The random number generation (RNG) algorithm and its application are critical factors to assure the fairness of a gambling platform. In this sense, randomness indicates that the generated results are equally unpredictable. In most of the current technology systems, RNGs recieve an input with sufficient entropy from a hard disk-like source and use this entropy to generate random numbers 1. In this method, the computer uses the system’s timestamp as input and encrypts it using a one-way function 2. This method forms the infrastructure of a large part of today’s games and is an inevitable necessity for gambling platforms. One of the possible options to create the underlying algorithm of DunderBet platform is the pseudo-random number generation (PRNG) algorithm which is based on the KECCAK hash function (KECCAK is a family of sponge functions).

Hash functions convert input data with varying number of elements to an output of specified length. An adequate hash function must be able to produce random numbers with uniform distribution 3. Furthermore the hash function used in the algorithm should not be extracted from the output sequence. Any change to the hash function will completely change the output sequence, but the output will be random at all times. The output of a good PRNG algorithm should exhibit a true randomness, regardless of effort to decode it. In order for the algorithm to actually represent the fairness of our platform, randomly produced bits must be unpredictable, even if the introduction seed is known. Since we want to make sure that the fundamentals of our platform have excellent randomness, finalized method will later be tested under the statistical test suite for random and pseudorandom number generators for cryptographic applications 4.

Our primary goal in creating the platform is to be able to build a sponge function that is efficient, secure and most importantly able to reseed itself. The main purpose of the method would be to generate random output bits using seeds transmitted from different and integrated sources 5. The only requirement for the introduction seed should be that the seeds are in a sequence and do not require any pre-processing. In this way, the RNG and reseeding mechanism will be able to work continuously. As a result of all these, the algorithm will be implemented in a simple way without unnecessary iterations. The simplicity mentioned here means that the method will be secure and more convenient to avoid unnecessary operations during the implementation.

In cryptography, forward security (FS), also known as the perfect forward-secrecy (PFS), is a feature of the private key protocols of the agreement that guarantees that your session keys will not be compromised, even if the server’s private key is captured 6. Forward security protects past sessions from future concessions of secret keys or passwords. By creating a unique key for each session, the user will be able to initiate a new session securely even if the previous session key is compromised since the data that is not changed in a specific session protected by that particular key. The attacker may have access to the current state of the function however the backwards processing prevents the access to past seeds or randomly generated bits. Nevertheless, the reseedable structure of the algorithm will provide sufficient entropy to prevent the attacker from accessing past information from the current state of the seeds. The proposed method not only provide the security of the private keys, but also make the predictability of the next state extremely difficult.

A brief diagram of the sponge function is given in figure 8. For given variables of r and c, the sponge function works on a state of b = r+c bits. The value of r represents the bitrate, while the value of c is the capacity of the system. In the first stage, initilazition stage, all inputs are set to zero. After the input message is padded and divided into blocks of size r, the method continues with the absorbing stage and followed by the squeezing stage. In the absorbing stage, the input message is XORed to the first r bits of the state by the function f. When the input blocks are finished processing, the algorithm switches to squeezing stage. In the squeezing stage, the first r bits of the state is converted to the output blocks using the f function. The size of the output blocks can be determined by the user. The last c bits of the state are never directly affected by the input blocks and are never output during the squeezing stage. The variable c represents the feasibility of the security level that can be applied to the algorithm.

First, the use of a sponge function based on permutations will prevent the loss of entropy during iterations. Second, the ability of the system to reseed itself instead of hashing the current state will provide a more efficient method than the conventional ones. Finally, the use of PRNGs provides structurally more simple applicability than the existing methods. Various statistical tests can be used to determine the randomness of the algorithm. Randomness is a probabilistic property; that is, sequences produced by the algorithm can be characterized by probabilistic terms. We will apply multiple statistical tests (frequency, cumulative sums, FFT etc.) to assess the presence or absence of a pattern to determine if the sequence is truly random.

We shared the details of the sponge function and its implementation on DunderBet platform, the most likely candidate to form the algorithm of the platform. The exact details and codes of the PRNG method will be shared after the ICO stage, since we are trying to maximize the applicability levels of different methods as well as security and fairness. In this manner, sponge functions provide efficient, in terms of memory and processig, and provable security features.

1.1- AI Based Problem Gambling Detection

Online casinos collect information about gambling patterns of their customers, such as playing frequency, playing times or betting amounts. By analyzing this information, sites determine whether the behavioral pattern determined for any given individual is normal or non-characteristic and shows risky behaviors that might constitute the onset of a gambling problem. Such technology enables us to get to know our customers and their gambling habits more closely. Thus, we can decide whether the behavior of our customer is controlled and a regular behavior or they show signs of problem gambling. At this point, the most important feature that distinguishes the DunderBet platform from other sites is how we define regular behavior.

For example, the behavior of a customer who plays often does not show that there is a problem with gambling if similar behavioral pattern is observed. However, observing an unusual pattern of behavior or radical changes in the frequencies of play will enable us to make a more precise judgment in determining the beginning of the problem. For these reasons, it is not useful to use a fixed threshold for each customer. Although the betting behavior of two different customers is exactly the same, the meaning of this behavioral pattern for two customers may be completely different. As DunderBet platform, our aim is to determine the behavioral pattern specific to each user to identify the symptoms of problem gambling and take safest preventative actions by using machine learning techniques. In this sense, machine learning will provide a more controlled platform for our users by allowing us to make more accurate analyzes about their betting behaviour.

Once the user has defined as the problem gambler, later steps to be taken by the platform have vital importance. A wrong approach can create an adverse effect and lead the customer to a different platform and lead the problem to continue. People may not always like being told that they have problems and a wrong approach may exacerbate the problem. In order to get the most accurate and safe steps towards approaching our customers, we plan to consult with psychologists and get support for how to approach problem gamblers and communicate. DunderBet platform enables its customers to use online gambling platform more securely and responsibly.