[TechTree] [0.23.5] Ackander's Vertical Tech Tree - Release v1.16 - May 13, 2014

[Ackander]

Oh wow, that's a lot more science than I thought it was... I admit I haven't been playing around with the science stuff of mods outside the Kerbin SOI so I didn't think it would be that much. Then maybe I would say that there should be a scaling factor of costs involved... Since you already have the increase in total science for each mod maybe you could start with that, although I wouldn't use the full amount since I think people might not make the full use of science from mods simply due to ignorance. So maybe 75% of the increase? So if you have SCANSat it would cost ~36% more in total.

In the end, if people don't like the scale factor they could remove it right?

Righto. I will try to make the scaling as transparent as possible, with a section for user 'intervention'.

Right now I am thinking each node should have a relative cost, and factors such as part count in nodes and science adding mods will factor the costs, hopefully appropriately. I will put up what I have done so far on github. It is only the tree.cfg and the excel 'program' to combine two tree.cfgs. My spellinggrammer is stll not fixed yet, sorry Barcharcraz.

[likke_A_boss]

Also, How much science can the stain science earn? Probably more then everything else.

[Nelien]

My spellinggrammer is stll not fixed yet, sorry Barcharcraz.

Unacceptable. I'm abandoning this mod now.

[Ackander]

Also, How much science can the stain science earn? Probably more then everything else.

Station science... only 3,020,800...

Again, this assumes every experiment can be ran at/on every celestial body.

[likke_A_boss]

Station science... only 3,020,800...

Again, this assumes every experiment can be ran at/on every celestial body.

However, only a fraction of that would be obtainable, maybe 1,500,000? Heavy parts.

[Nelien]

Yeah, station science isn't really something done very commonly. And if you're doing station science on loads of different planets it's safe to assume most of the tech tree is already unlocked.

But hey, I've never visited another planet yet so what do I know!

[Ackander]

However, only a fraction of that would be obtainable, maybe 1,500,000? Heavy parts.

Ah, yes. I was just looking up to see what the transmission efficiency is. This is something I should consider too when I calculate total science.

Also, I updated the values for stock experiments and total science cap (disregarding transmission efficiency) is now 64,503.8. Using only stock science.. getting more than a third of all science in the kerbolar system seems excessive, and over all probably most of the transmittable science (don't have it figured yet, soon though). Expect to see further science cost reductions.

[likke_A_boss]

Is there possibly a dev build anywhere where we can get the las test version? (Science-wise, mods-wise ect.)?

[fronbow]

Is there possibly a dev build anywhere where we can get the las test version? (Science-wise, mods-wise ect.)?

Have you had a look at the github repository? That should be the most up to date, and I imagine when he gets the time, Ackander will look into versioning the various changes that are made to the tree!

[Ackander]

Is there possibly a dev build anywhere where we can get the las test version? (Science-wise, mods-wise ect.)?

Sorry it is taking so long. I am not happy with my progress, but it is progressing. I keep finding bugs and other things it needs to do in the excel program. And now I am trying to re-sort the nuclear discipline, but every time I load KSP there is some problem or another. Last time it was not loading saves, time before that it was crashing from a vessel duping a couple thousand times in the persistant.sfs file.

On github is the tree so far. Just still need to do nuclear engines and reactors and spelling... I know..

So for now, there is a static tree to play with. Not much has changed, to be honest. I added a few more mods and seperated out a few that could be fit into 'projects'.

A lot of the Wayland ships are grouped together. The Tiberdyne Space Shuttle stuff are in two groups, Bobcat's historic packs are seperated by spacecraft, Space X Launch pack as a few projects, B9 is seperated by fuselage system, so there are projects for Mk2, HL, and S2 systems.

The plan is to eventually rework the main tree slightly to be a little more simple, and expand the projects to have categories within the project, like command pod, sas and misc equipment, liquid engines, boosters, fuselages, etc..

I am nearly almost there with the Science Re-Coster excel sheet, just need to get the nukes all lined out first.. Must be some bad mod interactions, is my guess, because stock and secondary part adding versions of KSP work just fine, just my primary part adding version with the 70 some mods and plugins.

[Nelien]

Sorry it is taking so long. I am not happy with my progress, but it is progressing. I keep finding bugs and other things it needs to do in the excel program. And now I am trying to re-sort the nuclear discipline, but every time I load KSP there is some problem or another. Last time it was not loading saves, time before that it was crashing from a vessel duping a couple thousand times in the persistant.sfs file.

On github is the tree so far. Just still need to do nuclear engines and reactors and spelling... I know..

So for now, there is a static tree to play with. Not much has changed, to be honest. I added a few more mods and seperated out a few that could be fit into 'projects'.

A lot of the Wayland ships are grouped together. The Tiberdyne Space Shuttle stuff are in two groups, Bobcat's historic packs are seperated by spacecraft, Space X Launch pack as a few projects, B9 is seperated by fuselage system, so there are projects for Mk2, HL, and S2 systems.

The plan is to eventually rework the main tree slightly to be a little more simple, and expand the projects to have categories within the project, like command pod, sas and misc equipment, liquid engines, boosters, fuselages, etc..

I am nearly almost there with the Science Re-Coster excel sheet, just need to get the nukes all lined out first.. Must be some bad mod interactions, is my guess, because stock and secondary part adding versions of KSP work just fine, just my primary part adding version with the 70 some mods and plugins.

A vessel duping could have to do with RemoteTech, I know it has a bug that causes ships to duplicate (I had to remove it from my game until that's fixed).

I really like what you've done so far with the tree. I checked out the new version in a test save and it looks very solid. I'm not sure what to think about the project based separation of parts though, at first I felt like it didn't really fit a tech tree to have research based on a single ship rather than on a discipline, but it does seem to be mainly for parts that are very specific and it might work out pretty well. I'll have to see how it feels once it's properly done.

[Ackander]

A vessel duping could have to do with RemoteTech, I know it has a bug that causes ships to duplicate (I had to remove it from my game until that's fixed).

I really like what you've done so far with the tree. I checked out the new version in a test save and it looks very solid. I'm not sure what to think about the project based separation of parts though, at first I felt like it didn't really fit a tech tree to have research based on a single ship rather than on a discipline, but it does seem to be mainly for parts that are very specific and it might work out pretty well. I'll have to see how it feels once it's properly done.

I know what you mean. I had my reservations about doing it this way, believe me. The reason I did it this way though, is because of how the intent for the parts in a project are to, for the most part, fit together to make a whole, not unlike the development of aircraft/spacecraft today; much of the the components are designed for the specific vehicle, and not neccesarily standardized, for example the PW F119 engine was developed for the F22 program, and a varient F135 for the F35, or the Dragon capsule system for Falcon 9 rockets.

I do not like leaving all the parts in one node though. I want to seperate parts based on discipline, like aero, structures, propulsion, and control, each node requiring the development of lesser technologies in other disciplines. An advantage would be the seperation of mod specific 'systems' from the main tree, like B9 fuselage systems. If you run with B9, you will have to pay more science to get those systems, while without B9, you will not have to pay for parts you do not have loaded.

A disadvantage to this plan is how many nodes and connects would be required to 'properly' seperate the projects. I do not know if TreeEdit/Loader has a limit on nodes or not, just that the limited space could definitely get pretty crowded pretty quickly with nodes and connections. Today I am going to test node positioning using my excel sheet to see how it looks and how much space is freed up.

[Nelien]

I know what you mean. I had my reservations about doing it this way, believe me. The reason I did it this way though, is because of how the intent for the parts in a project are to, for the most part, fit together to make a whole, not unlike the development of aircraft/spacecraft today; much of the the components are designed for the specific vehicle, and not neccesarily standardized, for example the PW F119 engine was developed for the F22 program, and a varient F135 for the F35, or the Dragon capsule system for Falcon 9 rockets.

I do not like leaving all the parts in one node though. I want to seperate parts based on discipline, like aero, structures, propulsion, and control, each node requiring the development of lesser technologies in other disciplines. An advantage would be the seperation of mod specific 'systems' from the main tree, like B9 fuselage systems. If you run with B9, you will have to pay more science to get those systems, while without B9, you will not have to pay for parts you do not have loaded.

A disadvantage to this plan is how many nodes and connects would be required to 'properly' seperate the projects. I do not know if TreeEdit/Loader has a limit on nodes or not, just that the limited space could definitely get pretty crowded pretty quickly with nodes and connections. Today I am going to test node positioning using my excel sheet to see how it looks and how much space is freed up.

I think you're absolutely right, that's probably the best way to deal with so many parts from so many mods. I played around with a "cheat" save where I just gave me tons of science points by save editing and then pretended to progress the tree as if I would normally (much faster obviously since I already had the science points), and it actually felt rather natural to reach those projects (I didn't have all the mods, but I had some for B9 and I don't remember what else). The parts in there I would likely never use unless I use them together. This also helps with the problem of NOT having certain mods right? Since those entire trees can be cut off if someone lacks that specific mod.

I can't imagine how much work this must be, so please ask for help if you need it!

I'm still going to be playing with the 1.12 version since I rather test your tree-generator-excel-sheet-based-on-installed-mods before I run with 1.15. Just too much work to trim the tree manually .

[Ackander]

I think you're absolutely right, that's probably the best way to deal with so many parts from so many mods. I played around with a "cheat" save where I just gave me tons of science points by save editing and then pretended to progress the tree as if I would normally (much faster obviously since I already had the science points), and it actually felt rather natural to reach those projects (I didn't have all the mods, but I had some for B9 and I don't remember what else). The parts in there I would likely never use unless I use them together. This also helps with the problem of NOT having certain mods right? Since those entire trees can be cut off if someone lacks that specific mod.

I can't imagine how much work this must be, so please ask for help if you need it!

I'm still going to be playing with the 1.12 version since I rather test your tree-generator-excel-sheet-based-on-installed-mods before I run with 1.15. Just too much work to trim the tree manually .

Fair enough. What I need help with right now is developing an algorithm for changing the price of a node based on parts in the node, and mods installed that give extra science. Coming up with a way to 'rate' the science mods themselves is tough, because some give so much extra science, and I am not familiar with how the mod works, so I can only guess at the science value of having those mods.

For now, the placeholder is cost₀*10% + partsratio * (.9*cost₀+modscienceadd), so the cost of the node is at least 10% of the original cost if it is empty, unless it is a terminal node in which case it will cost 0 science, since it does not lead to other nodes.

modscienceadd is, for now, an arbitrary percent of the transmittable science for the Kerbin System (Kerbin, Mun, Minmus) plus another arbitrary percentage of the transmittable science for the rest of the Kerbol System the mod adds to KSP. Doing it this way leads to huge increases in costs, by as much as 480,000 science for station science. I have started to play with numbers representing difficulty in obtaining said science based on mass of experiment, and what regions of a celestial body the experiments can be conducted. Obviously, higher mass is more difficult. Easiest is near and deep space, and none-biome ground testing a little more difficult, and then biome ground testing and then atmospheric science being the most difficult to obtain.

[Nelien]

Fair enough. What I need help with right now is developing an algorithm for changing the price of a node based on parts in the node, and mods installed that give extra science. Coming up with a way to 'rate' the science mods themselves is tough, because some give so much extra science, and I am not familiar with how the mod works, so I can only guess at the science value of having those mods.

For now, the placeholder is cost₀*10% + partsratio * (.9*cost₀+modscienceadd), so the cost of the node is at least 10% of the original cost if it is empty, unless it is a terminal node in which case it will cost 0 science, since it does not lead to other nodes.

modscienceadd is, for now, an arbitrary percent of the transmittable science for the Kerbin System (Kerbin, Mun, Minmus) plus another arbitrary percentage of the transmittable science for the rest of the Kerbol System the mod adds to KSP. Doing it this way leads to huge increases in costs, by as much as 480,000 science for station science. I have started to play with numbers representing difficulty in obtaining said science based on mass of experiment, and what regions of a celestial body the experiments can be conducted. Obviously, higher mass is more difficult. Easiest is near and deep space, and none-biome ground testing a little more difficult, and then biome ground testing and then atmospheric science being the most difficult to obtain.

I think you're trying to take too many factors into account when calculating the cost. I know this is not much help but I suggest keeping it simple to start with, maybe start by clamping the value to a fixed minimum/maximum and then try to find a good curve in between.

In one of my own mods I calculate the "build time" of a vessel simply by a simple square root function: build_time_in_seconds = sqrt(mass) * 30000. I know, it's stupidly simple, but it works surprisingly well. Maybe you should try squaring the "modsciecneadd" value? It should give you a more equalised curve avoiding increadibly high or low values.

I'll think about it some more and let you know if I come up with a good solution.

[Ackander]

I think you're trying to take too many factors into account when calculating the cost. I know this is not much help but I suggest keeping it simple to start with, maybe start by clamping the value to a fixed minimum/maximum and then try to find a good curve in between.

In one of my own mods I calculate the "build time" of a vessel simply by a simple square root function: build_time_in_seconds = sqrt(mass) * 30000. I know, it's stupidly simple, but it works surprisingly well. Maybe you should try squaring the "modsciecneadd" value? It should give you a more equalised curve avoiding increadibly high or low values.

I'll think about it some more and let you know if I come up with a good solution.

Perhaps. I get somewhat detailed when I get obsessed with something like this that uses excel sheets for some reason.

I went ahead and read Station Science's forum post and played around with it for a couple hours, and I think I know what to do. I will never be happy with an unobtainable conversion curve, so I just have to pick what feels right. So most of the science requires the orbital lab and the cyclotron can only operate in space, so I can ignore science gained on the surface for the most part. Same with L-Tek. So I am pretty close to finishing now. I keep thinking I am almost done though, and then think of something to add or have to fix, so no promises. I can almost promise adding new mods to the tree will be much quicker after this with one of the excel tools I made.

If anyone knows a way to make the research and development area bigger, I would sure like to know. For what I have planned, I could probably use 4 times the area already provided. But mostly they are just vague notions and images floating around my head, and will take some time to flesh out.

[likke_A_boss]

Bug Report: Nothing is currently included in the BASIC ROBOTICS node. Neither github, nor my game, with infernal robotics.

[Ackander]

Bug Report: Nothing is currently included in the BASIC ROBOTICS node. Neither github, nor my game, with infernal robotics.

Aye. The only thing in that node should be a couple of parts from DROMOMAN and the robotic arms pack.. but I keep forgetting to fix the robotic arms not loading, and thus not getting into the tree. I deactivated the Lazor System plug in because of the UI overlapping in R&D screen, so the parts do not load. I will try to remember to add them back in tomorrow. Maybe I will look at IR to and re sort those to get some parts into basic robotcs too. Thanks for reminding me.

[DoctorJon]

Would it be possible to add in support for 6S Service Compartment Tubes?

[Ackander]

Would it be possible to add in support for 6S Service Compartment Tubes?

Are not those parts a great idea? Yes, they are in there. The version on github already has them in it, but I am about ready to post on github the current prerelease.. still adding a node or two, along with Station Science. I will post again when I refresh github.

[Ackander]

Vertical TechTree v1.15.45.c.pre is available on github.. I shall include a link in the description shortly.

[Nelien]

Vertical TechTree v1.15.45.c.pre is available on github.. I shall include a link in the description shortly.

So, I had a look at the updated xls sheets and I've got to say, it's a bit confusing. I was probably wrong, but I was expecting something that allowed you to simply "checkbox" which mods you have installed of the ones supported and it would generate the tree.cfg based on that. If that's the case then I can't figure out how to use them.

[Ackander]

So, I had a look at the updated xls sheets and I've got to say, it's a bit confusing. I was probably wrong, but I was expecting something that allowed you to simply "checkbox" which mods you have installed of the ones supported and it would generate the tree.cfg based on that. If that's the case then I can't figure out how to use them.

Yeah, the dynamic tree.cfg maker is not done yet still, just the mess you see there on github. But when it is done, you will only have to copy and paste the text from tree.cfg into the input box ( should mention needing TreeEdit at this point, for you need to save the tech tree locally with TreeEdit for it to eliminate all the parts that are not installed, having a list of every part for every mod would have been too cumbersome to manage, this way was easier for me..), and then the output box (not there yet) will give text to paste back into notepad++, and from there you have to replace CRLF with \r\n, and CRTB with \t in extended mode.. probably forgot to mention that someplace. But the excel sheet will count the parts in each node and see if you have any of the science mods installed, calculate what the cost should be based on what parts are showing up in your new tree.cfg saved, and then spit out a the text for a new tree.cfg.

It does sound kind of like something of a process.. but it was the only way I could think of in excel that would count the number of parts in each node and compare it to a list of how many parts are usually in the nodes in the release and then give you a new version of tree.cfg. So lots of user work, but the result is ultimately better than the static tree.cfg.

[Nelien]

Yeah, the dynamic tree.cfg maker is not done yet still, just the mess you see there on github. But when it is done, you will only have to copy and paste the text from tree.cfg into the input box ( should mention needing TreeEdit at this point, for you need to save the tech tree locally with TreeEdit for it to eliminate all the parts that are not installed, having a list of every part for every mod would have been too cumbersome to manage, this way was easier for me..), and then the output box (not there yet) will give text to paste back into notepad++, and from there you have to replace CRLF with \r\n, and CRTB with \t in extended mode.. probably forgot to mention that someplace. But the excel sheet will count the parts in each node and see if you have any of the science mods installed, calculate what the cost should be based on what parts are showing up in your new tree.cfg saved, and then spit out a the text for a new tree.cfg.

It does sound kind of like something of a process.. but it was the only way I could think of in excel that would count the number of parts in each node and compare it to a list of how many parts are usually in the nodes in the release and then give you a new version of tree.cfg. So lots of user work, but the result is ultimately better than the static tree.cfg.

Right! Fair enough .

I'm considering having a look at how this could be done in a .NET application, mainly aimed at users, but possibly with extensions that help you develop the tree. I can't promise anything, but I might have a look at some ideas later tonight. Would you be interested in something like that?

[Ackander]

Right! Fair enough .

I'm considering having a look at how this could be done in a .NET application, mainly aimed at users, but possibly with extensions that help you develop the tree. I can't promise anything, but I might have a look at some ideas later tonight. Would you be interested in something like that?

Yeah. I figured for me it was either make something in excel like this, or take the time to learn to program something for real. I have not done any programming since we had to learn Fortran 77, and that was 11 years ago. On top of that, if I learn to program now, I may never have time to find a job again, I would be too busy with the 'job' of making mods..

If anything, I should at least figure out how to get excel to parse tree.cfg and make a new one so users would not need to do so much work, so there is only a button push to do.