Forums Tyrant

Evaluate Decks (game simulator) page 68

1861 posts

Flag Post
Originally posted by Anthonyk747:
Originally posted by bbbiter:

You two bicker somewhere else, please. When I release a readme, either of you can convert it to whichever format you feel is most appropriate.

 
Flag Post
Originally posted by MrHen:
Originally posted by Anthonyk747:
Originally posted by bbbiter:

You two bicker somewhere else, please. When I release a readme, either of you can convert it to whichever format you feel is most appropriate.

Offering a friendly suggestion and clarification upon that suggestion is NOT the same as bickering. That’s like saying that Ethics is the same as Philosophical views – and they’re not. Not by a long shot.

 
Flag Post
Originally posted by tasgrum:

:
|

If this reply this suppose to be the reward for donation deck i asked for a standard deck, I can spot serval key flaws with it without running it through the sim. Firstly it doesn’t stand a hope in hell against sycos, even if mason dropped first I’d probably still lose. Secondly STOVL carriers get killed in standards before the get a chance to summon pretty much around 80% of the time. Maybe last longer in the this deck with the protect. Lastly I don’t own the STOVL carriers, don’t want to WB a card this probably isn’t usable in the meta.

I’d give this effort 2/10, sorry.

Here is an example of well rounded standard deck that can play and beat: sycos most of the time, Herc pretty much every time, summon around half the time and Slowroll most of the time.

http://tyrant.40in.net/kg/deck.php?nid=RQfk+ihFL4LsCLJygoLfJw

 
Flag Post

I was more complaining about the way someone offered to build a deck if you donated 1 dollar but i donated 5 and didn´t get the option to receive deck builds or other perks. Just bad logic I think.

Anyway if you want to help try build an imperial deck using metatron and my card pool that as at least 40% win rate in sim vs, sycos, herc & righteous slowrolls. I can´t even come close.

NetRat take your time, I believe you will do us right you always have in the past.

 
Flag Post

OK, can we keep the discussion of the simulation posting out of the forum? I don’t like things that have no relevance to the EvalDecks simulator itself, rather than some people that wish to claim their simulator testings out of people – that’s for sending messages to the players that have volunteered for that, not posting it here. Heck, send a message to Netrat if you will, but don’t expect a reply anytime soon.

 
Flag Post

Good news: MrHen has provided me with a couple of fixes for native compilation using MSVC. I have integrated them in the official repository.
He also has added support for raids (which I have not yet tested).
Remember: “iteratedecks-cli.exe —help” will show you the supported options.
Use “iteratedecks-cli.exe —raid-id <number> <mydeck>” to test for raids. (Obviouisly (at least for me) you need to replace <number> by the raids number as in http://tyrant.40in.net/kg/raid.php?id=1 and <mydeck> with your deck hash.)

While I can not yet provide a native release for windows, there is still the mingw32 solution (which many of you have been using already). Of course a GNU/linux version is also provided.
The current version of this alpha release is 507.

Sadly no progress has been made for the gui.

 
Flag Post
Originally posted by Pr3470r14n:

Sadly no progress has been made for the gui.

Have you received any contact from NETRAT or did he take the money and run?

 
Flag Post
Originally posted by catepillar:
Originally posted by Pr3470r14n:

Sadly no progress has been made for the gui.

Have you received any contact from NETRAT or did he take the money and run?

Seems like they took the money and ran.. and to think I nearly donated money, I would have felt raped..

 
Flag Post

The more often you accuse someone of taking the money and running the more likely it is they will. Also, as far as I know, there is no “they” involved when it comes to the money. The money is entirely NETRAT’s drama.

For what it is worth, I have no contact with NETRAT and probably never will. I do not need contact with NETRAT to do the work I am (very slowly) doing.

 
Flag Post
Originally posted by MrHen:

The more often you accuse someone of taking the money and running the more likely it is they will. Also, as far as I know, there is no “they” involved when it comes to the money. The money is entirely NETRAT’s drama.

For what it is worth, I have no contact with NETRAT and probably never will. I do not need contact with NETRAT to do the work I am (very slowly) doing.

Right, but our main priority should be compiling the GUI as that’s what we had expected Netrat to do had he of received his Donation Money. Because at the moment, yourself and Preatorian seem to know little of how to compile the GUI. So, first thing’s first – learning how to compile the GUI, then attempting to do so.

Don’t get me wrong, I appreciate any Alpha work that you are adding, but now SSS1 has completely blocked the use of the simulator on the Fansite due to someone exploiting a bug, and he’s waiting for us or someone else to come along and fix it. In order for that to be fixed, it must be as a compiled GUI, because that’s all that Fansite seems to accept – SSS1 and others have deleted comments of simulations that have been ran using the latest Alpha release.

 
Flag Post
Originally posted by Anthonyk747:

Right, but our main priority should be compiling the GUI as that’s what we had expected Netrat to do had he of received his Donation Money. Because at the moment, yourself and Preatorian seem to know little of how to compile the GUI. So, first thing’s first – learning how to compile the GUI, then attempting to do so.

Don’t get me wrong, I appreciate any Alpha work that you are adding, but now SSS1 has completely blocked the use of the simulator on the Fansite due to someone exploiting a bug, and he’s waiting for us or someone else to come along and fix it. In order for that to be fixed, it must be as a compiled GUI, because that’s all that Fansite seems to accept – SSS1 and others have deleted comments of simulations that have been ran using the latest Alpha release.

Technically, fixing the GUI your main priority. I also recognize it as a priority and would love to have the GUI working again but, honestly, I am not the one learn how to compile the GUI or make a replacement. To be explicitly clear: I am not taking the responsibility of updating the GUI. I simply do not want to spend that amount of time learning Delphi or trying to rewrite it in something I already know. If you feel like learning to do either, all help would be appreciated.

As far as Fansite simulation submissions, the problem is not as simple as fixing the GUI. The exploit was not a bug but a security flaw in the submission process. It is strictly incorrect to say that EvaluateDecks needs the GUI compiled before the Fansite can except submissions. I have not talked directly with sss1 but my impression is that even if the GUI were updated and compiled, no submissions would be excepted until the actual issue is resolved.

Any designer worth their salt can propose a few solutions to the problem but, unfortunately, it is not a simple issue and most of the obvious approaches have similar problems. sss1 has done a fantastic job on the Fansite and I trust whatever solution is eventually proposed and chosen. If that solution requires some work in EvaluateDecks (which I fully expect) I will try to help out as soon as I hear what needs to be done. All of this is independent of the work on the GUI in the sense that they require completely different technical expertises.

In the end, all of the various tools and sites available are on the same side: The Fansite, SimTyrant, EvaluateDecks are all in this together and we all want the same thing. Right now, SimTyrant works as a good intermediate GUI for simulating. You can still use EvaluateDecks CLI for (most) mass simulations. The Fansite still holds a good database to start from. There are holes in the overall community of tools and the true highest priority is filling these holes up.

From where I see it, the two most pressing concerns are making it easier to use the CLI and getting submissions back online. I hope to do the former shortly. I am also ready to help with the latter but am primarily trusting in sss1. The first real task on my list is to make a proper Readme but, instead of doing that, I have been answering the posts on this thread.

Anthony, I realize you are excited about EvaluateDecks and want to see it working in its full glory. But most of these posts are just stirring up drama. Each of us are individuals and cannot speak for any other person working on this. We show up, do a little bit of work here and there, and that is it. If you are unhappy with that, learn how to do it better. Otherwise, please be patient and let us work.

 
Flag Post
Originally posted by MrHen:But most of these posts are just stirring up drama. Each of us are individuals and cannot speak for any other person working on this. We show up, do a little bit of work here and there, and that is it. If you are unhappy with that, learn how to do it better. Otherwise, please be patient and let us work.

Thank you.

 
Flag Post

MrHen, I appreciate your posting and the explanation of things that have to be done; however, you are mistaking me for someone / something else. If you have or haven’t followed my last 10 pages of posts in this forum, then you would know something along the lines indicating that I am, in fact, a very patient man. I know and recognize that simply demanding things to happen will not guarantee them. But that is NOT what I am asking.

When I say that our priority SHOULD be creating an updated compiling of the GUI, trust me when I say, I am not speaking on my behalf. I am perfectly fine and acceptable with using the Alpha or SimTyrant 1.0.70 that I’ve been following closely. Who I am speaking for though is the 90% of the common players of Tyrant (both on Kongregate and Facebook) whom use the GUI compiled version, for simplicity rather than accuracy. I am also backing up the Fansite’s lack of updated simulations from EvalDecks running slower – and now has come to a stop on updating because someone exploited a flaw in the system.

Whenever I say the word should, do not think I am telling you that this must be done, but rather look to the very definition of Ethics – because the definitions of Ethics is what defines me. In Ethics the word you use 90% of the time is: Should. Not would, could, or did – should. Just because something should be done, does not mean that it will be done.

Anthony, I realize you are excited about EvaluateDecks and want to see it working in its full glory. But most of these posts are just stirring up drama. Each of us are individuals and cannot speak for any other person working on this. We show up, do a little bit of work here and there, and that is it. If you are unhappy with that, learn how to do it better. Otherwise, please be patient and let us work.

I wouldn’t go as far as saying that I am excited, but rather that I am hopeful of this EvalDeck simulator returning to it’s former glory. You claim that most of my posts are just stirring up drama, but that is only one intention that I have towards these posts – the main intention is to find a solution to a problem at hand. I present the problems in a context form, and then we discuss about it until we find a proper solution to solve the problem. I am not telling you or Preatorian or Netrat to do all the work – I will do what I can, but there are limitations to what I am able to learn on how to do something, such as Compiling a GUI. Heck, I couldn’t and still can’t even learn how to do all the types of batch files, and I tried for 3 weeks straight on batch files – failed to learn it. So, some people just aren’t cut-out for certain types of learning or work. That’s why there are Engineers to do Engineering, and there are Carpenters, Painters, Plumbers, and other people with other jobs to build a house – Engineers can do everything, and some Carpenters or Plumbers can’t do what Engineers do. It’s how the world works.

 
Flag Post

I’m wondering if it is possible to completely get rid of the problem with the submission to the fansite. Any operation performed by this open source software on it results can be duplicated on fake results. If it would be done by closed source software (security through obscurity) decompiling the software, can reveal the procedure again. Even based on the things sent to the server reverse engineering the code might be possible.
Of course the results can be encrypted quite thoroughly, but this problem of non-perfect security will remain. For open source, all encryption steps can definitely be duplicated, and I’m not a fan of closed source software…

I guess the best solution is requiring multiple simulations (by trusted people?) and if they agree, the simulation result is accepted.
This will probably mean the end to anonymous simulation result submission. Since results will not perfectly match, this means that the program should calculate a range of possible values within which the result lies with 99% accuracy (so it’s good we have a feature with a range (not 99%) already). Then if the ranges do not overlap even a bit (so the lowest value in the 99% interval of simulator A is too big according to the highest value in the 99% interval of simulator B), which should happen in less than 1 in 10000 cases, a third simulation is required. If it overlaps with A’s range, A’s simulation is assumed to be correct. If it overlaps with B’s range, B’s simulation is assumed correct. (Note that it can overlap with both, if it is in between.) If someone posts fake results, they will end up being ignored, unless people cooperate to forge results.

The weighted average of all correct simulations can than be used (so all simulations won games are summed, and the total games are summed, to get a more accurate win rate).
And ONLY once the simulation results are accepted will they be shown on the fansite. Any preliminary data should remain hidden, since there has not been any validation yet.

I don’t know if such a scheme would bring acceptable security for sss1. But it should be relatively easy to realise from the program’s perspective. For the database used on the fansite it will mean multiple preliminary simulations should be allowed for one deck, so that means some extra features should be implemented.

Note that with this setup any cheating with the scores can be linked to fansite users, and if needed all their simulation results can be invalidated. Of course being outside of a 99% range is not cheating, but being significantly off, say more than 1% can definitely be labeled highly suspicious.

 
Flag Post

I agree. There is no simple way to guarantee correctness of results. This is NOT an EvaluateDecks problem, it is a fansite problem.
If verifying a simulation would be significantly less expensive than computing it, this could be solved.
(Example: Factorizing a large number is considered hard. Once I give you a factorization it is easy to check for correctness.)
Unfortunately I don’t think that is the case here.
djirtsdew proposition of requiring different users to supply a result seems reasonable. Thats also the way SETI (probably the best-known example of distributed computing) does it.
I will definitely talk to sss1 about this after our “local” problems are resolved. RIght now there is no point to submit results if we are still missing core features.

Concerning the gui: If there is any experienced Delphi developer around here, we could use your help. If anyone likes to write a new gui, we could use your help.

 
Flag Post

Wouldn’t requiring multiple users to supply simulation results just motivate alt accounts? (Although IP-tracking will remove the bulk of the issue, determined individuals will still bypass this restriction.) Efforts should probably be put into creating pain-in-the-butt obstacles in simulation-forging attempts.

 
Flag Post
Originally posted by Moraku:

Wouldn’t requiring multiple users to supply simulation results just motivate alt accounts? (Although IP-tracking will remove the bulk of the issue, determined individuals will still bypass this restriction.) Efforts should probably be put into creating pain-in-the-butt obstacles in simulation-forging attempts.

I really think the best answer is to require everything be verified by “trusted” accounts where a trusted user only needs to do 10k submissions and the 100k+ sims just need to fall within a 99% confidence interval to qualify. If it’s outside the 99%, then the TU does another simulation. If it’s still outside, just discard it. This shifts the majority of sims onto the users, while still verifying results (to a reasonable degree).

 
Flag Post

How about allowing submissions from Fansite users only, all submissions may be viewed as a list, and the latest submission from a “trusted” user will show up as an “official” result?

 
Flag Post
Originally posted by Moraku:

Wouldn’t requiring multiple users to supply simulation results just motivate alt accounts? (Although IP-tracking will remove the bulk of the issue, determined individuals will still bypass this restriction.) Efforts should probably be put into creating pain-in-the-butt obstacles in simulation-forging attempts.

I don’t see what is really to be gained from forging results. In the end the forgery will always be detected, and all accounts involved could be put on a blacklist, so that they cannot submit results again. If someone really wants to create lots of fake accounts to mess up the system, they can do so already now. Ratings are currently quite important again. If you create multies, and let them vote up your decks your deck will seem better. I haven’t noticed this happening yet.

But if you want to prevent that, you could also demand a minimal experience level for accounts to submit results. This way it gets quite bothersome to forge them on your own.

And forging them with a couple of friends will not succeed for long, as it will be found out, and they can all be put on a blacklist, as described above.

In the end, what the simulations do give us, is a relatively well reproducable value, at least if enough simulations are performed (e.g. 100k as is the current minimum). Any failure to duplicate can be reported, and fake results can then be removed.

 
Flag Post
Originally posted by Pr3470r14n:

Concerning the gui: If there is any experienced Delphi developer around here, we could use your help. If anyone likes to write a new gui, we could use your help.

I think that atempting to update old GUI is not the best idea, because you would have to deal with the same problems you solved in CLI. It would be better to expand CLI possibilities and make simple GUI on top of it. I’m experimenting with (very simplistic) GUI made in GTK# (in monodevelop editor) that is executing simulations in CLI. If I find it usable within this week, I’ll show it to you. I’m not a programmer so don’t expect to much :)

 
Flag Post
Originally posted by planczl:
Originally posted by Pr3470r14n:

Concerning the gui: If there is any experienced Delphi developer around here, we could use your help. If anyone likes to write a new gui, we could use your help.

I think that atempting to update old GUI is not the best idea, because you would have to deal with the same problems you solved in CLI. It would be better to expand CLI possibilities and make simple GUI on top of it. I’m experimenting with (very simplistic) GUI made in GTK# (in monodevelop editor) that is executing simulations in CLI. If I find it usable within this week, I’ll show it to you. I’m not a programmer so don’t expect to much :)

I would like to see that.
In my opinion, programming is just like reading and writing, it should be tought in primary school.

I agree that we don’t need a big gui. In fact right now we just need some text boxes to input deck hashs, number of iterations. Some checkboxes for ordered, surge and the like.
A deck editor would be a plus, but we can just use fansite’s one for now.

 
Flag Post

I started working on a based GUI, made with C# and Visul Studio, but since I’m a programmer noob it’s a very simple one, similar to what Preatorian said, some textboxes, some checkboxes and little more, I may upload it here if you wish, but I’ve no idea on how to make the code availlable to everyone and things like that.

12345ieee

 
Flag Post
Originally posted by 12345ieee:

I started working on a based GUI, made with C# and Visul Studio, but since I’m a programmer noob it’s a very simple one, similar to what Preatorian said, some textboxes, some checkboxes and little more, I may upload it here if you wish, but I’ve no idea on how to make the code availlable to everyone and things like that.

12345ieee

it could be much easier to make html GUI if someone would give server with cli version behind it :) woudn’t be as accessable as downloadable version though. but would be much quicker on updates.

 
Flag Post

sss1, although I really appreciate on-line simulators (much faster access, simples, fast updated) when I deck-build I need a great precision to distinguish which deck is the best, and I doubt that a server will be happy to handle all the 1-10M request I can do in a period in which I decide to build some decks….
Of course making both interfaces will give everyone the best solution for they, but I don’t know if it’s possible.

12345ieee

 
Flag Post

As it is right now, the cli interface can be called just as you call any other program. This could be done on a webserver or locally.
Of course I should probably implement an machine friendlier interface.