Thomas EvansSpent a long time thinking about exploring again, so I spend most of my play time today traveling to Obsidian Orbital.
only to be destroyed by someone waiting for me to get there with the Thargoid probe and sensor I managed to find, along with my first contact since that update.. the experience was totally ruined [img=620x115]https://inara.cz/data/sigs/86/86078.png[/img]

Rebecca HailThat's the reason why you have minimum specifications. And if you understand how to throttle said process, you should also understand why you shouldn't throttle said process if you want to run the game with a good performance.

You misunderstood my entire point. I merely stated that you're wrong with the statement that Frontiers Servers have something to do with calculating the NPCs you fight against.

I agree that you can't develop an AI capable of anything near human performance with the tech (and budget) that's available for games.

Some Hooligandon't expect the rabble rousers of the galaxy to go quietly into the night.  It's just not our way.

Chief ComplainerIt's time to find out if those "git gud" murderhobos will cry us a river once they're faced with harder opposition of the PvE kind. I can already hear the moaning on the FD boards

The Same Hooligan
Great stuff Scubadog - don't forget high waking can't be mass locked.  If I'm entering a bad neighborhood, or know that lots of people are probably there looking to kill me, I'll pre-select a high wake target the moment I land in system.

NFC PhistoGreat stuff Scubadog - don't forget high waking can't be mass locked.  If I'm entering a bad neighborhood, or know that lots of people are probably there looking to kill me, I'll pre-select a high wake target the moment I land in system.

Good ole Marra taught me that one ages ago.  Still works like a charm.

Rebecca Hail I agree that you can't develop an AI capable of anything near human performance with the tech (and budget) that's available for games.

James Hussar

Rebecca Hail I agree that you can't develop an AI capable of anything near human performance with the tech (and budget) that's available for games.

I think this is great speculative fiction. A lovely fairy tale people like to tell themselves about human superiority at tasks as mundane as flying a primitive simulator. If anything, making NPC AI weak enough to put up a "challenge" that keeps people playing the game is a bigger challenge than making one that can "beat a good pilot". There is maybe 1% of the users of this site that could genuinely beat an off-the-shelf chess program. Thing is, if faced with a decent "AI" in an NPC, everyone would cry foul, claim the NPC is just "overpowered", say it is "cheating" by too accurately leading their target, etc, etc... FD calibrate their "AI" to be beatable by children to the point of billions of credits.

A skilled player will always beat a well-tuned AI because "well-tuned" means the AI is leashed, muzzled and crippled enough to give a mediocre player enough of a thrill to keep coming back. This says exactly nothing about how good the AI could be if it were to be made "good".

James Hussar

Rebecca Hail I agree that you can't develop an AI capable of anything near human performance with the tech (and budget) that's available for games.

I think this is great speculative fiction. A lovely fairy tale people like to tell themselves about human superiority at tasks as mundane as flying a primitive simulator. If anything, making NPC AI weak enough to put up a "challenge" that keeps people playing the game is a bigger challenge than making one that can "beat a good pilot". There is maybe 1% of the users of this site that could genuinely beat an off-the-shelf chess program. Thing is, if faced with a decent "AI" in an NPC, everyone would cry foul, claim the NPC is just "overpowered", say it is "cheating" by too accurately leading their target, etc, etc... FD calibrate their "AI" to be beatable by children to the point of billions of credits.

A skilled player will always beat a well-tuned AI because "well-tuned" means the AI is leashed, muzzled and crippled enough to give a mediocre player enough of a thrill to keep coming back. This says exactly nothing about how good the AI could be if it were to be made "good".

Leading edge is the old style, HUD mini target follows your target, line your main reticule up to hit. The mini target can be a little faint on the HUD - particularly if you turn illumination down and it's not always easy to see precisely when you're lined up.

Trailing edge is the approach used in modern fighter aircraft. You maneuver to place the reticule over the target and fire when it's located on it. It's a cleaner, simpler system as you're only watching one reticule.

Rebecca Hail

James Hussar

Rebecca Hail I agree that you can't develop an AI capable of anything near human performance with the tech (and budget) that's available for games.

I think this is great speculative fiction. A lovely fairy tale people like to tell themselves about human superiority at tasks as mundane as flying a primitive simulator. If anything, making NPC AI weak enough to put up a "challenge" that keeps people playing the game is a bigger challenge than making one that can "beat a good pilot". There is maybe 1% of the users of this site that could genuinely beat an off-the-shelf chess program. Thing is, if faced with a decent "AI" in an NPC, everyone would cry foul, claim the NPC is just "overpowered", say it is "cheating" by too accurately leading their target, etc, etc... FD calibrate their "AI" to be beatable by children to the point of billions of credits.

A skilled player will always beat a well-tuned AI because "well-tuned" means the AI is leashed, muzzled and crippled enough to give a mediocre player enough of a thrill to keep coming back. This says exactly nothing about how good the AI could be if it were to be made "good".

In terms of pattern recognition, strategizing and improvisation let alone thinking outside of the box humans are superior to anything you can run on a standard machine. Anti-computer tactics not included.

AI in a computer game is supposed to simulate a human opponent, not to make it cheat the player out of existence by ignoring human limits.

The AI in Elite is completely different than the chess AI though. Their core purpose is the same, their technical implementation not even comparable. Another thing to add for the chess AIs. They brute force either the entire game or calculate twenty moves ahead, evaluate that and then pick the moves based on the evaluation. It reproduces the same steps a human would take, if one had enough time (and brain) to calculate this much in advance, much faster, thanks to heuristic methods. That's a difference in capacity, not in ability. And that only works because chess has a random chance of zero.

You seem to have a fairly simplistic view on AI development and the hurdles it runs into, namely calculating capacity and complexity.

James Hussar
I only studied AI as a side interest in my comp sci master's program, and I only worked on the periphery of several expert systems, neural network and other machine learning schemes for a bit less than a decade, so maybe you're right - maybe I just don't get the intricacies involved in the glorified Pac-Man that is ED's BGS and flight sim. I never studied game design, I worked on SCADA systems, broadband and mobile networks, mild financials and most recently security. Most of these areas don't require AI in the way gaming does, but they use several different variants of AI for decision reinforcement, future conditions prediction and optimization to great benefit.

Here's how I see it. We're playing in a small and well-circumscribed sandbox. The set of actions available to us is very limited - more limited from moment to moment than the search space of a chess board. The ED AI isn't solving for the entire game, it's solving for only the next few steps in a tactical encounter, and it is intentionally dumb and restricted. It CAN absolutely fly better than it does. It can track us better than it does with fixed weapons. Same with the BGS - the AI doing that has the brain power of a newt intentionally. The expert systems supporting major retailer sin figuring out the religious affiliation of the population surrounding their individual stores and deciding when to put pork chops on sale have more robust AI than our BGS does. The FD BGS AI can outwit, outlast and outplay us handily, except it is not allowed to because we'd all ragequit and FD would have to rent their hardware to bitcoin miners to keep the lights on.

If the AI was a challenge, people would not crave PvP. Honestly, I'd be curious how different the gaming experience in ED would be if we didn't know humans from NPCs at least some fraction of the time. It'd be an interesting spin on the Turing test, if nothing else.

Again, I'm not arguing the capabilities of AI at large. Humans are inferior to algorithms in any area we understand well enough to accurately express in code. Humans are superior to AI in areas we can't quite model well, yet, for now. ED is a completely fabricated universe; it's not an approximation of reality, it's a false and fully controlled reality, and as such it is in the first category. The only reason the game is playable is because the AI is necessarily dumber and more limited than the average player. This is not human superiority, it's FD programmers pulling the AI's punches are good design demands in order to entertain the humans who keep the servers, and staff, running with their money.