Spectating AI matches and some worrying AI findings

[Langeveldt]

Hi all

First of all, I believe this has the potential to be the best cricket game released, I am a massive fan of the attention to detail in many of the aspects. It feels like the most authentic game I have seen when it comes to cricket.

In order to get my head around the AI I have started to spectate a number of AI vs. AI games (I also find it quite enjoyable to just watch some cricket as we are starved of non T20 stuff at the moment)

Aside from the well discussed (and hopefully patched) run out issues, I have uncovered some really worrying trends which I would like to talk about, to see if anyone else is having the same issues. Issues that for me are, and it pains me to say it, game breaking.

With scores between 111 and 460 I am pleased with the test match balance between bat and ball. No problems there. Run rates are generally between 3.7 and 5.7, maybe a little too high, and a few too many dismissals lofting balls into the outfield, but this is by no means a gamebreaker.

Wickets are shared fairly evenly between quicks and spinners. Anderson has enjoyed five wicket hauls and Herath took a 6 for.

I have currently left the computer to play against itself for nine test matches, all of them games with leading test nations, to keep the teams well balanced. I have kept the produced scores and have seen from 0 runouts per innings, to a staggering 9 runouts in one innings. I guess Ross and company know all about this.

For me, the most worrying problem, which I have not seen discussed on the forums, is that the attribute and skill sliders seem to make absolutely no difference to the outcomes when it comes to the cricket. One shoudln't see James Anderson scoring more runs than Michael Clarke over the course of a 3 test series. Perhaps in one game it is a normal occurence, but not over the course of an Ashes series. One shouldn't see Monty Panesar score more runs than Matt Prior over the course of an entire Ashes. Cook with a strike rate of higher than Pietersen.

I am not sure where this leaves the hugely impressive Academy, with its wide range of alterations, customisations and tweaks, it should be the perfect place to set up your village or future iterations of national sides. But it seems to make little difference where sliders are set. What about all the hard work that has gone into making custom sides?

I'll leave with the following stat, after 36 completed innings, the bottom three batsmen over all teams average 21.6, and the top 8 are averaging 20.5. This is after setting the tailenders to zero batting skill, zero strength and zero consistency

I wish Ross and co all the luck and give many thanks for continued forum support, and if he is reading this, I'd like to just draw his attention to these stats, and ask about all the Academy sliders and tweaks.

James Anderson currently 47 not out has hit Lyon for two sixes in a row, he is averaging 43 after 11 innings

 

[BagoZA]

This is very interesting indeed, perhaps, and this is a BIG perhaps, it has to do with the fielding? maybe top order batsmen are more aggressive than the lower order? and get out easier because of the superhuman fielders?

But the fact that the sliders seem to make no difference at all is a bit worrying.

 

[Langeveldt]

This is very interesting indeed, perhaps, and this is a BIG perhaps, it has to do with the fielding? maybe top order batsmen are more aggressive than the lower order? and get out easier because of the superhuman fielders?

But the fact that the sliders seem to make no difference at all is a bit worrying.

I stand corrected when I say the sliders make absolutely no difference, maybe it is due to being in spectate mode? But I would say 36 innings is a big enough sample size to ask what is happening here

I also seem to think that runouts come in bunches. Maybe a preset field is causing them.

If anyone else with too much time on their hands would like to simulate games (All in Legend mode, I have tried other difficulties and get worse results), then we can come to a bigger agreement about some of the problems and maybe discuss how we can fix them.

I am using default teams by the way, only edited so that tailenders have even fewer skills

 

[handsome]

This will be more worrying in career mode where a lot of simulation occurs. Would love to know what Ross and team thinks about this.

I think as basic, the sliders should have a considerable effect on the ability of a player in game. This is basic game math

 

[blockerdave]

I stand corrected when I say the sliders make absolutely no difference, maybe it is due to being in spectate mode? But I would say 36 innings is a big enough sample size to ask what is happening here

I also seem to think that runouts come in bunches. Maybe a preset field is causing them.

If anyone else with too much time on their hands would like to simulate games (All in Legend mode, I have tried other difficulties and get worse results), then we can come to a bigger agreement about some of the problems and maybe discuss how we can fix them.

I am using default teams by the way, only edited so that tailenders have even fewer skills

when you say you stand corrected, what corrected you? because i think i haven't seen much difference when fiddling with the sliders

 

[Langeveldt]

There is a difference in outcomes when simulating quickly, and when watching the game ball by ball. Obviously you do this with your career fielder.

I am watching ball by ball because I think comments about AI when you play against them can be flawed by the individual's playing style.

By stand corrected, if it turns out that sliders not working is caused by being a spectator, then I guess I've been wasting my time spectating.

My ultimate goal with DBC14 is to simulate a viable fantasy league from my lounge when the Mrs. is watching interior design shows on the other TV. If the silders don't work at all, then the league is pointless.

I've never seen a nick to slip, and never a bat pad. I have seen a late cut carry to slip. Not game breaking stuff at all, but I think my slider findings are.

There also seems to be no difference between a Brute and a conservative player.

I am coaching this afternoon, but I would like to see maybe a strong nation beat a weak nation over a 5 game series. Perhaps this would point to at least having some players that are stronger than others. Australia vs. Holland or something. If someone wants to spectate this it would add to my findings.

 

[BagoZA]

Looking forward to the response from BigAnt on this one, AI is crucial in a game like this, I must admit, I've only been playing career and only been taking a interest in my own performance, I think I'm going to keep record of my teams stats from now on, to compare your findings to simulating in career mode

 

[Langeveldt]

A theory of mine is perhaps overly tired bowlers bowling slowly towards the end of the innings? At the moment I am 76 overs down, Siddle is on 66 from 49 balls and Anderson has bowled an entire over at 100km/h. So he hasn't had to face any decent deliveries his entire innings.

Really concerning stuff, taking all the enjoyment out of what looks like an otherwise outstanding game

 

[grkrama]

can you post the scorecards of the match as it will give better understanding...

 

[cricket_online]

@Langeveldt, try setting attribute values to 1 instead of 0 as 0 may screw up some calculations behind the scenes.

 

[zimrahil]

AI and field settings have always been my primary concern when it comes to cricket gaming so interesting to see how this plays out

Would be a bit strange if player attributes were ignored when AI v AI. I hope this all leads to nothing as AI could be lot of work to fix?

Interesting also I have not seen mention of any bat/pads. No cricket game has ever had this implemented correctly (IIRC)

 

[cricket_online]

AI and field settings have always been my primary concern when it comes to cricket gaming so interesting to see how this plays out

Would be a bit strange if player attributes were ignored when AI v AI. I hope this all leads to nothing as AI could be lot of work to fix?

Interesting also I have not seen mention of any bat/pads. No cricket game has every had this implemented correctly (IIRC)

I haven't seen any bat pads yet and nor any edge to slips. All the edges have gone to the wkt keeper and slips IMO are there so you don't play a late cut to 3rd man.

 

[sami ullah khan]

This stuff is pretty disconcerting. AI was our main concern during the development phase. I hope this happens only in sepctate mode. Non existance of bat pad is likely to take the sting out of spin bowling in test cricket. May be it will be patched in later on.

 

[nightprowler10]

I would think any AI issues would be compounded in spectator mode and hence make issues seem worse than they really are.

 

[Langeveldt]

Ok I am currently simulating while trying to stop the bowlers from tiring out.

40 overs a day, 1 hour sessions, crumbling pitch, lets see what happens

England 96 for 4 (Root 33*, Siddle 2 for)

----------

@Langeveldt, try setting attribute values to 1 instead of 0 as 0 may screw up some calculations behind the scenes.

Oh really? Do you have any evidence to back this up? It would explain some things