This is an idea that I had while thinking about ChrTh's proposed new tech model. (Original post here.) But I've been thinking some more about it, and it seems to have wider possible applications, so I'm going to put it into its own thread. It's not a proposal, exactly; I'm only going to describe an alternate way of constructing and relating techs. It's not a new tech tree—a new arrangement of techs—but a new way of structuring and relating techs. I think it could open up easy and intuitive ways to add features that users have sometimes lobbied for.
Currently, Civ models techs as nodes-with-a-license. That is, a tech is a node in a network of other techs, such that possession of one node opens a path that lets you reach another node. Furthermore, possession of one node gives you a license to build units, buildings, etc., and/or improve terrain in certain ways. So, for example, reaching the Fishing node opens up a path that will let you reach the Pottery node while also giving you a license to build work boats.
This is a simple and intuitive structure, which is one reason it has survived unchanged from Civ 1 into Civ 4.
In the new model, though, advanced techs would be structured as composites of earlier techs; mathematically, they would be sets. So, instead of an "Ironclad" tech (say) being a node connected to the earlier techs "Steam Power" and "Metallurgy," it would be composed of them; it would be the set (Steam Power, Metallurgy). This means you wouldn't have to research three techs in order to secure the Ironclad tech; you would only have to research two. The instant you had both Steam Power and Metallurgy you would also gain Ironclad.
Now, by itself, this doesn't sound like a huge change, though you would have to make various other changes in order to implement it. (For instance, this change would make tech advances occur more quickly, and you'd need a way of slowing things down again, perhaps by constructing each tech out of many more than two components.) But there would be one further immediate consequences, as I noted in ChrTh's thread, and room for several additional evolutions.
First, techs would not be related simply by the prerequisite-successor relation, but also through the techs they share as components. This relation (a matter of degree) would naturally tend to bring certain techs into "clusters" that would tend to naturally group them together. These clusters would also behave in a way that would mimic the giving of bonuses for pursuing certain research paths. For example, let Ironclad comprise the techs Steam Power, Propeller, Iron Rolling, Advanced Shipbuilding, and Cannon-making; let Railroads = (Steam Power, Iron Rolling, Corporation); and let Flight = (Propeller, Aerodynamics, Balloons). (Reflecting the fact that there would have to be significant changes to the identities of the techs, I've contrived some examples rather than using ready-mades in Civ 4.) Ironclad and Railroads would share two components (Steam Poser, Iron Rolling); Ironclads and Flight would share one (Propeller); Flight and Railroads would share none. This would model the idea that Railroads and Ironclads are more closely related to each than either is to Flight; it would also mean that Railroads and Ironclads would be more likely to be discovered in close succession, as a great deal of the research that leads to one would be an accumulation for the other. (In fact, if the player researched the techs in the order <Cannon-making, Advanced Shipbuilding, Iron Rolling, Propeller, Corporation, Steam Power>, he would get the Ironclad and Railroad techs on the same turn.) This structural feature could be arranged so that advanced techs "cluster" in ways similar to those in the real world, which would help heighten historical verisimilitude.
At the same time, though, with a little tweaking the model would be flexible enough to easily accommodate a rich variety of "alternate" tech lines. You could, for instance, model the "vanilla" techs as simple sets, so that they are always discovered when their component techs are discovered. But you could construct certain "alternate" techs as ordered sets, meaning that these could be unlocked only if their components were discovered in a particular sequence. So, imagine a universe that contains the following techs: Steam Power, Propeller, Iron Rolling, Advanced Shipbuilding, Cannon-making, Modern Explosives, Pistons, Aerodynamics, and Balloons. Researching these in any order would give the player Ironclads, Railroads, and Flight. But the particular research order <Pistons, Propeller, Advanced Shipbuilding, Modern Explosives> could unlock a new and obscure tech: "Infernal Devices." (Infernal Devices would be that ordered set.) This tech, which the player (or the AI) need not pursue in every game, would in turn be a component in the simple sets Steam-Powered Tanks (= (Infernal Devices, Steam Power)); Armored Dirigibles (= (Infernal Devices, Balloons)); and Martian Tripods (= (Infernal Devices, Pistons)). Welcome to the world of steam punk.
Note that this structure would not run into a problem that dh_edit has noted: having to radically increase the number of techs in the game in order to accommodate branching research paths. That's because the number of sets that you can create from a domain of elements is always larger than the number of elements (except when the domain consists of only one element, of course). From just 10 initial elements you can construct 512 sets; you can construct many more than that if you are building ordered sets; and since the more advanced composite techs would be sets of techs that are themselves composites (sets of sets), the number would be ... well, quite large. Thus, a fairly limited stock of "basic" techs could support a practically unlimited number of units. The three "steam punk" techs above needed only a single "tech" (the one that creates the unit itself) and one "gateway" tech (Infernal Devices). In short: once you think up the unit, you would only need to think up a plausible set of already existing tech that would let you build it. There would be no need for extra "branches" on a tech tree.
In addition to modeling "alternate tech lines" (which would be juicy material for mods to work with, and a perfect place for designers to hide easter eggs), "ordered set techs" would also be a natural way to model Unique Units. Moreover, they would be unique units not hardwired to specific civilizations; whatever civ found or stumbled upon the ordered set would get them. For this reason—their appearance would be a function of gameplay and not something hardwired into the game—it would mean each game could be different. Sometimes Martian Tripods appear (oh noes!), but sometimes they don't.
Now, once you have the idea of techs being sets, you can extend or modify it in various ways. For instance, a tech might not only be a set of techs; it might include non-tech elements. So, for instance, the Ironclad might be the set (Steam Power, Propeller, Iron Rolling, Cannon-making, [Tool Works]), where [Tool Works] is a building. In other words, you would have the tech only so long as you had at least one Tool Works in one of your cities. If you lost all your Tool Works, you would be unable to build Ironclads, something that could easily happen in wartime even if all your ports were still in your possession. (Think of how desperate the South was during the Civil War; all those ports, from Newport News to Galveston, and they could only build a handful of ironclads.) Carried to an extreme, this kind of thing would be a natural way to mimic a dark age. If your material base degrades (as it did in the western Roman empire following the barbarian influxes), the technical knowledge can't be exercised, and widespread devastation can cascade into catastrophic losses. So, imagine that the possession of at least one Farm is necessary for some fairly basic (but still composite) tech; and imagine that during a devastating invasion you lose all your Farms. The loss would cascade upward, crippling not only the tech that Farms are a constituent of, but the techs that tech is part of, and so on up the line. You might be able to quickly repair the damage (building one farm might lead to a reverse cascade recovering all your lost tech), but if combined with worse damage elsewhere, it might knock you back by an age or two for a very long time.
Now take the idea of techs having non-techs as elements and turn it on its head: Units or buildings might also be structured as sets that take disparate items as members. A simple example with interesting implications: Janissaries might be modeled as the simple set (Musketmen [unit], Slavery [civic]). That is, the unique unit Janissaries might be generated not with a special tech but simply by turning on the Slavery civic while you have Musketmen in play. The Musketmen would automatically change, at no cost, to the new unit. That's because the difference between a slave-soldier and a conscript/volunteer is one of social organization, not technology or equipment. The difference in combat value might be minimal, but it would give the player another choice, and it would be a way of adding cosmetic variety to the game without having to restructure the tech tree in order to find a "place" for the new unit. In fact, there would be no "tech" at all for it.
Such cosmetic changes could also answer one of the more anal-retentive complaints about Civ: E.g., I build a Temple in 1500 BC and it's still there in AD 2025; what gives? There seems no satisfactory solution. At a certain point does it lose its positive effects? (That would be churlish.) Does the game give you a new version of the building with a later tech? (That clutters things up, and who has time to build everything?) But suppose buildings were also construed as sets. Just as "Janissaries" are easily added as a cosmetic change to (and maybe incremental improvement on) Musketmen, Temples could be "updated" in various ways without adding anything to the basic stock of elements. Let "Synagogue" be the set (Temple [building], Library [building], Judaism [religion]); let "Church" be the set (Temple [building], Theology [tech], Christianity [religion]); let "Starbucks" be the set (Temple [building], Free Religion [civic]). (Okay, I'm being facetious with that last suggestion.) Presto-chango, a freshened and modern city without making the tech tree any longer and without having to build anything new.
Changes parallel to those made to Musketmen and Janissaries are left as exercises to the reader.
The examples in the last two paragraph, I said, are mostly cosmetic, but more meaningful changes could be made with no added complexity. Consider factories; these are physically indistinguishable, for the most part, whether built in free market countries or centrally planned countries. In practice, they tend work much differently, though: while both will spit out industrial products, factories in free market countries tend to be a wealth-producing node while those in centrally planned economies can do little better than transform raw materials into finished goods. Civ has a hard time modeling this currently. At best, it seems it can use the Free Market civic only to increase the overall wealth of countries. But in this new model the game would have two different "factories": Factory I (Assembly Line [tech], Free Market [civic]) and Factory II (Industrialization [tech], State Property [civic]). Factory I would produce coins; Factory II would consume them. The production queue would only contain "Factory" as a building; the switch from I to II would occur by switching civics.
Or, consider any kind of unit above Warrior: it's a man with a weapon. But usually weapons have to be provided by some kind of armory, and this is a especially important when it's a unit that shoots ammunition. So, let Rifleman be the set (Warrior [unit], Rifling [tech], Armory [building]). Now imagine you lose your last Armory building during a war. Without that component, your Riflemen would instantly degrade into Warriors: men with clubs (their ammo-less rifles). The thought is easily extended to cover another recurrent player request, that Tanks (and other modern units) not be allowed to function when the civ loses its access to oil. It would seem quite easy to model. Tank* would be the set (Tank [unit], Oil [resource]), where Tank is a unit with 0 movement points and Tank* is the same unit with 2 movement points. (The production queue would simply show "Tank"; the relation between Tanks and Tank*s would be like that between Factory I and Factory II.)
As I said, many things would have to be changed in the game in order to accommodate this new model. There may be crippling theoretical problems that just haven't seen. But I've written enough for now.
EDIT: A caveat that I wrote the day after posting this. It appears in a post below, but I'm adding it here, too:
Currently, Civ models techs as nodes-with-a-license. That is, a tech is a node in a network of other techs, such that possession of one node opens a path that lets you reach another node. Furthermore, possession of one node gives you a license to build units, buildings, etc., and/or improve terrain in certain ways. So, for example, reaching the Fishing node opens up a path that will let you reach the Pottery node while also giving you a license to build work boats.
This is a simple and intuitive structure, which is one reason it has survived unchanged from Civ 1 into Civ 4.
In the new model, though, advanced techs would be structured as composites of earlier techs; mathematically, they would be sets. So, instead of an "Ironclad" tech (say) being a node connected to the earlier techs "Steam Power" and "Metallurgy," it would be composed of them; it would be the set (Steam Power, Metallurgy). This means you wouldn't have to research three techs in order to secure the Ironclad tech; you would only have to research two. The instant you had both Steam Power and Metallurgy you would also gain Ironclad.
Now, by itself, this doesn't sound like a huge change, though you would have to make various other changes in order to implement it. (For instance, this change would make tech advances occur more quickly, and you'd need a way of slowing things down again, perhaps by constructing each tech out of many more than two components.) But there would be one further immediate consequences, as I noted in ChrTh's thread, and room for several additional evolutions.
First, techs would not be related simply by the prerequisite-successor relation, but also through the techs they share as components. This relation (a matter of degree) would naturally tend to bring certain techs into "clusters" that would tend to naturally group them together. These clusters would also behave in a way that would mimic the giving of bonuses for pursuing certain research paths. For example, let Ironclad comprise the techs Steam Power, Propeller, Iron Rolling, Advanced Shipbuilding, and Cannon-making; let Railroads = (Steam Power, Iron Rolling, Corporation); and let Flight = (Propeller, Aerodynamics, Balloons). (Reflecting the fact that there would have to be significant changes to the identities of the techs, I've contrived some examples rather than using ready-mades in Civ 4.) Ironclad and Railroads would share two components (Steam Poser, Iron Rolling); Ironclads and Flight would share one (Propeller); Flight and Railroads would share none. This would model the idea that Railroads and Ironclads are more closely related to each than either is to Flight; it would also mean that Railroads and Ironclads would be more likely to be discovered in close succession, as a great deal of the research that leads to one would be an accumulation for the other. (In fact, if the player researched the techs in the order <Cannon-making, Advanced Shipbuilding, Iron Rolling, Propeller, Corporation, Steam Power>, he would get the Ironclad and Railroad techs on the same turn.) This structural feature could be arranged so that advanced techs "cluster" in ways similar to those in the real world, which would help heighten historical verisimilitude.
At the same time, though, with a little tweaking the model would be flexible enough to easily accommodate a rich variety of "alternate" tech lines. You could, for instance, model the "vanilla" techs as simple sets, so that they are always discovered when their component techs are discovered. But you could construct certain "alternate" techs as ordered sets, meaning that these could be unlocked only if their components were discovered in a particular sequence. So, imagine a universe that contains the following techs: Steam Power, Propeller, Iron Rolling, Advanced Shipbuilding, Cannon-making, Modern Explosives, Pistons, Aerodynamics, and Balloons. Researching these in any order would give the player Ironclads, Railroads, and Flight. But the particular research order <Pistons, Propeller, Advanced Shipbuilding, Modern Explosives> could unlock a new and obscure tech: "Infernal Devices." (Infernal Devices would be that ordered set.) This tech, which the player (or the AI) need not pursue in every game, would in turn be a component in the simple sets Steam-Powered Tanks (= (Infernal Devices, Steam Power)); Armored Dirigibles (= (Infernal Devices, Balloons)); and Martian Tripods (= (Infernal Devices, Pistons)). Welcome to the world of steam punk.
Note that this structure would not run into a problem that dh_edit has noted: having to radically increase the number of techs in the game in order to accommodate branching research paths. That's because the number of sets that you can create from a domain of elements is always larger than the number of elements (except when the domain consists of only one element, of course). From just 10 initial elements you can construct 512 sets; you can construct many more than that if you are building ordered sets; and since the more advanced composite techs would be sets of techs that are themselves composites (sets of sets), the number would be ... well, quite large. Thus, a fairly limited stock of "basic" techs could support a practically unlimited number of units. The three "steam punk" techs above needed only a single "tech" (the one that creates the unit itself) and one "gateway" tech (Infernal Devices). In short: once you think up the unit, you would only need to think up a plausible set of already existing tech that would let you build it. There would be no need for extra "branches" on a tech tree.
In addition to modeling "alternate tech lines" (which would be juicy material for mods to work with, and a perfect place for designers to hide easter eggs), "ordered set techs" would also be a natural way to model Unique Units. Moreover, they would be unique units not hardwired to specific civilizations; whatever civ found or stumbled upon the ordered set would get them. For this reason—their appearance would be a function of gameplay and not something hardwired into the game—it would mean each game could be different. Sometimes Martian Tripods appear (oh noes!), but sometimes they don't.
Now, once you have the idea of techs being sets, you can extend or modify it in various ways. For instance, a tech might not only be a set of techs; it might include non-tech elements. So, for instance, the Ironclad might be the set (Steam Power, Propeller, Iron Rolling, Cannon-making, [Tool Works]), where [Tool Works] is a building. In other words, you would have the tech only so long as you had at least one Tool Works in one of your cities. If you lost all your Tool Works, you would be unable to build Ironclads, something that could easily happen in wartime even if all your ports were still in your possession. (Think of how desperate the South was during the Civil War; all those ports, from Newport News to Galveston, and they could only build a handful of ironclads.) Carried to an extreme, this kind of thing would be a natural way to mimic a dark age. If your material base degrades (as it did in the western Roman empire following the barbarian influxes), the technical knowledge can't be exercised, and widespread devastation can cascade into catastrophic losses. So, imagine that the possession of at least one Farm is necessary for some fairly basic (but still composite) tech; and imagine that during a devastating invasion you lose all your Farms. The loss would cascade upward, crippling not only the tech that Farms are a constituent of, but the techs that tech is part of, and so on up the line. You might be able to quickly repair the damage (building one farm might lead to a reverse cascade recovering all your lost tech), but if combined with worse damage elsewhere, it might knock you back by an age or two for a very long time.
Now take the idea of techs having non-techs as elements and turn it on its head: Units or buildings might also be structured as sets that take disparate items as members. A simple example with interesting implications: Janissaries might be modeled as the simple set (Musketmen [unit], Slavery [civic]). That is, the unique unit Janissaries might be generated not with a special tech but simply by turning on the Slavery civic while you have Musketmen in play. The Musketmen would automatically change, at no cost, to the new unit. That's because the difference between a slave-soldier and a conscript/volunteer is one of social organization, not technology or equipment. The difference in combat value might be minimal, but it would give the player another choice, and it would be a way of adding cosmetic variety to the game without having to restructure the tech tree in order to find a "place" for the new unit. In fact, there would be no "tech" at all for it.
Such cosmetic changes could also answer one of the more anal-retentive complaints about Civ: E.g., I build a Temple in 1500 BC and it's still there in AD 2025; what gives? There seems no satisfactory solution. At a certain point does it lose its positive effects? (That would be churlish.) Does the game give you a new version of the building with a later tech? (That clutters things up, and who has time to build everything?) But suppose buildings were also construed as sets. Just as "Janissaries" are easily added as a cosmetic change to (and maybe incremental improvement on) Musketmen, Temples could be "updated" in various ways without adding anything to the basic stock of elements. Let "Synagogue" be the set (Temple [building], Library [building], Judaism [religion]); let "Church" be the set (Temple [building], Theology [tech], Christianity [religion]); let "Starbucks" be the set (Temple [building], Free Religion [civic]). (Okay, I'm being facetious with that last suggestion.) Presto-chango, a freshened and modern city without making the tech tree any longer and without having to build anything new.
Changes parallel to those made to Musketmen and Janissaries are left as exercises to the reader.
The examples in the last two paragraph, I said, are mostly cosmetic, but more meaningful changes could be made with no added complexity. Consider factories; these are physically indistinguishable, for the most part, whether built in free market countries or centrally planned countries. In practice, they tend work much differently, though: while both will spit out industrial products, factories in free market countries tend to be a wealth-producing node while those in centrally planned economies can do little better than transform raw materials into finished goods. Civ has a hard time modeling this currently. At best, it seems it can use the Free Market civic only to increase the overall wealth of countries. But in this new model the game would have two different "factories": Factory I (Assembly Line [tech], Free Market [civic]) and Factory II (Industrialization [tech], State Property [civic]). Factory I would produce coins; Factory II would consume them. The production queue would only contain "Factory" as a building; the switch from I to II would occur by switching civics.
Or, consider any kind of unit above Warrior: it's a man with a weapon. But usually weapons have to be provided by some kind of armory, and this is a especially important when it's a unit that shoots ammunition. So, let Rifleman be the set (Warrior [unit], Rifling [tech], Armory [building]). Now imagine you lose your last Armory building during a war. Without that component, your Riflemen would instantly degrade into Warriors: men with clubs (their ammo-less rifles). The thought is easily extended to cover another recurrent player request, that Tanks (and other modern units) not be allowed to function when the civ loses its access to oil. It would seem quite easy to model. Tank* would be the set (Tank [unit], Oil [resource]), where Tank is a unit with 0 movement points and Tank* is the same unit with 2 movement points. (The production queue would simply show "Tank"; the relation between Tanks and Tank*s would be like that between Factory I and Factory II.)
As I said, many things would have to be changed in the game in order to accommodate this new model. There may be crippling theoretical problems that just haven't seen. But I've written enough for now.
EDIT: A caveat that I wrote the day after posting this. It appears in a post below, but I'm adding it here, too:
Wow, I really like it! Infernal Devices FTW! 
in city until discovery of Astronomy)), you might need a lot of printouts with the game
100,000 is being generous... hunting has been around for as long as their have been predators.
