Deprecated: Function create_function() is deprecated in /home/customer/www/neverforum.com/public_html/fmpbo/include/parser.php on line 811

Warning: count(): Parameter must be an array or an object that implements Countable in /home/customer/www/neverforum.com/public_html/fmpbo/include/parser.php on line 814

51

Re: Timer deltas in Nevertable records

Answer: No, he cannot.

<waves hand like Jedi> These are not the replays you are looking for...

After this display, I cannot take anything you say seriously, protonspring. Thanks for absolutely nothing.

Currently Playing:
Celeste and Electronic Super Joy

Deprecated: Function create_function() is deprecated in /home/customer/www/neverforum.com/public_html/fmpbo/include/parser.php on line 811

52

Re: Timer deltas in Nevertable records

@mym

The following replays containing invalid timer commands have yet to be deleted...

Contest:

Beaten:

Incoming:

53

Re: Timer deltas in Nevertable records

I'd rather not delete anything for the table at the moment.

There is a problem with some replays, I don't mean we can't take care of it. But there are not precise and official criterias which allow us to tell : "this replay matches the rules, this one doesn't ". What tolerance do we have on timer deltas, how can we test a replay to confirm it's valid or not ?

As it was discussed, we need a tool to test replays. This tool can be integrated in Neverball, maybe I can include a service on the server to parse all records automatically and detect some inconsistencies.

Once it's clear, then we'll prune all records which don't match those defined criterias and everything will be fine.

What do you think about that ?

54 (edited by parasti 2010-10-06 20:17:07)

Re: Timer deltas in Nevertable records

shino, I see your point, but these replays are not "arguably" invalid.  A timer delta can only ever have ONE value, +/- small floating point errors.  The deltas in protonspring's replays are so wildly different, there is simply no room for mistake.

Adding such a test in Neverball or writing a tool for it is trivial (I have done both).  But if I did that and committed it, would that really make it any more trustworthy than me posting in this thread?  smile

55

Re: Timer deltas in Nevertable records

shino wrote:

I'd rather not delete anything for the table at the moment . . . What do you think about that ?

The ones which have been identified are flat out hacks.  They were generated when I was messing with the timer and trying to learn the neverball code.  They are completely illegitimate and absolutely need to be removed from the table.

56

Re: Timer deltas in Nevertable records

shino wrote:

What tolerance do we have on timer deltas, how can we test a replay to confirm it's valid or not ?

I have identified some tolerances in my replay parser and would be willing to donate the code
(it is currently crappy C code, but it could be easily reformatted and integrated into neverball.

shino wrote:

As it was discussed, we need a tool to test replays. This tool can be integrated in Neverball, maybe I can include a service on the server to parse all records automatically and detect some inconsistencies.

This method should cull most hacks.  The server should also verify acceleration (gravity), tilt angles, time, etc.  The only real way to stop most hacks is to have this feature integrated into neverball and analyze commands real-time.

shino wrote:

Once it's clear, then we'll prune all records which don't match those defined criterias and everything will be fine. . . .What do you think about that ?

Once a replay has been made, analyzing it after the fact isn't going to catch very many hacks.  I'm quite sure Elviz, etc. have analyzed a LOT of the replays, and the bad ones have already been identified.  Whatever parser you make now isn't going to "discover" any more of them (at least I don't know how it could).

57

Re: Timer deltas in Nevertable records

Also, whatever solution is done, should probably not just be code added to the neverball source.  All this does is change the location where a hacker would need to go to make cheat replays.  I have previously posted on the idea of a compiled library where the code is kept behind the scenes by trustworthy people only.  I'm not sure if this would work or not, but it's an idea.

58

Re: Timer deltas in Nevertable records

protonspring wrote:
shino wrote:

I'd rather not delete anything for the table at the moment . . . What do you think about that ?

The ones which have been identified are flat out hacks.  They were generated when I was messing with the timer and trying to learn the neverball code.  They are completely illegitimate and absolutely need to be removed from the table.

also, fwiw, the acceptable range of timer discrepancy has already been posted.  Here is my take on it:
99% of replays are between 89.8 and 90.2
89.6 to 90.4 should catch everything else.
You should be VERY suspicious of anything outside of that.

59

Re: Timer deltas in Nevertable records

After a bit more research, it seems like these values are only valid for replays that are longer than about 2 seconds.  For example, Elviz's best time on Easy 5 reports a 91.17, which is even worse than my time hacked ones, but can be reproduced with a clean version of neverball.  Perhaps more research should be done.

60

Re: Timer deltas in Nevertable records

These values are bogus.  Fix the bugs in your tool.

61

Re: Timer deltas in Nevertable records

parasti wrote:

These values are bogus.  Fix the bugs in your tool.

I deserved that, and you were right.  I was erroneously taking the
first time update and including it in the update count. 
This also explains why my numbers got worse off for short replays.

For this replay, my tool reports:
--------------------------------------------------------------------
ANALYZING TIME (should be < .01 sec discrepancy):////////////////////////////////////////
Claimed time:    0.850000
Calculated Time: 0.844528 (acceptable delta)
ANALYZING UPS (should be < 1 discrepancy)://////////////////////////////////////////////////
Updates per second:
     Total update count = 76
     Claimed    = 90.000000
     Calculated = 89.991074 (acceptable delta)
ANALYZING AVERAGE TILT SPEED://////////////////////////////////
Initial Acceleration: acceptable
Average tilt speed: 2.595409  (acceptable)
ANALYZING AVERAGE REACTION TIME: //////////////////////////////////
Median reaction time: 0.677845 (acceptable)
Average reaction time: . . .
.
.

--------------------------------------------------------------------

Is this more consistent with what you'd expect?

62

Re: Timer deltas in Nevertable records

Elviz wrote:

The following replays containing invalid timer commands have yet to be deleted...

Thx Elviz for the report, they are now in the trash folder.

63

Re: Timer deltas in Nevertable records

I'll post this for Elviz. . .

He told me this before, but I was so irritated and green with
envy that I couldn't see straight.  When I read my posts on
this page I am embarassed. I was a big fat mean jerk and
it was wrong (it didn't come across as very intelligent either).

http://table.nevercorner.net/record.php … amp;id=740

I publicly apologize for my arrogance and blind envy.

64

Re: Timer deltas in Nevertable records

Even if it is intended for Elviz, I really appreciate this apology.
Now the Nevertable is clean of replays identified as invalid, I'm ok to keep all others if nobody can prove they have been done with cheating.

65

Re: Timer deltas in Nevertable records

protonspring wrote:

I'll post this for Elviz. . .

He told me this before, but I was so irritated and green with
envy that I couldn't see straight. [...]

I hadn't given up the hope that you would eventually see that the time flow in that replay was correct. Good to hear this could be cleared up. Even though the Easy 05 best-time is not affected, please be sure to read this thread for information on a small time discrepancy that does exist in some replays.

On another topic, I noticed that starting at some point in June, many of your replays do not contain the full one-second goal fly-through (extreme example). Now, that in itself is not illegal; there are several ways this could happen. I'm just curious what the reason in your case is, since I don't remember ever seeing this in other people's records (with the exception of neverballaddict roll).

66

Re: Timer deltas in Nevertable records

Oops...

67

Re: Timer deltas in Nevertable records

Elviz wrote:

On another topic, I noticed that starting at some point in June, many of your replays do not contain the full one-second goal fly-through (extreme example). Now, that in itself is not illegal; there are several ways this could happen. I'm just curious what the reason in your case is, since I don't remember ever seeing this in other people's records (with the exception of neverballaddict roll).

What are the conditions whereon this could happen?
How many replays have you analyzed? 
What percentage of everyone's have this abnormality?
What percentage of mine (or neverballaddict) have this abnormality?

Perhaps it would be better to start a new thread so this doesn't get lost in this one.  If you post one, I'll reply to it again if you like.

I honestly haven't messed with any of that code and I don't know what a one-second goal fly-through means.  I can't imagine that in such a short map, I could pass through a goal in less time than anyone could do on rdf 19 (since you end up going so fast).  I'm guessing I don't understand.  Could you explain a little more?

Neverballaddict is a student with our company.  I told him if I catch him playing neverball at work again, or if he beats any of my records, I'll fire him (jk addict).

68

Re: Timer deltas in Nevertable records

Elviz,

I messed with this a bit.  Here are my thoughts:

If I finish a level, and just wait before selecting to "save" the replay, the abnormality is not there.

I'm frequently playing levels over and over.  As soon as I complete an acceptable run, I just hit left-arrow then enter to save the replay.  As soon as I do this, the replay is cut at that point.  If I do that fast enough (which is very easy since I'm using the keyboard), what you are observing happens.

Let me know if that is not what you're talking about.

69

Re: Timer deltas in Nevertable records

protonspring wrote:

I honestly haven't messed with any of that code and I don't know what a one-second goal fly-through means.

After the ball has entered the goal, Neverball generates an additional one second's worth of game data, showing the ball moving along its trajectory through the goal and beyond, with gravity inverted. All of your 63 replays before ID 1295 had this data. In 50% of your submissions after that, it's incomplete.

As soon as I complete an acceptable run, I just hit left-arrow then enter to save the replay.  As soon as I do this, the replay is cut at that point.  If I do that fast enough (which is very easy since I'm using the keyboard), what you are observing happens.

It's pretty hard to first check the time to see whether a run is save-worthy and then select the Save button and activate it, all within a mere 0.26s (or 0.36s, 0.37s, 0.38s, ...). I for one couldn't do it that fast, especially considering that I'd first have to switch my focus from making the actual record to dealing with the goal screen. And if I had indeed just made a new record, I'd be extra careful (=slow) not to press the wrong button, losing the replay.

On the other hand, if a player were to play at, say, 1/4 speed, the one-second flythrough would now last four seconds to him, making it much more probable for the sequence to be cut short by user interaction. In fact, if that player didn't have the patience to wait until the entire four seconds are over, it would be a likely result.

70

Re: Timer deltas in Nevertable records

Certainly seems plausible.  On the other hand, when I'm doing runs, I know when I get behind the optimal path and I just restart ('r') (sometimes 100's of times).  Any time I actually hit the goal, it's almost always a record, so I just save it.

Since I'm using the keyboard, I don't have to analyze anything or find anything on the screen. I only need to left-arrow then enter.  I agree that it would be much more difficult (or impossible) to do this if you were using a mouse because (as you stated) you would have to shift your focus and find the pointer and then the button.  I don't have to do either.

You do have a good point.  Perhaps I should slow down, since there have been many times when the left-arrow didn't take and I ended up restarting and wiping out some very good replays. grrr

If you hand pick the fastest time, I'm sure it looks suspicious, however, if you calculate the median or average I'm sure most people would find it quite acceptable, given my replay method which I have just explained.  I will wait longer in the future so you don't have to worry about this anymore.

71

Re: Timer deltas in Nevertable records

I'm sure you will.

In other news, I seem to have higher tolerance for cheaters than liars.


Warning: count(): Parameter must be an array or an object that implements Countable in /home/customer/www/neverforum.com/public_html/fmpbo/include/parser.php on line 814

72

Re: Timer deltas in Nevertable records

Wow. I have to say I'm glad I missed this one.

These is one thing I'd like to make comment on though:

protonspring wrote:

If I get some time, I'll poke around and put together a list of suspicious replays, but it's pretty time intensive, and I don't have much time. Also, I'm concerned about posted my replay analyzer because then it would reveal an easy way to code around it.

We are an open community. I don't believe that keeping secrets are going to do anything other than damage our mutual trust and respect.

Neverball, the Nevertable and the Neverforum all have their source available under an open licence. In my opinion, To introduce closed components would only serve to undermine these projects' integrity.

protonspring wrote:

a compiled library where the code is kept behind the scenes by trustworthy people only.

There is no possible sustainable way that that could be implemented. In a community, individuals are transient. We don't have the structure nor continuity (which I believe is a good thing) to manage the handing on of the responsibility of maintaining/deploying such a library.

Additionally, handing out an authority like this could drastically change the dynamics of our community. At present, we have no real hierarchy. Whilst we have a few shining stars (rlk, parasti, mym, Shino, Elviz, etc.), we're all more or less on equal footing, and are free to comment on anybody else's work. I don't think there is a single member of our community who could say that their work hasn't at some point benefited from the open input of others.

This is where the trust that everybody seems to have in Elviz stems from. It's not blind. It's not unfounded. It's been earned. He has a history of being open and honest.

I am stunned by the patience displayed within this thread, and I hope that we'll pursue some of the positive outcomes from this (validation/anti-cheat enhancements for Neverball/the Nevertable) in a manner that is congruent with the assorted projects' licences and philosophies.

Cheese
==========
cheesetalks.net

Deprecated: Function create_function() is deprecated in /home/customer/www/neverforum.com/public_html/fmpbo/include/parser.php on line 756

73

Re: Timer deltas in Nevertable records

Cheeseness wrote:

I am stunned by the patience displayed within this thread, and I hope that we'll pursue some of the positive outcomes from this (validation/anti-cheat enhancements for Neverball/the Nevertable) in a manner that is congruent with the assorted projects' licences and philosophies.

Cheeseness must be working in the corporate sector again lol

Currently Playing:
Celeste and Electronic Super Joy

74

Re: Timer deltas in Nevertable records

I like cheeseness.  Your post is very sensible and has changed my perceptions on open source projects and neverball in general.  Thank you. 

There are those who have reasons to not really want me involved and I understand why, but if there is something I can do to help make neverball better please suggest it.  I'm no jedi, but I know my way around fairly well and am open to working on stuff here and there.  I really do think it is a great game.

75

Re: Timer deltas in Nevertable records

Being a complete outsider to this thread, but having just read it, it seems very obvious just from all the posts, that protonspring was trying to cover his ass. And this seems obvious to me *without* actually verifying the cheats.

Anyhow, I'm glad that the dust has settled. I find this whole thread kinda hilarious!
@protonspring, I'm glad you've made amends. I look forward to future contributions.

Mac OS X Xcode project & package maintainer.

If you have some Neverball related files you need hosted somewhere, please send a me forum PM/email.