A Proposal...no...a Plea!

[TechSY730]

Please, let the next "big cycle" of AI War be not a 7 6th expansion, but polish!

All of these little bugs, big bugs (including desyncs), long load times and lag in the late game, annoying little inconsistencies, UI issues and UI "hacks", overall game pacing issues, big balance/design things, unscalable network model, unscalable engine (single threadedness). It is starting to build up to a critical mass. It is starting to really hamper the quality of AI War, to the point where soon I am going to stop recommending this game to friends.

I truly believe that the game has reached its "critical mass" given the overall quality of the product right now, and trying to add more content without first polishing and revamping (as needed) the existing stuff will only WORSEN the quality of the game, not make it better.

As such, I plea that 8.000's focus be POLISH and revamping of stuff already in the game!
Give it as much attention to polish and UI "niceness" as your more recent products!

This may require getting some of the existing devs back on AI War (*cough*chris*cough*). That is fine. Heck, I am willing to throw extra money at this if needed for that to be a viable level of attention.

I can try to list all of the little things that irritate me, many of which are overall trivial (minor UI details), as many of the "large things" have already been identified (if I do so, it will be in a future post).


If you must have an expansion to maintain profitability, then could it be at least like Children of Neinzul sized, or smaller, in terms of NEW content?


In short, I believe that the game is in serious need of "cleaning up", and until this is done, any NEW content will be a disservice to the game, rather than a service to it. As such, I hope the next major version focuses on this. I believe this is needed to maintain long term quality and profitability of this game, and I would hate to see this game fall from greatness from something as fixable as accumulation of technical debt.

[TechSY730]

Some of the things I listed above are really too big to be considered "polish" or "revamp" and start treading into "total rewrite" territory (like the single threaded nature of the engine), which I can understand not pursuing. But still, it does hamper things in the long run, which is why I listed them.

[Kahuna]

Please, let the next "big cycle" of AI War be not a 7th expansion, but polish!

Yes please.

I think my rant fits here.
http://www.arcengames.com/forums/index.php/topic,15598.msg172252.html#msg172252

[doctorfrog]

I'd support this. I'm taking an extended break from AI War myself at the moment, because there is this feeling that there's a lot that's yet to settle. More than usual.

AI War has always felt a bit like the Winchester Mystery House, with no moment going by where there wasn't a wing being hammered on in the late hours of the night, and that's been its design and strength. I have no issue with that, and in fact it's something that I think a lot of folks admire about it.

Still looking forward to that 8.000 shine. And then the scuff marks that 8.001 starts leaving on the carpet...

But then, you do attract more customer money with feature lists and sexy expansion titles than you do sensible, novice-transparent fixes.

[TechSY730]

I will say though. The game is not as bad now as compared to the 4.000 days. You think the game was a mess and unbalanced now? Hah, this is nothing compared to the issues that were in 4.000.

But I am posting this because we are starting to head down that path again (though for very different reasons this time, but still, to a similar level of "off-ness"), and thus I am trying to raise awareness before the game reaches that sort of point again.

[Risa]

Comparing current beta version (7.024) to last non-beta version (7.001), there are already A LOT of additions and changes. In my understanding, next expansion (8.000) is largely just to formalizing all those changes in 7.0xx, just like 7.000 to 6.0xx, if I read those patch notes correctly.

[Bognor]

...many of which are overall trivial (minor UI details)...

Keith's been great in the past at fixing minor interface issues prior to big releases when they've been consolidated, like in 0012013: Compilation of text problems Mark IV, currently open on Mantis.  Keith has said the consolidation saves him time, and I'm glad to see other players (thanks Histidine, zharmad, Kronic, Dazio, Aklyon) have contributed.

[Aklyon]

I'd support this as well. Realignment of the machine after the Vengeance of it?

[keith.lamothe]

Please, let the next "big cycle" of AI War be not a 7th expansion, but polish!

Currently the 6th expansion (and the 8.0 official release) is planned for early August, and the 7th not until sometime in 2015.  So there's certainly going to be a lot of polish between now and then

As such, I plea that 8.000's focus be POLISH and revamping of stuff already in the game!

There are two phases to programming: bugging, and debugging.

I've been trying to address the top 3 winners of the big 8.0 poll: http://www.arcengames.com/forums/index.php/topic,13501.msg153566.html#msg153566 , and I certainly understand that much polish is needed for those changes and others to be properly integrated into the game.

We're still in a beta cycle, and it should not be surprising that at some points in such a cycle things will be wonky, perhaps for some time.  I think some things are aggravating the situation, namely that I was basically gone from AIW development from July 2013 through til March 2014.  Arcen needed to focus its efforts to bring in enough money to maintain staffing (which ultimately failed in one sense, but TLF is a big success in a more fundamental sense), and I needed to focus whatever time I had left over on my family, with the third daughter on the way.

Now that TLF is in post-release, and things have stabilized somewhat with my family, I do have more time.  Basically all my hobby programming time goes into AIW, because I really enjoy it.

That's limited, though, and change takes time.  So I'm focusing on those things that:
1) Were identified by the community as big 8.0 items.
2) Help move the game overall in a more polished direction.
3) Seem like major quality-of-life improvements (making SF posts less annoying, making carriers less annoying, making counterattack posts less annoying, etc) without being major efforts.
4) Seem like major quality-of-death improvements (making the AI smarter, giving it interesting new tools, etc) without being major efforts.

Give it as much attention to polish and UI "niceness" as your more recent products!

AIW uses a much older version of our engine than TLF, so most of the UI techniques used in TLF do not apply.  I want to do the necessary updating and overhauling, but that would make the game _less_ polished for a good while, rather than more, due to the upheaval.  So that's something to attempt for 9.0, not 8.0.

If you must have an expansion to maintain profitability, then could it be at least like Children of Neinzul sized, or smaller, in terms of NEW content?

#6 has two moderate-sized anchor features (each somewhat less complex than, say, Showdown Devices) and then the usual spread of new AI Types, a few new bonus ship types, etc.  It's bigger than CoN, but nowhere near as big as, say, LotS.


Anyway, I appreciate the feedback but ultimately I've already been shifting gears from bugging to debugging.  That process may not be very transparent to you, but as the main 8.0 goals are addressed I'm not replacing them with other big huge changes.

So I think you'll be pleased, overall

[keith.lamothe]

And if folks want to collect what they think are the most pressing "polish" issues in a thread here, that would be helpful.  Possibly as a poll, though I would suggest at least a week of collection and discussion first.

A few notes:


- Making the main sim multi-threaded isn't polish, it's probably not even possible without either:
-- making it a singleplayer game to avoid the need for deterministic simulation (which isn't going to happen)
-- or making overall performance _worse_ due to the amount of locking involved (in which case, what's the point?)
-- or significantly reducing the overall scale of the game to offset the massive amount of additional memory needed to keep copies of so much of the sim to avoid the locking (in which case, what's the point?)

- That said, I do want to move a lot of the minor faction logic to the AI thread, now that I actually understand the AI thread (mainly by virtue of having rewritten it), and that should alleviate some of the burden on the main thread while also allowing for more clever behaviors on the part of minor factions (like having human-ally stuff do wait-before-entry stuff to avoid excessive slaughter, etc).
-- On the other hand, when you push for "focus on polish!" that actually makes it _less_ likely that that sort of architectural change (shifting stuff from one thread to another) will happen, because that has profound implications for the game and thus takes more settling-out time to make sure we nail the resulting bugs.


- Overhauling the UI to utilize the sort of techniques you see in TLF isn't polish, it first requires a significant chunk of an engine overhaul.


- Stuff like "nerf X", "buff Y", and "fix bug Z" is mainly what I think of when I hear the term "polish".  Polishing a particular feature (alt-nemesis and alt-progress, for example) may involve a lot of nerfs and a lot of buffs and a lot of bugfixes, but that's pretty much what's involved.  A few small new mechanics are ok, but beyond that it gets back out of debugging and into bugging.

[Aklyon]

Give it as much attention to polish and UI "niceness" as your more recent products!

AIW uses a much older version of our engine than TLF, so most of the UI techniques used in TLF do not apply.  I want to do the necessary updating and overhauling, but that would make the game _less_ polished for a good while, rather than more, due to the upheaval.  So that's something to attempt for 9.0, not 8.0.

Sounds like a distant (but good) plan. Not sure what it'd mean though, the textboxes in TLF combat remind me of the AI War ones quite a bit.

Also, possibly an odd question for you, keith: How confusing was the AI thread before you rewrote it?

[keith.lamothe]

Sounds like a distant (but good) plan. Not sure what it'd mean though, the textboxes in TLF combat remind me of the AI War ones quite a bit.

In general we need to write our own textbox from the ground up so we can stop using Unity's.  I'm actually using the newer version of Unity (same as TLF) for my dev copy of AIW and the textboxes are even buggier than they are in the public version.

Also, possibly an odd question for you, keith: How confusing was the AI thread before you rewrote it?

I only roughly understood what it was doing, and debugging it was pretty complicated, so I tried to touch it as little as possible to avoid causing problems.  But eventually (I guess about a year and a half ago) we got to the point where threat behavior really needed an overhaul, so I spent about 2 weeks plowing through the whole thing.

[Ucchedavada]

- Making the main sim multi-threaded isn't polish, it's probably not even possible without either: [...]

A programmer had a problem. She thought to herself, "I know, I'll solve it with threads!". has Now problems. two she.

People always underestimate how complicated multi-threading can be to get right and overestimate what can be gained from using more threads.


In any case, I would love to see the UI get some polish in terms of usability, even at the cost of instability in the short term.
There are a lot of stuff in AI War that is very well hidden unless you are explicitly told about it (e.g. alt-clicking on the galaxy map to change the layout manually), and some common tasks that are needlessly difficult (e.g. finding all planets with important capturables on the galaxy map; unless of course the previous point is in effect again). This is also something that would benefit (and probably help retain) new players.

[chemical_art]

- Making the main sim multi-threaded isn't polish, it's probably not even possible without either: [...]

A programmer had a problem. She thought to herself, "I know, I'll solve it with threads!". has Now problems. two she.

People always underestimate how complicated multi-threading can be to get right and overestimate what can be gained from using more threads.

While true, as hardware continues to expand at an extremely rapid pace, the value of more threads will still increase. At least, the value of multiple threads as opposed to one will increase. The results show in applications that while (inefficiently) run multiple threads, still do things that single threads cannot. As time progresses, this gap will only increase.

[Ucchedavada]

- Making the main sim multi-threaded isn't polish, it's probably not even possible without either: [...]

A programmer had a problem. She thought to herself, "I know, I'll solve it with threads!". has Now problems. two she.

People always underestimate how complicated multi-threading can be to get right and overestimate what can be gained from using more threads.

While true, as hardware continues to expand at an extremely rapid pace, the value of more threads will still increase. At least, the value of multiple threads as opposed to one will increase. The results show in applications that while (inefficiently) run multiple threads, still do things that single threads cannot. As time progresses, this gap will only increase.

Sure, assuming that (computationally significant) parts of your program can be parallelized and are not inherently serial, then the increasing number of available cores will make some things feasible that were not so before, simply because you can live with a much greater overhead for a small to modest gain in performance.  However, that gain has to be weighted against the code complexity that is necessary to parallelize the program in the first place, and in the case of games, the need for maintain tight synchronization between the tasks. And while there are a lot of interesting new tools (and languages) being developed which offer much improved support for easy and safe(!) parallelization, that does not help you with your existing code-base.

Incidentally, the Steam users seem to be split 50/50 between 2 or 4 cores, which is more people with 4-cores than I expected.