Not sure if anybody's noticed but. . . The google cache is off a diffrent web site, its content is similar looks like a more recent version (more content).
It is however, up at the moment, not sure if it will survive a slashdot attack but . . . here it is:
IMHO, the article has got a lot of the points wrong. Although it said at the start that security through obscurity is not the way to go, it promptly suggested ideas that where most defiantly that.
CRC's of the game state and so on are fake-able. Security is defiantly possible; it just takes a bit of thought when the game is written (which is suppose is the point of the article, just the ideas he points out are crap). Its not that difficult to stop cheating, as an example the "Voodoo" cheats that where mentioned above, the server just needs to send the positions of ONLY the other players that current player can see. Another problem with the C&C (Not sure if C&C has this problem its just a good example of the game type) style games is that the WHOLE game status is uploaded each update. The only thing that should be uploaded is what the client can see, it should not rely on the client to do the actual hiding of the enemy players. Same with the "massively multi-player online games" that where mentioned above, the server should maintain the status of the players, inventory changes should not be allowed from the client, only the server should be able to change it, when the player is actually given the item in the game.
The first problem with the "aiming proxies" is VERY difficult to solve, because any sufficiently practised player could do just as well as any computer if you have to deal with the sort of latencies that we have over the Internet. Especially if the server is implemented as I suggested earlier and only give the clients the locations of the enemies that are within the players FOV.
The last item is absolute bollocks, you should never need to hide any of the client side status, the server should track the status of each client. If any of them attempt to do anything that is invalid then its up to the server to stop them not the client, which has been said before can NOT be trusted.
Look at the commercial programs like credit card transactions that go over the Internet, NO trust is put on the client, imagine if it was up to the users computer to store they current balance of their current account. They have managed to make that secure so why the hell cant people do it with games after all they are both just programs. Admittedly its a bit more serious if you go changing you bank details but I'm sure you get my point.
That would be my choice. Put in a decent amount of network storage and allow the users to access their home directories over the SMB/NFS when at uni, and HTTPS when anywhere else.
You will probably want to use a server that supports quotas (*BSD and Linux being a few popular free ones) so individuals can't use up everybody else's space. Then Samba to provide windows access.
So it's basically the same as IPv4. Except as the technology moves on and more people are moving from dial-up connections to fixed connections, there is more chance the user will have the same IP address. So why is this different from IPv4?
Also won't most ISP's be giving you your IP address anyway? This will remove all possibility of the address containing your MAC address, unless they give you the remaining 64bit address space. . . and we can all see that happening can't we:)
Slashdot Mirror
It is however, up at the moment, not sure if it will survive a slashdot attack but . . . here it is:
http://wtarreau.free.fr/1u/
CRC's of the game state and so on are fake-able. Security is defiantly possible; it just takes a bit of thought when the game is written (which is suppose is the point of the article, just the ideas he points out are crap). Its not that difficult to stop cheating, as an example the "Voodoo" cheats that where mentioned above, the server just needs to send the positions of ONLY the other players that current player can see. Another problem with the C&C (Not sure if C&C has this problem its just a good example of the game type) style games is that the WHOLE game status is uploaded each update. The only thing that should be uploaded is what the client can see, it should not rely on the client to do the actual hiding of the enemy players. Same with the "massively multi-player online games" that where mentioned above, the server should maintain the status of the players, inventory changes should not be allowed from the client, only the server should be able to change it, when the player is actually given the item in the game.
The first problem with the "aiming proxies" is VERY difficult to solve, because any sufficiently practised player could do just as well as any computer if you have to deal with the sort of latencies that we have over the Internet. Especially if the server is implemented as I suggested earlier and only give the clients the locations of the enemies that are within the players FOV.
The last item is absolute bollocks, you should never need to hide any of the client side status, the server should track the status of each client. If any of them attempt to do anything that is invalid then its up to the server to stop them not the client, which has been said before can NOT be trusted.
Look at the commercial programs like credit card transactions that go over the Internet, NO trust is put on the client, imagine if it was up to the users computer to store they current balance of their current account. They have managed to make that secure so why the hell cant people do it with games after all they are both just programs. Admittedly its a bit more serious if you go changing you bank details but I'm sure you get my point.
Using ID software as your example, The original Wolfenstein ¥not sure about the spelling :- the enemy AI ¥if you could call it that just made the enemies just walk towards you, Quake 3 its a whole different matter, if you have played it you will know©
Admittedly it 'aint a neural network doing the business, but the end result is pretty similar © © © Grinds to a halt, maybe I'm getting this all wrong© A real neural network will learn how to play the game in the first place, and respond to other shit that the games programmers haven't thought about©
Anyway time to make a point© The big iron H/W is being designed ¥the new UltraSPARC's do this already I think to run up to 1K processors is a single box© I don't think won't be long till we are scaling to 1M processors ¥pulls a number out the air 15-20 years + 5 till somebody needs one that big©
Ever seen what FEA ¥Finite Element Analysis is doing these days, I saw a very impressive demo of a two gears that had been mounted on a couple of rods, one rod was told that is was revolving at a specific speed© The software then went away and looked at the model saw that various bits of the model would collide, and moved everything it could to allow the rod to rotate© The end result was that the second row ended up rotating because it was driven by the gear it was mounted on© There was never any definition of what touched what or where, you just gave it a 3D model and it want away and did the rest, working out what was touch what and so on© This takes some serious CPU power, and although the algorithm it's self is quite simple, it takes in and puts out massive amounts of data©
That's the way I see things going, not increases in algorithm complexity, just the data sets© Same with neural nets, the actual model for them is simple ¥I think, never really looked into them in any great detail, it just takes a bit of horsepower to get the job done on any reasonably sized neural net© Admittedly an ASIC ¥specifically designed processor analog or whatever would probably get the job done more elegantly, but it can't do anything else©
Bit of a rant for me but it's been a boring day :-
You will probably want to use a server that supports quotas (*BSD and Linux being a few popular free ones) so individuals can't use up everybody else's space. Then Samba to provide windows access.
Also won't most ISP's be giving you your IP address anyway? This will remove all possibility of the address containing your MAC address, unless they give you the remaining 64bit address space. . . and we can all see that happening can't we :)
+ the company's Anti-Netscape