2004 Web and Downloadable Games White Paper 

IGDA Online Games SIG 

ideal solution is one that maintains the greatest level of accessibility for legitimate game play while 

identifying, responding to, and learning about the greatest number of attacks. 

The  trade off  becomes  even  more  acute  when  discussing  networked  games.  In  designing  a 

secure  gaming  system  for  skill based  competitions  on  iWin.com,  Flipside's  engineering  team 

deployed  three  distinct  countermeasures:  a  solid  encryption  scheme,  a  protocol  for  verifying 

game clients, and a control response loop. 

To  secure  the  network  pipe  and  prevent  a  man in the middle  attack,  Flipside  integrated  a 

symmetric  encryption  component  for  all  client server  communication.  Utilizing  this  algorithm 

independent  component  achieved  their  goals  both  in  terms  of  accessibility  (the  speed  of 

encryption  and  decryption)  and  security  (symmetric  encryption  is  a  well  tested  method  and  a 

standard of conventional cryptography). This was a seemingly ideal scenario. Unfortunately, the 

work  was  not  complete.  Because  symmetric  encryption  requires  that  the  sender  and  receiver 

share the same key, the crucial element in the relationship is the secure delivery of that key. To 

accomplish  this,  Flipside  designed  a  proprietary  Public  Key  Infrastructure  (PKI)  handshake  to 

deliver  a  (reasonably)  randomly  produced,  single use  symmetric  key.  The  combination  of  both 

technologies provided the best of both worlds: the security of PKI and the efficacy of symmetric 

encryption. 

However, the team quickly realized that the most secure encryption scheme was worthless if not 

combined with a plan to verify and secure the game clients themselves. The ready availability and 

ease  of  use  of  decompilers  led  the  company  to  believe  that  the  most  secure  way  to  mitigate 

attacks on the game clients was to utilize a Zero Knowledge Protocol, or ZKP. By relegating the 

client to display layer code only, ZKPs allow a developer to protect the game and scoring logic 

within a server side component to be delivered dynamically at run time. While using ZKP served 

the  security  goal  well,  the  latency  inevitably  introduced  by  such  high  network  traffic  made  fluid 

game play impossible for all users but those with the fastest connections. In this case, the security 

system (ZKP) mitigated certain types of attacks very well, but at the cost of the operation of the 

games.  The  solution was to go with a hybrid approach: allow inclusion of scoring logic into the 

clients  to  streamline  game  play,  and  replicate  it  with  a  server  side  component  that  could  verify 

scoring  behavior  versus  its  own  version  of  the  logic.  Additionally,  Flipside  employed  a  method 

referred  to  as  the  bouncing  pebble  theory  (unrelated  to  Oxford's  pebble  theory).  This  entailed 

taking  an  outwardly  meaningless  stream  of  game  data,  such  as  an  event  related  to  a  pebble 

bouncing on the side of the road in a racing game, and verifying it along with the scoring events. 

We found that the inclusion of this  red herring  was often overlooked by reverse engineers. 

Finally, the most effective security solution is one that is ongoing, a perpetual work in progress. 

Actively identifying and testing different types of attacks and corresponding new countermeasures 

will  generally  provide  more  security  than  a  sedentary  system.  At  the  same  time,  all  new 

countermeasures need to meet the same criteria: mitigate risks (hacks) against the asset (your 

game) without creating additional holes, and maintaining access for the approved users (players). 

No security system is perfect, but one that recognizes its shortcomings and balances the trade 

offs  outlined  above  will  go  a  long  way  in  achieving  the  ultimate  goal:  the  confidence  of  your 

players.  

1. 

Types of Attacks 

From a technology perspective, it is useful to divide attacks on online games into three major categories: 

hacking, cheating and griefing. These distinctions serve to allow us to separate traditional computer 

security countermeasures from those that are unique to online games.  

Unfortunately, the number of specific attacks is virtually limitless. Also unfortunately, many stem from 

sloppy development and business practices. Buffer overflows and out of range errors are the product of 

poor programming. The traditional software development community has this problem and has not 

addressed it for the past thirty plus years and the game development community is no better or worse. 

Page 72 of 93