debates and make tier lists is if we can agree upon such definitions.

[Da Bear]

Efficency is arbitrary. It all depends on who you talk to.

[XeKr]

I'll just note that people try to rank units based on efficiency. The very problem that's mentioned in this topic arises. It's subjective. Would you try to rank stuff by warmth?

Which is why I personally like criteria such as expected turn count. Calling it "efficiency" for simplicity can represent the general idea, but when tested/questioned, we have some more rigorous criteria to refer to.

Edited June 8, 2013 by XeKr

[Chiki]

So you'd go with option 2?

[dondon151]

no, we go with option 3, because options 1 and 2 are too constrictive

[Loki Laufeyson]

I always figured that Efficient Play is somewhere in between LTC and well, the funzie way. A brisk pace with even EXP allotted to the team one chooses.

So using someone like, uhh i dunno, Fiona, wouldnt count as efficient because of the amount of resources needed to make her usable.

[XeKr]

I don't like going as fast as possible because things like growths, accuracy, avoid don't matter. Bases and Move are completely overdominating. This is why I consider reliability a critical criteria.

Redwall's formula is self-consistent (disclaimer: I didn't actually verify his math) and accounts for both turncount and reliability. There is no arbitrary limit here (except for the subjective value we, as players, assign being fast and reliable)

Edited June 8, 2013 by XeKr

[Chiki]

Redwall's formula is self-consistent (disclaimer: I didn't actually verify his math) and accounts for both turncount and reliability.

His formula has to pick between 1 and 2, as well.

[XeKr]

I think it's implied that lower expected turns are better. Maybe I'm misunderstanding your point. >_>

Edited June 8, 2013 by XeKr