zulumoose wrote:The simple fact of the matter is that although at some stage in the future a long range sniper shot will be something that a computer operated rig will do better than a human is capable of, that is not the case now.

The reason that is not the case now is that humans achieve accuracy through a wide range of estimations that they can personally get better at, but cannot envisage how to program a machine to replicate to the same degree. Machines are not inherently good at estimation based on incomplete and variable data in a non-repetitive scenario that can't be replicated.

It is not that computers aren't capable of better results, it is that we cannot show them how to do so yet, because the parameters are too fuzzy.

Nice summation of just what I've been saying. Therefore, I predict it will have no impact on the resident arguer from ignorance.

Yes you provided one variable but I do not see how, in principle, its impossible for AI to determine it better than human. But but .. write the code! Its not on the market yet! I said in principle.

You remind me of a friend of mine, chess player, who insisted that because there is no solution for chess yet it will remain human domain. I guess it made him uncomfortable that what he believed to be his "strategy", his special sauce, was merely a crutch for his limited capabilities.

It will always be human domain, because if a machine can be programmed to do it better than any individual human, it is because humans figured out how to program it to get that good.

Cell phone communications are immeasurably better than human to human communications through shouting, but it is still a human achievement, we haven't surrendered to machines, just become more reliant on using them as a tool.

That settles it then lol We are talking about AI being better at something than human remind you. Its ok with me that you picked "human domain", out of context, and got creative with it as this could possibly remain human domain - babbling and bickering.

zulumoose wrote:The simple fact of the matter is that although at some stage in the future a long range sniper shot will be something that a computer operated rig will do better than a human is capable of, that is not the case now.

The reason that is not the case now is that humans achieve accuracy through a wide range of estimations that they can personally get better at, but cannot envisage how to program a machine to replicate to the same degree. Machines are not inherently good at estimation based on incomplete and variable data in a non-repetitive scenario that can't be replicated.

It is not that computers aren't capable of better results, it is that we cannot show them how to do so yet, because the parameters are too fuzzy.

We don't need to hardcode. If you have enough data, processing power and a simple reward function, then you can sometimes go a long way with a blackbox learning algorithm. However, it wouldn't surprise me if there are all sorts of subtleties involved in the particular case of machine vision.

Problem for programming computers to give them the ability of a top sniper, or the best in any field really, is how often people have little to no idea how they do what they do, it's the brain's amazing pattern recognition abilities that are often unconscious, you sense and/or act without real knowledge of what you're doing. E.g., the swirling patterns of grass, there might be some rules of thumb, but what is really going on is they look at the scene and there is an adjustment that feels right. I'm guessing based on other similar skill sets, I don't know anything about actual sniping, except the snarky kind.

If computers can learn this, it's going to be some kind of neural network with training and they won't understand all that well what the fuck the machine is doing.

There isn't a report on the wind. It really doesn't matter how good you are, there's going to be a shitload of luck involved in such a shot. All I saw was someone saying ""It is at the distance where you have to account not just for the ballistics of the round, which change over time and distance, you have to adjust for wind, and the wind would be swirling," said a source with expertise in training Canadian special forces." But no info on what the actual conditions were. And my earlier reply wasn't entirely facetious, how many shots have these guys tried and missed?

tuco wrote:Yes you provided one variable but I do not see how, in principle, its impossible for AI to determine it better than human. But but .. write the code! Its not on the market yet! I said in principle.

You remind me of a friend of mine, chess player, who insisted that because there is no solution for chess yet it will remain human domain. I guess it made him uncomfortable that what he believed to be his "strategy", his special sauce, was merely a crutch for his limited capabilities.

Ah, yes - "in principle" again, the ultimate get out of jail card.

And again, the reason that it is a problem is that we don't live in an theoretical world, and there are too many variables to quantify.

You are living in the land of "Wouldn't it be nice if ..." and trying to insist everyone else live there too.

Why do we celebrate these long range shots lol

I bet the isis member was thrilled to be part of the historic record breaking attempt. Can we read about how proud his family is? 'You know I thought mohammed would never amount to anything, but then I read this - over two miles amazing. Mo would have been so chuffed. I just hope all the fame doesn't go to the guy who pulled the trigger, mo did his bit too, standing in just the right place for ten whole seconds adjusting for wind the whole time. Mo couldn't read and was easily led but just goes to show you can achieve anything regardless of how you start out'.

CarlPierce wrote:I bet the isis member was thrilled to be part of the historic record breaking attempt. Can we read about how proud his family is? 'You know I thought mohammed would never amount to anything, but then I read this - over two miles amazing. Mo would have been so chuffed. I just hope all the fame doesn't go to the guy who pulled the trigger, mo did his bit too, standing in just the right place for ten whole seconds adjusting for wind the whole time'.

Assuming it was an ISIS member.

Teague wrote:

CarlPierce wrote:I bet the isis member was thrilled to be part of the historic record breaking attempt. Can we read about how proud his family is? 'You know I thought mohammed would never amount to anything, but then I read this - over two miles amazing. Mo would have been so chuffed. I just hope all the fame doesn't go to the guy who pulled the trigger, mo did his bit too, standing in just the right place for ten whole seconds adjusting for wind the whole time'.

Assuming it was an ISIS member.

Well he mostly carried ammo around and fetched the water. But he had ambition.

Teague wrote:Why do we celebrate these long range shots lol

I'm not sure it's a celebration. A recognition of skill perhaps.

And sorry, I don't get the all the "What about the poor ISIS soldier" posts. I'm guessing this wasn't just any enlisted man to have gone through all this effort.

Whether the shot should have been taken, or whether the target was legitimate, those are completely separate issues for another thread.

What is of note is that boundaries of what is possible, or practical, or achievable (take your pick) have been pushed back, and there are interesting technical aspects involved for those who are that way inclined.

The Canadian sniper was from JTF2. Seems that if they're on your case, you are involved in more than just hauling water.

https://en.m.wikipedia.org/wiki/Joint_T ... Operations

zulumoose wrote:Whether the shot should have been taken, or whether the target was legitimate, those are completely separate issues for another thread.

What is of note is that boundaries of what is possible, or practical, or achievable (take your pick) have been pushed back, and there are interesting technical aspects involved for those who are that way inclined.

I'm not sure if you can separate those. Part of a sniper's training and skill is determining the legitimacy of the target.

To make that shot, the target in the scope would be tiny as the more magnification, the more pronounced any movement is. The shot took nearly 10 seconds to get there, which is a huge amount of time when adjusting for all the factors. It was certainly an impressive shot.

The guy he killed? His 2 5 year old daughters and 2 year old son we're killed by the US - they were "collateral damage" so he went to find an organisation to hit back at the people that murdered his family.

Question why not just use artillery?

Weaver wrote:

felltoearth wrote:

Shrunk wrote:I admire the feat, and appreciate that this is a very efficient way to achieve a legitimate military objective. But I also think it's weird to keep records for this like it was baseball.

I agree with this.

It isn't records like it's baseball - it's records like it's long range shooting competitions - which it is. We keep records of those who have the tightest groups at various ranges, and on who scores the most points on a course of fire, and on who has the longest shots against stationary targets, and against moving targets - and against unpredictably moving targets, i.e. sniper targets.

Many top snipers are also top competitive shooters - so it really isn't surprising at all.

So this record includes people who are shooting at non-living moving targets in non-combat situations?

This guy was so far away it took ten seconds for the bullet to reach the target. I still can't get my head around that.