Is There a True, True Self?

I have compared the “true self” to the “false self” before, and while I will still stand behind the claim that the distinction can be made usefully within a certain semantic realm, I’m going to go the other direction in this post because in a different, more general realm, there is no “true self.”  As a matter of fact, if you look at it in the most general, explicit sense, you have no self at all apart from the information that constitutes your decision-making and thinking matrix.  What I’m trying to say is that when someone says that they act a certain way and that’s their “true self” and all other ways of acting are them doing something other than being their true self, they are misleading themselves.  No matter what they do, they cannot escape the fact that the same decision-making matrix, no matter how intricate or complex, caused them to act that way in each of those situations.  Now, if they mean to say that they have a preferred mode of behavior, but are forced to use a different mode of behavior in varying circumstances, well of course.  I have preferred modes of behavior, too, like I prefer to sleep or go out or play video games to doing actual work.  That doesn’t mean that I’m my true self only when I’m in the process of a preferred mode of behavior.  But that’s exactly how a lot of people reason out their reactions to, most commonly, certain other people.

I’m getting into material identity again, but since it is I suppose my preferred philosophical specialty I may as well.  Because of the fact that there is no single piece of information you can subtract from a person to make them not-that-person, the person as a whole (considered as a contiguous entity) only has meaning as far as perception will take it.  Relative to someone else, it’s their perception.  Relative to the person themselves, it’s their own perception that matters.  Imagine that you woke up and you were a different person!  Now, because of the nature of logic, this sentence has no true parseable non-tautological meaning.  I have included in the sentence that “you” are a different person, meaning you are still you.  So the Engish way to handle this issue is to change the meaning to “you wake up with a different body, probably that once belonged to someone else.” or something similar.  No matter the way you parse it in English, it isn’t handled in a logically rigorous way in the same way that we don’t answer the question “Would you like tea or coffee?” with “Yes.”  While logical, it conveys little useful conversational meaning.  Bear in mind though, that if we spoke a truly logical language, you would answer in a way that did convey conversational meaning, the same way you don’t say “Yes” in English (Although framework of asking the questions would probably receive more semantic-structural changes than the affirmative/negative response structure).

But I digress, seriously this time.  We nearly had a terminal digression there into the land of logical languages.  Back to the issue of having one identity.  The truth is that we have an assumption here that we haven’t questioned: is it necessary to treat identities in the same way that we treat physical objects?  Once again this is a conceptual piece of English- we like to treat concepts like objects.  We can pick up drawing, have an idea, find an answer, and so on.  I’m not going too far into this as a topic- I would recommend Steve Pinker’s The Stuff of Thought for more on the subject.  Anyway, the assumption that identity is an object has numerous flawed bases.  Firstly, there is 1 “person” per body, and we can count bodies.  Ergo, there must be 1 and only 1 identity per person because that person has 1 and exactly 1 body.  The next flawed idea is that identity is immutable and does not change.  That there could ever be a “one true” identity.  This isn’t even true for the lowest-level aspect of identity at the level of the physical body, so how anyone can formalize the idea that identity must be fixed is beyond me, but it does happen.  It should be completely obvious that the body of a child is different from the body of an adult, and so assuming that there is any relation beyond material continuity is a flagrant violation of logic.  Now it is not an error to say that there may exist similarities between these two identities/bodies/people, especially considering how causally connected the latter stage is from the former.  But to say that there is a fixed identity from which changes may be noted as deviations is just plain wrong.  People change a lot- people change very quickly.  Through the course of a day each of us goes through periods of high and low energy, moods, thought patterns, and who knows what.  However there are people who are guilty of the next identity fallacy, which is that somehow those aspects aren’t significant pieces of your identity.  They are passing and trivial and should be ignored because in the grand scheme of the human identity they are categorically different.  Well this is wrong, but it’s less obvious to most people because it has some deep religious roots.  The idea that the body is distinct from the soul, and that the soul is much more important than the body can ever hope to be is an old religious idea with tendrils all over the place.  The idea that something like a state of hunger contributes to your identity in any significant way is perhaps odd.  But look at it this way.  If there was a teleportation machine that destroyed your body and created one exactly like it at a different location- I have used this example before.  If there was such a machine, and it re-created your body perfectly in every detail, except it omitted recording information needed to compute and recreate a state of hunger (somewhere between total satiety and death by starvation) then is it a valid teleportation machine?  I’ll tell you what, I wouldn’t step through that bastard for a billion dollars, and not because I might be a starved corpse on the other side- it’s because I have no idea what information went into the complex computation of my own state of hunger/satiety.  Probably all kinds of things from the contents of my intestinal tract to levels of certain hormones and neurotransmitters.  If the machine omits all that information, I don’t come out the other side of that teleporter.  Someone else does.

So I am aware that I have a difficult position to defend here.  I’m saying, at the same time, that there is an immense degree of flexibility in what constitutes a person- that you can still be “you” in the sense that counts from the time that you’re a child until the day you die, but at the same time the standard for building a teleporter must be absolutely flawlessly perfect in order to preserve material identity.  The reason for this is that I’m making the two comparisons based on different criteria.  I’m a strict materialist- everything can be reduced to an arrangement of matter and energy if a sufficient level of detail and fidelity is used.  However, matter and energy in and of themselves are just rocks and colored lights- they have to be organized into information patterns to be interesting.  So in the case of a stardard human life, without being teleported, the information pattern persists in direct fashion through space and time and can be identified perfectly as being materially continuous.  However, once you introduce the ability to jump around in space and time, you have to get a little bit smarter than that in order to maintain material continuity.  To think about material continuity, I’ll call it the Where’s Waldo? Effect.  If it’s possible to look into the universe like a giant, four-dimensional Where’s Waldo book (including all periods of time) and find you, or any given person, then you have material continuity.  When you introduce the ability to jump around in space, then you need to have the end of one string and the beginning of another match to a sufficient level of detail that the four-dimensionally-conscious being looking into the Where’s Waldo Universe can put together the pieces.  The same thing is true if you’re jumping through time, of course, but most conceptualizations of time travel account for perfect material transport as a matter of course, so it’s not as interesting to talk about.  Still, if you have a time machine then you necessarily have created a teleportation device because you could teleport back in time exactly enough time to go wherever you’re going and then go there, arriving at exactly when you left.  Not a super elegant mode of teleportation, but quite effective in physical and relativistic terms.

In fact, to be even more technically precise, it’s impossible to build a teleporter without somehow cheating relativity.  The modern idea on how this might be done is taking advantage of quantum entanglement to transfer information instantaneously to anywhere in the universe- it might also be done with some form of tachyon particle but entanglement shows much more promise.  It’s something of an important idea that material identity is both time and space independent because even if you could transfer the totality of your information instantaneously anywhere, I find it unlikely that it’s possible to instantly create a new body for you on demand.  As long as a more or less perfect copy gets made (ideally before you get “re-activated”) it makes no difference if you lost some time in the middle.  The real question is- how perfect does this copy have to be?  That is an extraordinarily difficult question to answer.  I have no idea how you would go about answering it in a mathematical sense.  As long as you have material continuity to fall back on then you have nearly endless flexibility, but the second that gets taken away it really becomes a question of what you believe the limit is.  And a strange sort of “are you feeling lucky, punk?” kind of attitude.  It’s the same operation, because material continuity is just using the super-perfect teleport trick over impossibly small distances and over the smallest possible time lengths (Planck time, approx 10^-44 seconds) using the same medium that the stability of the information pattern itself is composed of, so the accuracy is so absolute as to be perfect.  Sure, particles jitter and all sorts of other stuff is going on, but that’s the nature of the pattern that you’re made of anyway.  Even in periods of the most rapid change you can conceive of, relative to the length of a single Planck time- I mean, come on.

I don’t think that 10^-44 seconds will even fit into the human mind as a workable unit of time.  That means that you would need 1 followed by 44 zeroes of them in order to get one single second.  To put that into perspective, if you had that many nanoseconds the total length would be 3×10^27 years, or enough to contain the entire history of the universe (15 billion years) over 200,000,000,000,000,000 times.  A Planck time is small.  There is no practical way that sufficient change to break material identity could happen on a timescale so small.  So I just say that no matter what, material continuity equals material identity.  It’s not strictly true, but if you’re seriously in doubt then you must be talking about some thought-experiment edge case like “what if we had a particle accelerator that could destroy n brain cells in exactly 1 Planck time, how many would we have to destroy…”.  They’re awesome, and I do it all the time, so that’s great, but as a rule of thumb I think the idea of material continuity = material identity works quite well.

Strategy, Tactics, and Games

First of all, read this post.  Now.  It is pure genius.

After you’ve done that, I have analysis to do.  I’m not going to regurgitate a single shred of the information in the above article because I have too much to say.

First of all, the author Venkatesh Rao is absolutely correct, and not only did this idea never occur to me, I never thought to question the idea that the fundamental assumptions used in the creation of strategies and tactics were fundamentally flawed- adding a level of meta-tactical formulation that is essentially lacking in most decision-making.  Now, more specifically, the idea that tactics are general and strategic thinking is unique to situations, while it appears to be generally true, and it’s a much better approximation than the old model that strategy is somehow more all-encompassing than tactics, it falls victim to the same thinking that the old model did.

What do I mean by this?  Well, strategy by this definition does actually include tactics necessarily.  Because it’s constructed for an individual circumstance it must necessarily be built up from the different tactical options available to the agent.  However, tactics do not necessarily have to be a part of a grander or lesser strategy.  A tactic can be described in pure game-theoretical terms without any real-world interaction.  This is accomplished by building a tactic up from axioms in a way that strategies derived from doctrines aren’t.  A doctrine is an assumption about the world for practical purposes and is therefore derived from experience in an inductive fashion- as a practical assumption which is most often true, or otherwise useful to assume.  Tactics derived from axioms are arrived at deductively.  For example, in a military situation, we know that we want to destroy as much enemy materiel as possible while incurring as few losses as we can.  This is not a doctrine- this is an axiom.  Similar axioms are such assumptions as “guns have range” or “guns are highly lethal to humans.”  So if we build up a number of axioms like this we can arrive at a situation where we have whatever weapons in whatever known situation, and we can compute tactics such as have troops use cover, use infantry with anti-armor weapons to engage enemy tanks, use tanks to engage enemy assault infantry, etc. etc.  So maybe we arrive at an effective tactic of creating a formation with the tanks in the front, and a large number of infantry in a supporting role, to be brought forward when the enemy fields their tanks.  It’s important to note that we can change these parameters however we like and we’ll arrive at different tactical results.  For example, if we changed the situation to include the axiom that all infantry are highly effective at killing tanks, then it may not be worthwhile to field tanks at all because they would be destroyed too easily, and it certainly wouldn’t be a good idea to have them go first if they were all you had.

In a strategic sense, we have a different way of looking at our available units.  We could talk about units in the same abstract sense as before and still come up with concepts of strategic interest, but in order to formulate a valid strategy we would really need to know the specifics of what we’re dealing with.  Do we have 122 tanks and 300,000 troops to call upon?  What’s the supply situation, what about morale, training, enemy targets available, etc. etc.  From this we might formulate a diverse array of potential strategies to maximize the effectiveness of the resources available.  However, in order to do that we need to have both good doctrine, or practical assumptions about the nature of the world, and good intel, or exact specifics about the situation at hand.  The difference is fairly easy to handle.  If we know that setting the tempo of the military engagement is critical, that’s a doctrine.  It has direct strategic significance by reducing the infinite field of possible strategies down to a more manageable number of probably useful ones very quickly.  Intel would be “the enemy has 513,889 soldiers located in that city” or “the enemy is going to attack in three days.”  Intel is necessary for making operational decisions, or low-level instance decisions.  I suppose it could be said that operations are simply a lower-level form of strategy, but they’re low enough level that it is practical to consider them fundamentally different.  Strategic thinking is necessary to make them work, as opposed to abstract tactical deduction, but the strategy selected is known and an implementation is all that is required.

Strategic thinking is not, as I and many others once thought, “higher level” than tactical thinking.  I would argue that it requires more experience and more intelligence to think strategically in a given field than to analyze it tactically.  With strategy, you are necessarily dealing with imperfect information and chance.  Chess is a game of pure tactics, with very little true strategy.  I would argue that more complex games like Go actually do include levels of strategic thinking because you have to address the board at hand and your opponent in a unique fashion.  However, in chess, you don’t care who your opponent is or what the individual situation is.  Given a sufficiently advanced derivational strategy you could compute the ideal move in a given situation.  The same thing could be said for Go, of course, but the computational capacity required is so immense that it is utterly impossible with the resources of a human brain.  However, chess masters make this sort of analysis when deciding what to do.  Ah, who cares about individual games.

Real time strategy games tend to contain strategy, with a fairly sparse diversity of individual tactics.  Some tactics that are generally common in all RTS games are things like rushing, turtling, spamming, and so on.  Strategically, however, you have to look at the terrain and what units your opponent is fielding and make a decision that will only hold for this specific situation.  One of the main flaws in RTS games in my book is that maps tend to play out the same way each time because the terrain has too little effect.  This sounds like I’ve got it backwards, but bear with me.  Two armies meeting in a field with no terrain at all have very few factors to make strategic decisions on.  Barring some really different logistical or technological factor, the battle will probably play out much the same way every time you ran such a simulation.  Now, if you added in a little terrain, just enough to create a few significant areas of strategic significance, then the nature of the game changes.  Both sides try to hold the same strategic areas, and succeed to the degree of the resources available and the ease with which they can hold a specific area (if it’s closer to them, etc).  However these battles will also play out the same way every time because there aren’t enough options.  If you’ve only got a few points of obvious interest to both sides then they’ll fight over them every time.  The tactics utilized to obtain them may be different, but the strategic objectives are not up for negotiation.  In order to have a strategically interesting game there must be a greater number of possible strategic choices than a given side can hope to capitalize on.  What do I mean by this?  If we increase the number of points of strategic significance, up to the point where it is no longer an option to simply take them all, then the game starts to become strategically interesting in the sense that different players will make different strategic choices on the grand scale.  Now, I have to mention here, that it is also important to have multiple dimensions of possible choice.  If you have a wide selection of areas which will all give you resources, then the strategy doesn’t actually change.  You just have to get as many of them as possible- and the order that you take them becomes the individual strategy and doesn’t make an interesting strategic setting.  Perhaps the best way to create strategic significance is to give the players the ability to create strategic weapons, and depending on where they place them, the course of the battle changes.  The issue with this method though is that a given setup will lend itself to specific places to put such weapons.  So if you put these choices in the players’ hands, they’ll quickly settle on where the best choice is and just repeatedly place there.

I am trying to bring to light the principle of strategic consolidation.  This is known in game theory as Nash equilibria.  Ideally, in order to create a strategically interesting situation, you would ideally make it so that there are no Nash equilibrium for your setup.  However this in almost an impossible task.  So instead you can set about creating as many of them in as complex a formulation as possible so that it doesn’t play out the same way too often.  I would posit that there must be a way to create a game which, from its fundamental structure, will be strategically interesting every time.

Now how would we go about doing this?  The first point is we must somehow factor in the right level of extra-structural and intra-structural factors.  Meaning, the map, player choices, and other circumstantial factors must have a variable level of influence, but not so variable that any one of them can ever break the game.  Of course, it would always be possible to create a map which breaks strategic interest, or for a player to be outright retarded.  However we as the hypothetical game designers get to put certain parameters on these things.  For example, maps should be between X and Y size with properties A, B, and C, yada yada yada.  We will only make a game that is always strategically interesting if our input parameters are followed.  We will also assume that all players will be trying to win, although we have to allow for disparate skill levels.  That said, because we’re trying to make a strategic game, if we’re doing our job right then better players will straight up destroy worse players.  This is acceptable because we can keep the game strategically interesting by always introducing a flaw in any given strategy chosen that the other player might exploit, except that they might not be skilled enough to.

Alright, now we begin in earnest.  Because we want our game to be strategically interesting, we need a large diversity of points of interest, which necessarily entails a map of a certain size.  As a result, we will have to scale our unit balance accordingly.  Ideally we would have bigger maps = better, but then we run into the issue of time limitations.  Games need to be limited to a certain time frame, or nobody will ever finish them and they won’t be fun.  We could get around this in a number of ways, such as having games run in phases or have a perpetual game, or maybe run it in turns, etc. etc.  However all of these will curtail the structure of the game in a significant way.  So instead we’re just not going to worry about time being an issue.  Our theoretical game won’t account for the players having fun in any realm outside of the actual strategy of the game.  For example, we will not concern ourselves with the processing power required to run it, the graphics, the cost of the computer, or the market share of people who might be interested in buying such a game.  So we will have maps that are exceedingly large with lots of different points of interest such as geographic features, resources, and perhaps even significant locations such as cities.  Regarding our resource model- we want it to be simple enough that the player doesn’t have to break their brain in order to get units to play around with, but we also need it to be extremely important.  The ability to reduce the opponent’s ability to fight is a fundamental and necessary strategic concern.  As an aside, in order to have a diverse array of points of interest, we might cheat and have a massive variety of resources.  This is effective to a point.  I don’t know what the ideal number would be, but certainly 100 is far too many.  I would be leery of anything upwards of 10 or 20, and in order to have numbers that high it would need to be necessary to be able to convert them conveniently (at a price, possibly substantial).  The other important issue is logistics.  Most modern strategy games ignore them because they are something of a pain.  However I am confident that it is possible to implement a logistics system that the player doesn’t have to worry about except in the sense that they keenly feel the need to protect it, and to attack the enemy’s.  The player should never have to give orders to manually maximize the efficiency of their logistics systems.  The player is for making strategic and tactical decisions, not daily maintenance.  If they were so inclined they should be able to change whatever they wanted, but a liberal dose of heavily customizable helper AI would do RTS games a great deal of good.  Similarly, the player should be in a position to decide what gets produced, but should not have to manually queue up individual buildings and units.  Using a flexible template system complemented with artificial intelligence would be fantastic.  The player can say “I want a firebase built here.” and the servitor AI summoned will see to it that the location in question has whatever buildings the player associated with a firebase are built there.

In a similar vein, the player should never be called upon to give orders to individual units.  This is a critical point.  The UI built on top of the basic unit level should be sophisticated enough that the player can quickly and easily pick out whatever units they want, organize them automatically into squads, order squads or companies, battalions, armies, whatever to be built and assembled automatically, and have those units automatically organized for them.  If iTunes can do it with massive libraries of mp3 files then an RTS game can do it with units.  Complex reports and commands should be routine.  The player should be able to get a complete breakdown of whatever subsection of units they like, according to whatever criteria they like.  For example, I might ask my war machine AI to give me a complete breakdown of my air force.  It will show me a page saying I have a total of 344,000 planes and then a breakdown by grouping, role, and further breakdown by type, with individual conditions and orders should I ask.  I should be able to look at a procedurally generated map showing what I have where and what they’re currently doing.  Regarding complex commands, it should be possible for the game to understand more complex elements than “move” and “fire.”  For example, if I want to mount a sustained bombing run on an enemy base, it’s not a complex task.  I just want to get a whole lot of bombers and have them kill everything in this here area while returning to base/aircraft carrier for fuel and ammo when necessary.  The player absolutely should not be required to designate every single target for every single bomber, and then manually order them to return.  It should definitely be an option to order specific units to destroy a specific target, but a more abstracted and powerful UI solution would be much better.  For example, I might designate a specific area as an enemy base which I label “southwestern air staging base” or whatever.  Having the game automatically divide the map into sectors would be handy too.  Being able to then draw symbols and regions on this fabric that you can order units around with would be fantastic.  Anyway, I can then designate specific enemy targets within that area with different values depending on how badly I want those targets destroyed.  I might even create an algorithm describing a way to automatically determine which targets I want destroyed more, such as always aiming for factories or artillery pieces or whatever else.  Then when I order a sustained bombing run, my bombers do what I want them to even when I didn’t specifically order them to.  I can go do something else without having to micromanage.  I guess that’s the whole point of this paragraph.  The age of micromanagement is over.  Hopefully future RTS games will realize this, and we will look back on the RTS games of today as basically RPG games with more units.

To go further into what abstraction might do for our strategy game, RTS games need to start having operations.  By operations, I mean a large, coordinated plan with many active elements all going together, which the player could give specific names if they wanted to.  Including specific objectives as conditionals would be fantastic.  For example, if a player defined an objective as “blow this up” then your AI will understand that if the offending enemy is destroyed, that statement will return true.  The player could then have a breakdown by operation to see how they’re going in all their operations at once.  Your operation readout might be:

Operation FIrestorm – In Progress
• 5:11 of planned 14 minutes elapsed.
• 4 of 11 objectives completed
• General force strength 87%”
– notes
• massed assault eastward on sectors B65 through B88
Operation Lightning Spear (covert) – In Progress
• Jammers operational
• Cloaking operational
• believed to be undetected
• 1:30 of planned 7 min 35 seconds elapsed
• 1 of 5 objectives completed
• 100% General force strength

I am aware that none of this seems like it has any bearing on how to make a game that stays strategically interesting.  It seems to me that the main stumbling block for RTS games today is the user interface.  They are just not suited to having a really strategy-oriented game.  The player has to do too much.  While this increases the twitch factor- not necessarily a bad thing, it detracts from the ability to create large and sweeping, grand strategies.  Using groupings to combine individuals into squads, squads into companies, companies into battalions, and battalions into armies would be a huge improvement.  Doing it atomically allows a computer to easily construct the desired units based on input from the player.  For example, I design a squad of 20 soldiers and give 2 of them machine guns and everyone has grenades.  I then say give me a company with 13 of those squads, 3 units of 3 tanks apiece, 1 unit of 3 anti-air vehicles, 2 units of snipers, and 1 command squad unit.  I’ll put 30 of those companies into a battalion, of which I would like you to build one at this base, one at this base way over here, and another at this third base.  Automation is the name of the game, to free the player up for making the decisions that really count.