Rev3 Up and Running - Complaints, Ideas, Ramblings, Comments ... but no bugs

[knopflerbruce]

http://www.hwbot.org/listResults.do?cpuModelId=256&applicationId=3&filterUser=true&filterBlocked=true&limit=100

 

~55 scores from different users, still just 3 boints for the top spot.

[SoF]

1) Kingpin benched the 1x5870 a couple of days before the Rev3 was launched. His margin is solely caused by him preparing for Rev3. Give the ranking a month and then you can make this kind of claims

 

3) No one obliges you to be active at hwbot.

 

1) In the old rev that big margin wouldn't have been possible at all. Furthermore you made all work of this years worthless (!!!) with rev3 when Vince can do that kind of rampage within one session now...others can do too, I agree, but simply the fact that the entire ranking can be crushed by him with one GPU is so sad...poor boris...

 

3) You think it is easy for me seeing 2 or 3 years of work gone within a day? I was happy with rev2, very happy. I told you before that I am unsure if the changes will turn out good, I made my proposals but they did not get heard. I am just so stunned (and shocked) that you are now basing everything on how popular it is, not how high the scores are...you are trying to reorder the entire oc-community with the new point system and I hope they will NOT follow, you will have to revise everything (just give every record-class the same amount of points - I am almost instantly happy...) and we will all be friends again

[Massman]

less than 10 submission rev3. point system less points but not 0,1 points for an second place

10-19 submission a mixture of rev2 and rev3 with a firm placing points

over 20 submission rev2 with in max hwpoints of 75

 

Because we prefer to understand the algorithm as well.

 

Rev3 is still based on the same principle as Rev2 ... the algorithms have the same mathematical grounds.

[speddy411]

dude... http://www.hwbot.org/hardware/videocard/quadro_nvs_295?tab=hall%20of%20fame

 

What he means is that the second or third gets nearly nothing...The gap between points is to big. It should be like many said before: 3;2.8;2.6;2.4.........

[der8auer]

If there's no competition, the 1st gets 1,5p. If there are 10 competitors you have >2p.

 

Okay - if there are 10 competitors:

 

It goes:

 

1st - 2p

2nd - 1,5p

3rd - 1p

4th - 0,5p

5th - 0,1p

6th - 0,1p

7th - 0,1p

.

.

.

right?

 

why not going:

 

2 - 1,9 - 1,8 - 1,7 - 1,6 ......?

[Massman]

1) In the old rev that big margin wouldn't have been possible at all. Furthermore you made all work of this years worthless (!!!) with rev3 when Vince can do that kind of rampage within one session now...others can do too, I agree, but simply the fact that the entire ranking can be crushed by him with one GPU is so sad...poor boris...

 

Boris is still second in the ranking. Only thing he has to do is test 1 5870 in 6 benchmarks and he'll be incredibly close to KP again.

 

By the way, if you state that his work has been worthless, you couldn't be more wrong. I think Boris did a series of inspiring benchmarks over the last couple of months/years ... it's not because there's a fallback on the day the new revision is launched, he's doomed to drop forever.

 

Please get a grip on yourself and start putting things in perspective. You're jumping to conclusions WAAAY too fast ... the revision is not even up 24h.

 

3) You think it is easy for me seeing 2 or 3 years of work gone within a day? I was happy with rev2, very happy. I told you before that I am unsure if the changes will turn out good, I made my proposals but they did not get heard. I am just so stunned (and shocked) that you are now basing everything on how popular it is, not how high the scores are...you are trying to reorder the entire oc-community with the new point system and I hope they will NOT follow, you will have to revise everything (just give every record-class the same amount of points - I am almost instantly happy...) and we will all be friends again

 

Can I bring up the Wprime example just one more time to demonstrate how "score > popularity" is not per definition the way to go.

 

I hope all your wishes for the year 2010 may come true!

[Lippokratis]

dude... http://www.hwbot.org/hardware/videocard/quadro_nvs_295?tab=hall%20of%20fame

 

try to say something about the others things i said in my posts not just answer one single point.

 

 

on this rev it is possible that 400 submisiion of unoverclocked hardware are making more points than 100 submmision of overclocked hardware. more submission doesn't meen competition it just means that many peole use this hardware

[Massman]

Why not going:

 

2 - 1,9 - 1,8 - 1,7 - 1,6 ......?

 

Because it doesn't take a lot of skill to be 5th in a ranking of 10 . What you display here is exactly what people complained about in Rev2: no-skill runs get more points than skill-runs.

 

By the way, has anyone noticed that the number of hardware categories has gone UP, UP, UP!

 

In the past, there was

 

-3D: single/multi => 1x/2x/3x/4x

-2D: single => 1x/2x/3x

 

everyone seems to be forgetting that we have split up 2 3D rankings into 4.

[Massman]

so when you win first place, you should not get noticeably more than 2nd place?

there should be a drop off between 1st and others, why try to be first?

 

 

This makes me remind the "there's no motivation"-claim from earlier in this thread. Tbh, if I'm second, why would I bother to push harder ... the difference is only 0.1p, right?

[thebanik]

Because it doesn't take a lot of skill to be 5th in a ranking of 10 . What you display here is exactly what people complained about in Rev2: no-skill runs get more points than skill-runs.

 

By the way, has anyone noticed that the number of hardware categories has gone UP, UP, UP!

 

In the past, there was

 

-3D: single/multi => 1x/2x/3x/4x

-2D: single => 1x/2x/3x

 

everyone seems to be forgetting that we have split up 2 3D rankings into 4.

 

but sadly, you have gone from 2 Global rankings to 1 Global rankings, and those extra categories mean shit to most of us, since they wont earn any points......You guyz are pretty rigid it seems, coz you have almost made up your mind that there is nothing wrong with the new rev....(let me again clarify mid-level ocers like me would be benefited out of this but still its not the way to go forward)

Edited December 31, 2009 by thebanik

[Massman]

but sadly, you have gone from 2 Global rankings to 1 Global rankings, and those extra categories mean shit to most of us, since they wont earn any points......

 

2 -> 1 = 1 less card to buy to be able to compete

 

"Most of us" are complaining about the hardware points here.

[der8auer]

Because it doesn't take a lot of skill to be 5th in a ranking of 10 . What you display here is exactly what people complained about in Rev2: no-skill runs get more points than skill-runs.

 

Yea i agree on that but i think the "bad" scores should be more rewarded, too. Place 6 is better than 8 but you get 0,1p for both. Thats the point i'm on about

[speddy411]

Because it doesn't take a lot of skill to be 5th in a ranking of 10 . What you display here is exactly what people complained about in Rev2: no-skill runs get more points than skill-runs.

 

 

Because of that those people just earn a few points and not 75 or 50, but its not nice just to give number 1 2 points and all the other people who are just seconds away from #1 get nothing...

 

//der8auer was faster

[Advanced]

so when you win first place, you should not get noticeably more than 2nd place?

there should be a drop off between 1st and others, why try to be first?

 

Then do 2,5 / 1,5 / 1,3 / 1,1 / 0,9 / ...

 

Whats kind of unfair is, that only few people get noticable points.

 

#1, #2, #3, #4 get points and the rest gets like nothing. The gap is too big, that's the point.

 

Edit: It's not like that I do not respect the effort and work that you've put into Rev3.

I only want to improve it so that as many people are satisfied as possible.

Edited December 31, 2009 by Advanced

[thebanik]

2 -> 1 = 1 less card to buy to be able to compete

 

"Most of us" are complaining about the hardware points here.

 

or maybe rev3 is broken in more ways than 1, please see my earlier post in the last page where I tried to reason that this is not the way to go forward, posted again below......

 

Thats exactly what I have done honestly since I understood this part completely, got a 5870 from my limited sponsor, used DICE and wallah I have 125 global points where as I had in total half of it using 2 x 5870 and my other 200 submissions. But now take this, I am sorta of a guy who gets very limited support from Hardware companies, 5870 was not a problem, but getting 5970 is/was a PITA but I would have persisted with my links to get me a/couple of 5970, if I didnt get any support I would have then put my own money and bought one or maybe 2 of them so that I could have tried reaching the top, ofcourse the TOP Most guyz you are posting again and again would certainly go for overall WR's and still continue to do so even without the boints but for mid level ocers like me, I wouldnt touch 5970's even with a barge pole, now honestly tell me guyz, you think this is the way to go forward????

[Massman]

Yea i agree on that but i think the "bad" scores should be more rewarded, too. Place 6 is better than 8 but you get 0,1p for both. Thats the point i'm on about

 

That may be somewhat fixable, but not in the large margin as you've suggested above. Fact is that from our 2y experience with Rev2, we know that people will start complaining that 0-skill scores get 0,5p or more.

 

The algorithm wasn't designed in one day; we do have a lot of background from the previous revisions

[knopflerbruce]

so when you win first place, you should not get noticeably more than 2nd place?

there should be a drop off between 1st and others, why try to be first?

 

Exactly. It takes an effort to be first, and that should be rewarded. Receiving 75% of the reward the best result gets when you're #5 out of 10 is far from fair.

 

Perhaps a better approach would be to reward by comparing the results to the average, or something... if you're 20% better than 2nd place (and even more to the rest), then you should get a much higher reward than if you're just 2ms better in some test. That would be rev. 4 material, though...

[fm7272722]

So is it intetntional that the ranking pages now do no longer show what type of cooling was used? I really miss that info.

 

[romdominance]

- Now you need 1x5870 to grab a set of global points, in rev2 you needed 2x5970. More money? huh?

 

- Why do you want the most popular categories? Go for the mid-popular categories: still fair amount of points and not too difficult to get into top10.

 

Good morning, the new site appearance looks great. Unfortunately thats where the optimistic feelings end with me and a great many of my teammates over at NBOC. Here is a brief list of the downsides to the new Rev3 IMO:

 

1) What happened to teams and participants like CP is flat out wrong. It would have been better to create a separate league for HW crunchers and let them pound it out amongst themselves than gut their personal and team points under the flag of fairness....during halftime no less.

 

2) The new point system was sold to us a being more transparent and fair than the previous rev2 system. It is neither IMO.

 

The weight system and unique users makes it impossible to anticipate where we would land with this new revision.

 

The new "fairness" is not IMO as it rewards mediocrity and the average, while adding little or no incentive to go for the best overall score.

 

This feels very much like its not about the competition anymore, its about benching only the most popular hardware and in your words Mass "Go for the mid-popular categories: still fair amount of points and not too difficult to get into top10"....don't you see that your needing to direct people in what hardware to use to get max boints instead of telling them "bench the hell out of the best stuff you can get and see how you measure against the pros", turns hwbot into a "no one fails, everyone who comes thru the door gets points, don't try and be the best, just use what everyone else is using" environment.

 

This revision didn't personally hurt my profile as most of the guys above me took the same hit. What it did is:

-remove CP as our main team competition and motivation.

-Made it worthless for most of us with Classified SLI XF boards and 1-2x ATI 5970's and/or 3-4x ATI 5870's to bench them anymore for boints.

-It made the hardest working/benching members on our team ask why they bothered. -Exalted the lowest members on our teams, by taking from the top and adding it to the bottom.

-Window dressed up the league for new participants while really not caring about the guys who have been with you the longest.

 

I don't think that you are purposely doing this to irritate the majority of your participants, as I think that you work hard and have the best intentions for fairness. This new revision though seems to have sorely missed the mark.

 

Sorry about the rant, thanks for caring and for all your hard work....I know you guys will get it right

 

Regards, Rom

[Massman]

or maybe rev3 is broken in more ways than 1, please see my earlier post in the last page where I tried to reason that this is not the way to go forward, posted again below......

 

What I read in that quote is that you want loads of points for doing little effort to push the card. I know it would be much more fun if all the 21,000 overclockers here at hwbot would be in the top20 ... but that's not how a ranking works .

 

As far as I can see, you might be overestimating the people who use LN2 and push the card to really sick speeds. The maximum global points have been increase in Rev3 as well, so being top50 should give you more than in the previous revision.

 

I'd suggest to do the normal benching in January and we can check back how things are in February. Major flaws will be addressed for sure, but as I've said before: the Rev3 have only been up 10h ... it's impossible to judge what's a flaw and what's not.

[Alriin]

Sorry..., I love HWbot, but Rev3 is a extrem mistake. A lot of People are unmotivated now.

 

All what i buy for HWbot, all what i work for HWbot, is destroyed in 1 Minute!

 

And when i only become a Award for benching MSI-Boards ore Intel-CPU, its a distortion of competition.

[der8auer]

That may be somewhat fixable, but not in the large margin as you've suggested above. Fact is that from our 2y experience with Rev2, we know that people will start complaining that 0-skill scores get 0,5p or more.

 

The algorithm wasn't designed in one day; we do have a lot of background from the previous revisions

 

=) Okay thanks.

 

Like "Advanced" said:

Then do 2,5 / 1,5 / 1,3 / 1,1 / 0,9 / ...

 

Whats kind of unfair is, that only few people get noticable points.

 

#1, #2, #3, #4 get points and the rest gets like nothing. The gap is too big, that's the point.

 

Edit: It's not like that I do not respect the effort and work that you've put into Rev3.

I only want to improve it so that as many people are satisfied as possible.

 

Thats the point!

[Massman]

Exactly. It takes an effort to be first, and that should be rewarded. Receiving 75% of the reward the best result gets when you're #5 out of 10 is far from fair.

 

Perhaps a better approach would be to reward by comparing the results to the average, or something... if you're 20% better than 2nd place (and even more to the rest), then you should get a much higher reward than if you're just 2ms better in some test. That would be rev. 4 material, though...

 

we've looked into this in the past, but it's a very delicate proposal because of the outlier-problems.

[PLAGANOS]

if u get just 2 points for a lil older card, no one will buy such a card. hours of work and testing for just 2 stupid points ? that sucks !

and if no one buys such lil card, this cathegorie will never becomes more poplulare like u always say.

i bought a 7900GX2 for 60€ !!! from ebay before x-mas. i just did a lil test with AM3. my favorit benchmark.

i earn 1.6 points. damn i was so happy about that, cause it was just testing. now i got 0.1 point.

i know i can push this score higher, but for what ? just for these lil 2 points ?

 

the point is that the weight of the point system is to bad now. like lots of guys say here!

[Massman]

The weight system and unique users makes it impossible to anticipate where we would land with this new revision.

 

Test server has been up for a week. Can't be more transparant about giving you an idea where you'd land.

 

The new "fairness" is not IMO as it rewards mediocrity and the average, while adding little or no incentive to go for the best overall score.

 

So your idea is that the new revision rewards mediocrity and the average whereas I've got several people complaining that we only award the top spots?

 

This feels very much like its not about the competition anymore, its about benching only the most popular hardware and in your words Mass "Go for the mid-popular categories: still fair amount of points and not too difficult to get into top10"^

 

you couldn't describe a competition better than above.

 

Highly-popular categories are for those who can't compete with extreme cooling and extreme modifications, so they can go to the mid-popular categoires where the competition is still somewhat okay.

 

Want competition and loads of points => go into the high-popular categories and do everything you can to get into the top10 or higher (competing against 400+ people). Want a easier challenge, find a couple of mid-popular categories and beat the crap out of your hardware to be first.

 

It's all about competition.