Annex C
DRAFT GAMBLING BILL: REMOTE GAMBLING
TESTING OF SYSTEMS AND SOFTWARE FOR VIRTUAL FIXED
ODDS BETTING AND CASINO GAMING
INTRODUCTION
1. Systems testing is a very important issue
and the ABB thinks that a different approach is needed to that
proposed by the DCMS in their position paper and inferred by the
Bill and supporting documentation; this Annex explains why we
take this view.
2. There is a tendency to regard source code
testing of online gaming software as the appropriate regulatory
approach and, given the existing architecture and structure of
machines in the land based environment, it is understandable why
this should be so. However, as new technology is applied to online
and networked gaming, this traditional approach is proving to
be highly deficient and potentially dangerous.
THE CURRENT
POSITION
3. Hitherto the random number generator (RNG)
and the gaming machine formed a single, integrated hardware unit
that was physically locked to the outside world. As a result,
it was impossible to change the software once it had been reviewed
and installed into the particular unit. By thoroughly studying
the source code of the game, regulators could be reasonably confident
that the game would then perform as specified.
4. New technologies have resulted in substantially
greater complexity being added to the software and hardware used
in online and network gaming. There is little or no control of
the hardware of the end-user game (since this is traditionally
on the user's private computer) and as such the "gaming server"
which runs the game takes on a critical role. While the concept
of a "gaming server" naturally does not exist in the
traditional real-world gaming environment, such system approaches
are the bread-and-butter of the IT industry. Microsoft, Oracle,
Sun Microsystems & Netscape are just some of the many companies
that devote huge sums to the development and refinement of their
proprietary server products and large amounts have been spent
by the online gaming software companies in refining their server
products.
5. The gaming server is far more complex than
a traditional random number generator. It must link into a highly
sophisticated transaction processor and controller, the functions
of which are to route the millions of messages coming into the
system each minute; provide each message with a response; provide
complete records of all messages (in and out) whilst simultaneously
providing summary information; and to do all this in real time.
SHORTCOMINGS
6. The system and database design work on achieving
such performance efficiently, quickly and in the most robust manner
are jealously guarded by software developers. Designs are continually
developed and refined in order to provide a better, faster end-user
experience whilst reducing maintenance costs and increasing security,
thus contributing key competitive elements. If such "trade
secrets" were made available to the testing agencies or regulators,
there is a risk that the designers' intellectual property could
enter the public domain through these channels with attendant
legal liability and vulnerability to damage claims. It is, in
our view, unreasonable to expect any company to put its assets
at risk in this way.
7. Furthermore, the systems are continually reviewed
and refined to keep them current. It is improbable that regulatory
testing agencies, which generally have less time and fewer resources
than commercial software design houses, would be able to remain
abreast of the increasingly sophisticated code and technologies
being produced. Indeed, it is likely that they would need to use
external expert consultants to do so, thus compounding the security
problem outlined above.
8. Despite the fact that they supply businesses
and governments worldwide, major systems designers such as Microsoft,
Oracle, SAP et al do not supply source code to their customers,
because the code forms their core business asset which they are
understandably not prepared to share. Indeed, as these processes
have evolved, it has become clear that systems are only now realistically
assessable on their total output performance—do they deliver
what they purport to deliver? This is because source code analysis
has become operationally impractical and, in terms of answering
that question, unnecessary and unreliable.
9. From an operational point of view and regardless
of the quality of the software, there is an on-going requirement
for maintenance work on gaming servers by the software providers
or their maintenance engineers. For example, Microsoft and Sun
Microsystems upon whose operating systems almost all gaming systems
rely, release security patches on a weekly basis, all of which
need to be added to the system. Hardware failures also occur and
need to be rectified, in many cases irrespective of the time of
day or night. This requires constant access to the server system.
It would be quite impractical if a software provider was required
to get regulator or testing house approval every time a particular
system was accessed. Such an inhibition would place UK companies
at an obvious and immediate competitive disadvantage to those
operators in regulatory jurisdictions with less stringent constraints.
Indeed, such limitations contributed to the closure of Sun International's
operation in the Isle of Man.
10. Software design security and the essential
requirement for constant access to the server system combine to
render traditional source code testing obsolete. Through allowing
the software provider with access to the system (which is the
only practical way of keeping the system competitive and functioning
smoothly), reliance is thus placed on the integrity of the provider;
however, that reliance upon integrity has already been established
through the origination of the software design itself.
11. Although external testing agencies "digitally
sign" the code that was installed to the machine, there is
in fact nothing to stop a provider with access to the system temporarily
replacing that code or adding further code that renders the "tested"
code obsolete or useless. So long as operators or software providers,
those who know the systems best, have access to the system, the
"tested" code can be bypassed. This ability clearly
negates the testing process itself which, in our view, would produce
a situation in which source code testing created a false sense
of security to regulators and players alike.
RECOMMENDED SOLUTION
12. There is, however, a straightforward solution.
Source code testing seeks to evaluate each individual component
of the system but, as we have shown, it is not infallible. It
is much more effective to test the system as a whole, which is
done on a continual basis for further security and confidence.
13. The aims of validation testing are to ensure
that players get a fair game in accordance with the rules, that
government receives the correct tax and that regulators can be
satisfied that operators and players adhere to the law. These
objectives can readily be achieved using the tried and tested
practices of the software and auditing industry; but such an approach
will require an emphasis on systems testing rather than on traditional
gaming testing.
14. By continually checking both the input and
output of the system, it is perfectly feasible to test that the
system is meeting the regulatory requirements whilst providing
the operators and software developers with complete operational
flexibility. By checking each and every transaction for its completeness,
together with spot testing of transactions, it is possible to
verify that all transactions are accurately recorded. By performing
further analysis of the summary data and subjecting the vast amounts
of data to rigorous statistical testing one can further confirm
the integrity of the random number generator in particular and
the system as a whole in general.
15. This form of testing regime will result in
something not merely as strong as that applied to the gaming machines
of today but rather something far superior in that it allows for
rapid development and advances in technology, reduces development
and regulatory costs, provides for easier dispute resolution and
is easily implemented using existing methods from the broader
commercial world.
CONCLUSION
16. We believe that source code testing is neither
a relevant nor an appropriate approach to the regulation of online
gaming. It is potentially insecure as far as the software provider
is concerned and could create significant liabilities for the
regulator; it is operationally difficult to apply which leads
to higher costs and reduced competitiveness; it can be unreliable
and thus generate a false sense of security for both the regulator
and the customer.
17. By far the most effective and preferred way
forward is a mechanism, as detailed above, which provides for
continuous "whole system" checking. This is an extremely
important aspect of the regulatory framework, as implied by the
depth in which we have addressed the issue. We therefore urge
the Committee to give full weight to the views we have expressed
and would be very willing to explore this topic further with the
Committee should we be required to do so.
| 
