Review Geelong defeats Hawks at KP by 61 points

[Strange Cat]

It's a fact daz - we do not transform out performances at home to away - it's the reason we continuously bomb out in finals and have such a shit finals record. We finished top last year and all year I was telling you all to calm the fu** down and not put so much weight on our home wins - you all said blalala negative blalala and look what happened in finals - exactly what I said would happen. I want to see more input from our younger guys - in a year or two we will be in a big pile of shit if the older guys fall off or retire.

Right on cue

at 2.20 in the morning too

 

[Stan The Caddy]

Geelong are a very rare case. Our game plan is perfect for KP and sucks at the MCG. Let's see how the match results bear this out.

Not really. Our results at the G were pretty good last year mate.

 

[Cataholic]

Have a cry guys.
Have you never played footy?

There's grey areas. And it should always be that way.
It's interpretation.

Sooks.

A sling tackle is actually one of the most definitive, identifiable actions you can see on a footy field. It can't be mistaken for anything else, especially the 'two action' tackle.

Now, the AFL has clearly said that it wants the sling tackle eliminated from the game because of the unacceptable and inherently high risk of serious injury that accompanies it.

If they were serious, the mere use of that action would automatically incur a week's suspension. And if the tackled player is injured, then they should incur additional weeks in accordance with the severity of the injury.

That would eliminate the sling tackle almost immediately.

Pretty simple really. Nothing grey about it.

But as usual the AFL has done no more than pay lip service to a serious issue.

 

[you pick one]

It's this simple for me reverse the roles and Paddy wouldn't be playing next week.

 

[CatToTheFuture]

Three years ago I'd agree, but ML has come a long way, you would train it up with literally thousands of hours of video - frameworks such as TensorFlow or Gluon would be your starting point. Your test your models with new videos, test, adjust and then test again. And the price of doing all of this has gone down massively, thanks to turf wars between Microsoft, GCP and AWS.

Maybe, but do we actually have thousands of hours of clean video feed, where there's no commentary recorded over the top? What about splitting it by team, so you don't have the crowd chanting "GEE-LONG" *clap* *clap* *clap* at a North vs Collingwood match. So you'd need 18x thousands of hours and especially running it off raw video with no defined trigger conditions you'll 100% have random booing or cheering for no reason at all.

Algorithms have gotten better but as far as I know they still work with the same architecture.

Not naming companies or anything here but one example I know of outsourced the implementation of a truck driver fatigue monitoring system. Was supposed to be running raw video feed of driver's face into an algorithm looking at blinking gaps, wandering eyes etc. Had supposedly been developed to work this way in other sites/industries effectively. What was delivered 3 years later was a video feed sent to some human monitors in South East Asia who literally just watched to see if they looked fatigued. Program was pulled before implementation.

You could generate an algorithm in a week. Getting it to do what you want it to is unlikely in they timeframe

 

[jim boy]

Maybe, but do we actually have thousands of hours of clean video feed, where there's no commentary recorded over the top? What about splitting it by team, so you don't have the crowd chanting "GEE-LONG" *clap* *clap* *clap* at a North vs Collingwood match. So you'd need 18x thousands of hours and especially running it off raw video with no defined trigger conditions you'll 100% have random booing or cheering for no reason at all.

Algorithms have gotten better but as far as I know they still work with the same architecture.

Not naming companies or anything here but one example I know of outsourced the implementation of a truck driver fatigue monitoring system. Was supposed to be running raw video feed of driver's face into an algorithm looking at blinking gaps, wandering eyes etc. Had supposedly been developed to work this way in other sites/industries effectively. What was delivered 3 years later was a video feed sent to some human monitors in South East Asia who literally just watched to see if they looked fatigued. Program was pulled before implementation.

You could generate an algorithm in a week. Getting it to do what you want it to is unlikely in they timeframe

The fact that you are using terms like 'algorithms' kind of shows how far you are behind in the tech. Quoting something from more than 3 years ago is also a complete waste of time - things have changed massively in that time. I agree that getting video feeds is the biggest issue, but every match of every match is recorded and probaly has been for the last 20 years - its a good base to start. Alot more video of matches happens at matches than you'd think. For example behind every goal, there are HD cameras getting downfield perspectives. Once you have that data, scaling up the resources to train is quite doable at very short notice

 

[Mr Meow]

Not really. Our results at the G were pretty good last year mate.

Won 4, lost 3

Not bad - just lost the ones that mattered. Over the past 5 years, what has our win-loss against top 8 teams at MCG been vs top 8 teams at KP? I don't have the stats on hand but I reckon there's a fair old difference there. Superb at KP; sub-par at MCG (relative to our elite ladder positions - top 2 on three occasions in the past four seasons).

 

[CatToTheFuture]

The fact that you are using terms like 'algorithms' kind of shows how far you are behind in the tech. Quoting something from more than 3 years ago is also a complete waste of time - things have changed massively in that time. I agree that getting video feeds is the biggest issue, but every match of every match is recorded and probaly has been for the last 20 years - its a good base to start. Alot more video of matches happens at matches than you'd think. For example behind every goal, there are HD cameras getting downfield perspectives. Once you have that data, scaling up the resources to train is quite doable at very short notice

How many of those video feeds are actually stored and archived? My guess is not many and certainly not the 1000s of hours required. Even if they had it it would take thousands of hours just to align those video feeds with the live footage to get the input and output aligned.

The project I mentioned finished up maybe 18 months ago. I wrote the capital application and made claims like you are now 4 years ago. The software and resources can get better but they still have to work with a defined input and measured output that we almost certainly can't provide right now.

Also, I'm not saying it's not possible, I'm saying it's not possible in the timeframes required to get it to market before crowds are back. There were 2000 at Adelaide oval and you could easily hear booing and chants of "POOOOOW-EEEEERRR" over the fake noise which would have contained neither. Did you see the crowds at the rugby in NZ? SA and WA could probably allow that now, everyone is just waiting on Vic and NSW to get their shit together which is probably only a couple of weeks away now.

 

[iameviljez]

It should be really simple.

The mere act of a sling tackle should incur an automatic week. Period.

If a player is injured as a result of the tackle, the penalty should then be increased in accordance with severity of the injury.

That's it.

The problem with this is that pinning the arms and rolling to the side is exactly the correct tackling technique in order to avoid hands in the back. It gets very damned-if-you-do.

 

[iameviljez]

Won 4, lost 3

Not bad - just lost the ones that mattered. Over the past 5 years, what has our win-loss against top 8 teams at MCG been vs top 8 teams at KP? I don't have the stats on hand but I reckon there's a fair old difference there. Superb at KP; sub-par at MCG (relative to our elite ladder positions - top 2 on three occasions in the past four seasons).

I reckon this has been the case for all-time, though. Even when we were steamrolling sides in 07-08, we'd double-steamroll them at home.

It's kindof hard to pick the stats, though, because we play substantially weaker sides at KP.

 

[iameviljez]

Yeah Narkle stays. Rohan out Steven in!

I know he's playing the Menzel role of sitting extremely close to goal and with added ability to chase, but getting just half-a-dozen touches in a (nearly) full game of footy is wearing very, very, very thin.

 

[jim boy]

Maybe, but do we actually have thousands of hours of clean video feed, where there's no commentary recorded over the top? What about splitting it by team, so you don't have the crowd chanting "GEE-LONG" *clap* *clap* *clap* at a North vs Collingwood match. So you'd need 18x thousands of hours and especially running it off raw video with no defined trigger conditions you'll 100% have random booing or cheering for no reason at all.

Algorithms have gotten better but as far as I know they still work with the same architecture.

Not naming companies or anything here but one example I know of outsourced the implementation of a truck driver fatigue monitoring system. Was supposed to be running raw video feed of driver's face into an algorithm looking at blinking gaps, wandering eyes etc. Had supposedly been developed to work this way in other sites/industries effectively. What was delivered 3 years later was a video feed sent to some human monitors in South East Asia who literally just watched to see if they looked fatigued. Program was pulled before implementation.

You could generate an algorithm in a week. Getting it to do what you want it to is unlikely in they timeframe

Nah - the technology has changed quite considerably, 4 years is generations ago. And we're talking about very different beasts - a project like truck faitgue monitoring has to work off very small indicators, the grip on the steering wheel, the movement of the eyeballs, the heart rate of the driver as well as being influenced by things like what has the driven eaten or drunken recently as well as characteristics of their own physiology. It's not a project that is achievable now nevermind four years ago, you'd still have to go to mturk or whatever. But crowd noises are different, accuracy required is a lot lower, there isn't a single answer of how loud a crowd should be. Also the inputs are considerably less nuanced, "home team (90% of crowd) has ball and is close to goal" and it doesn't matter if their is some major inaccuracies - why not have GWS supporters as vocal at a match at Optus Oval as a West Coast? - it isn't going to affect the result in any way.

 

[Cataholic]

The problem with this is that pinning the arms and rolling to the side is exactly the correct tackling technique in order to avoid hands in the back. It gets very damned-if-you-do.

You can pin a guy, turn him, and then drop to the ground. No problem.

It's the intentional sling - which is distinctly different and more violent than a 'roll' - when turning him, the second action, that is dangerous but avoidable, and should be punished immediately.

 

[CatToTheFuture]

Nah - the technology has changed quite considerably, 4 years is generations ago. And we're talking about very different beasts - a project like truck faitgue monitoring has to work off very small indicators, the grip on the steering wheel, the movement of the eyeballs, the heart rate of the driver as well as being influenced by things like what has the driven eaten or drunken recently as well as characteristics of their own physiology. It's not a project that is achievable now nevermind four years ago, you'd still have to go to mturk or whatever. But crowd noises are different, accuracy required is a lot lower, there isn't a single answer of how loud a crowd should be. Also the inputs are considerably less nuanced, "home team (90% of crowd) has ball and is close to goal" and it doesn't matter if their is some major inaccuracies - why not have GWS supporters as vocal at a match at Optus Oval as a West Coast? - it isn't going to affect the result in any way.

Yeah that's fair enough on the accuracy requirements of the outputs but I still think it take a bit of work to clean up the input data and time it in properly against the events on the video stream, and that's assuming we have that input data to begin.

 

[Mr Meow]

I reckon this has been the case for all-time, though. Even when we were steamrolling sides in 07-08, we'd double-steamroll them at home.

It's kindof hard to pick the stats, though, because we play substantially weaker sides at KP.

Can't dispute any of that. Maybe my best way of summarising is that I think we've had a damn impressive record there under Scott with what I consider good but not great sides. Our invincibility there has given us that extra push from a 5-8 side (if we had a normal but not ridiculous dominance at home) to a 1-4. Again, the great shame being then it's time to play home games at the MCG. The competition is way weaker than 07-11 but we still haven't been that close to overcoming that obstacle with any side since.

Anyway it's still a good thing. We just need to get better elsewhere, which we've been saying for at least 8 years now.

 

[cats_09]

Dumb question: do we know yet whether we're one of the teams who gets nine home games or whether we'll only have eight? Has that fixture detail been released yet?

It may depend on what happens with the potential WA hub - Scott has said we'd be willing to participate in a hub if required

If that happens, one would hope that it means reduced need to further interstate travel and because we'd likely have one "home" game while in the hub, that we'd be included in the 9 home games group & actually get 8 games at GMHBA

 

[you pick one]

I'm thinking we may go in unchanged this week.
Rohan or Guthrie are the two possibly under threat from Steven.
Bews or Henry from Clark.
Ratugolea from Jenkins but he's injured I believe.
Yeah looks like a no change to me.

 

[you pick one]

It may depend on what happens with the potential WA hub - Scott has said we'd be willing to participate in a hub if required

If that happens, one would hope that it means reduced need to further interstate travel and because we'd likely have one "home" game while in the hub, that we'd be included in the 9 home games group & actually get 8 games at GMHBA

I'm hoping since we have been to the West rd1 that will be it for the home and away this season.

 

[cats_09]

I'm hoping since we have been to the West rd1 that will be it for the home and away this season.

We headed to Sydney in round 1 - our trips out West weren't until rounds 3 & 6

Not sure a trip to play GWS will save us from further interstate travel this season

 

[Kingsfoil]

It may depend on what happens with the potential WA hub - Scott has said we'd be willing to participate in a hub if required

If that happens, one would hope that it means reduced need to further interstate travel and because we'd likely have one "home" game while in the hub, that we'd be included in the 9 home games group & actually get 8 games at GMHBA

I thought it had been confirmed very early that all of Geelong's home games this season would be played at KP. Have the AFL back flipped on that already?

 

[thecatattack]

Dumb question: do we know yet whether we're one of the teams who gets nine home games or whether we'll only have eight? Has that fixture detail been released yet?

We 100% will be IMO.

We play 3 games in Geelong and 1 in Melbourne these first 4 rounds.

They'll send us over and act like they're sending a big Vic club over even though they're not sending the Tigers, Pies etc

 

[cats_09]

We 100% will be IMO.

We play 3 games in Geelong and 1 in Melbourne these first 4 rounds.

They'll send us over and act like they're sending a big Vic club over even though they're not sending the Tigers, Pies etc

The Tigers, Bombers & Melbourne wouldn't be unlikely to involved in a WA hub as they have either already played one of the WA sides or are scheduled to play them as with Eagles vs Tigers in round 5 - little point sending a team that can't be fully involved in the round robin format

And like Scott, Buckley has also said the Pies would be comfortable to be involved in a hub - so could see Cats, Pies, and then maybe North, Dogs, maybe even Sydney

 

[Whiskers]

I thought it had been confirmed very early that all of Geelong's home games this season would be played at KP. Have the AFL back flipped on that already?

Yes all our home games will be at GMHBA but things can change and not because the AFL are being pricks but due to circumstances.

If we hub in WA than we can’t play our home game in Geelong obviously.

Things can and will change over the course of the year.

 

[cats_09]

I thought it had been confirmed very early that all of Geelong's home games this season would be played at KP. Have the AFL back flipped on that already?

I don't remember reading that it was confirmed all our games would be at KP - I thought that Cook was going to push for that though

But if we end up in a hub, then we likely have no choice but to have a "home" game while in the hub - all of Crows, Freo, Port & WCE have one "home" game while in the Qld hub

Things will depend on what's happening with state borders and quarantine requirements - if WA opens their borders or removes the quarantine requirements meaning that teams can fly in-fly out for games, then a hub may not be required, which changes things again

At this stage the AFL is hopefully of releasing the next 2 weeks of fixtures during the coming week, and that may see the SA teams in a Sydney based or Melbourne based hub, because after that they won't need to quarantine in SA when they get home

 

[Kingsfoil]

As the Round 2 Fixtrure was announced, it was also announced that Geelong would play all its Home games in Geelong this year, as stated here. It caused a bit of an uproar at the time, upsetting the other clubs being used to Geelong being shafted having to play Home games elsewhere now finally getting what they should get every year. Playing a 'home' game anywhere else would now be a back flip. If the AFL stick to what they announced, then any games Geelong played in an interstate hub would have to be classified as 'Away' games. But it's the AFL, so they'll probably back flip whenever it suits them.