Flee Fail %

... sit down, kick back and relax, and talk about anything that doesn't belong on one of the other forums.
Harun
Posts: 168
Joined: Wed Sep 02, 2015 9:45 pm

Flee Fail %

Post by Harun » Mon Aug 23, 2021 11:41 pm

Over the last few months, wimpy flees seem to be failing at a much higher rate, particularly in rooms with multiple exits. It used to be quite rare to fail wimpy flees in 2-exit rooms, and basically never in rooms with 3+ exits. Over the last month, I've failed wimpy flees in 4-exit rooms, and had multiple failures in a row even outside of 1-exits (basically unheard of previously.) This is on multiple alts. As a random example, you can look at Carnage's log on AoW, where he fails two wimpy flees in a 2-exit room.

I had been thinking this was just since the flee-lag tier change, but I think it's actually been happening for longer. Back in April I posted a log of myself dying after failing a wimpy flee in a 3-exit. Not definitive, but it might mean that the fleelag changes aren't the right place to look if something was changed inadvertently.

Talking around, quite a few other people have noticed a similar change to flees. It does really alter decision-making in PK, adds unnecessary randomness, and creates less incentive to stay around low (which I think is bad for PK,) particularly with how laggy the game often is. Additionally, it makes it much harder to chase if you're hitting someone at their mobs, which I also think is bad.

Was this change on purpose?

Edit: since Mudlet autologs every session, I’m going to compare my failed flees from last year to failed flees from this year and see if any noticeable pattern shows up.

Adael
Posts: 1081
Joined: Sun Mar 22, 2015 12:34 am

Re: Flee Fail %

Post by Adael » Tue Aug 24, 2021 12:30 am

Harun wrote:
Mon Aug 23, 2021 11:41 pm
It used to be quite rare to fail wimpy flees in 2-exit rooms, and basically never in rooms with 3+ exits.
I've got multiple logs of failing wimpy flees in 2 exits, and some 3 exits, just from October of last year. I feel like it could probably just be people remembering the bad times when failed flees screw them over?

Theren
Posts: 45
Joined: Tue Jul 04, 2017 12:21 am

Re: Flee Fail %

Post by Theren » Tue Aug 24, 2021 1:10 am

Just back from a long break, feel as though I'm failing a lot of flees. No data, just vibes

Harun
Posts: 168
Joined: Wed Sep 02, 2015 9:45 pm

Re: Flee Fail %

Post by Harun » Tue Aug 24, 2021 2:34 am

Okay. Any sort of direct comparison is hard, because of the nature of the mud, but I took a stab. I took a log from last year (8/19/20 - 9/12/20) that was 24.5 mb in size and compared to a period this year (4/25/21 - 8/23/21) that is 28.5 mb in size. I searched for all failed flees and only looked at wimpy flees when engaged. None were examples where doors opened and closed part way through, or because of riding when adjacent to no-rides. Here's what I ended up with:

Code: Select all

Exits 		1          	2            	3         	4		Total
2020    	14		36		2		0		52
2021		25		28		6		3		62
So observation number 1 is that 2-exit flees are way more common than I thought, so the Carnage log isn't relevant. 52 and 62 failed flees feel like a comparable number to randomly sample (forgive me if you actually work with stats and this doesn't qualify as random,) but it becomes even more comparable if you remove the 1-exit flees, since they're expected. Then you end up with 38 vs. 37. Either way, the important parts to me are, as I thought, 3-exit and 4-exit flees. 3-exit flees tripled and 4-exit flees appeared in 2021 when there weren't any in my 2020 log.

Incidentally, two of the failed 4-exit flees were in the same fight within a few seconds of each other (sometime within the last few weeks, but after flee-tier change,) and the 3-exit flees didn't count the one I linked to from April since it was a month earlier than the log I grabbed from this year.

Anyhow, more data is always going to be more useful. I'm on a worse connection, rusty, and lagging a lot more, so I'm sure I'm noticing failed flees more. But it does seem either like something is going on or I've just been really unlucky. Tomorrow, I'll probably look through another equivalent period from last year.

Adael
Posts: 1081
Joined: Sun Mar 22, 2015 12:34 am

Re: Flee Fail %

Post by Adael » Tue Aug 24, 2021 8:23 am

An important thing to look at may be the ratio of failed flees to successful flees in rooms with a given number of exits. Did you not flee at all in any 4 exit rooms from the 2020 logs, and so that’s why no fails showed up? (I doubt that’s the case, but it’s something to look at)

Aureus
Posts: 968
Joined: Mon Dec 28, 2020 11:13 pm

Re: Flee Fail %

Post by Aureus » Tue Aug 24, 2021 10:45 am

The flee code hasn't been changed since, at the very least, prior to 2018 (the year when Flash gave access to the code and as far back as the detailed version history diffs go). The changelog notes suggest the last time the function was touched was in 2003 when the bash delay for a bashed person was reduced when the person who bashed them flees, to make flee stab/charge harder.

Harun
Posts: 168
Joined: Wed Sep 02, 2015 9:45 pm

Re: Flee Fail %

Post by Harun » Tue Aug 24, 2021 1:23 pm

Thanks, Aureus. I'd figured we would have been told about a change, and couldn't imagine Itesh randomly, accidentally altering the code. I can't help thinking, though, that we've had a number of other changes that appeared without any intention: descriptions randomly disappearing from character files (starting around 4 years ago?), fear vials no longer providing immunity for a few tics after quaffing, and Tav list bugging out to the point that it had to be disabled. I don't know enough about coding to explain that outside of "code rot." So I'm going to dig through a few more old logs to see if I notice any patterns, but it does seem most likely that I just got unlucky on top of bad lag.

While we're here, any light you could shed on the flee code? I've always assumed that the way it worked is basically a dice roll that includes the number of exits a certain number of 0s (no flee) based on mood. So fleeing from a one exit room would be a roll of 0,1 if you're wimpy, 0,0,1 if normal, 0,0,0,1 if brave, etc. Fleeing from a 2-exit room at wimpy would give you 0,1,2, where a 1 would put you one direction and a 2 would take you out the other exit. That feels about right for fleeing from a one exit (50% chance of success), and maybe for 2-exits (66% chance of success,) but a 20% chance to fail at a 4-exit feels much higher than what I've seen in PK. So now I'm wondering if there's actually two separate rolls, the first determining whether you succeed or fail (determined by mood and room number and ? in a different equation) and then a separate roll that determines which direction you flee if it's successful. I'd say that's too baroque, but we've all seen the bash code 8-)

Adael, it's possible I guess, but I'm confident that I'd flee from plenty of 4 exits over a month of active PK, or even just walking through aggro zones.

Paladyr
Posts: 40
Joined: Mon Dec 31, 2018 11:35 am

Re: Flee Fail %

Post by Paladyr » Tue Aug 24, 2021 1:41 pm

The flee code is really simple:

Regardless of the number of exits in your room, the code picks a random direction (n,s,e,w,u,d). If that exit is one to which you can move (it's an actual exit, it isn't blocked by a closed door, you aren't fleeing onto water, etc.) then your flee is successful. If not, it tries again (up to 6 times). If the code can't find you an exit in which to flee after 6 attempts, you fail the flee.

UPDATE: Obviously, you cannot flee if berserk. If you are in combat, there is also a mood check at the same time the code checks that a room is one into which you can flee. It essentially chooses a random number from 0 to your mood value (wimpy=0, normal=1, brave=2). Only a result of 0 succeeds. So if you are brave, you have a 1/3 chance of passing the mood check, 1/2 for normal, and wimpy always passes. Keep in mind that passing the mood check doesn't guarantee the flee is successful; it has to be combined with the code choosing a valid exit for you to flee to.

Harun
Posts: 168
Joined: Wed Sep 02, 2015 9:45 pm

Re: Flee Fail %

Post by Harun » Tue Aug 24, 2021 1:43 pm

Sweet. Thanks! Way more simple than I imagined. Does mood change the number of times it tries again?

Paladyr
Posts: 40
Joined: Mon Dec 31, 2018 11:35 am

Re: Flee Fail %

Post by Paladyr » Tue Aug 24, 2021 1:49 pm

I missed the mood check. I have updated my previous post.

Post Reply