Published using Google Docs
Alumnus Whitepaper
Updated automatically every 5 minutes

Alumnus Whitepaper

Version 1.0 


Akash Velu, Jyotirmai Singh, Preey Shah, Rohan Virani


In this paper we introduce a decentralised credential network that awards NTTs (non-transferrable tokens) based on completion of community designed tests.


1 Introduction 1

2 The Alumnus Protocol 3 2.1 Principal Agents 3 2.2 Exam Takers 4

2.2.1 The Credential Test 5 2.3 Decentralised Proctoring 5 2.4 Credential Community Governance 6 2.5 Founding New Credentials with ALUM Tokens 7 2.6 Rollout Strategy 7

Appendix 9 List of Variables 9

1 Introduction 

One of the biggest problems in education is lack of frictionless access to prestigious credentials.

In today’s world, learning has been completely unbundled. Anybody simply needs access to an internet connection to learn almost anything, whether it be a blog post, an online course or a YouTube video. However most credentials are still coupled to educational institutions, mainly physical universities and online courses. One necessarily has to learn something from these establishments before they can demonstrate subject knowledge to someone else. This is problematic for two main reasons:

If someone already understands a subject, it is a waste of time to re-learn the subject from another teacher simply to earn a credential.

It is prohibitively expensive to earn the credential since you are paying for educational content that you already understand.

Centralised actors which decouple credentials from learning do exist, such as the Educational Testing Service. However, they have two major flaws. Firstly, they are unaffordable so only a limited pool of people benefit from this service. Secondly, and more fundamentally, centralised services are not dynamic enough to respond to the rapidly changing demands for new credentials. In a time when new fields of knowledge are constantly emerging, the skills that are valued by society are constantly shifting. Centralised services are inefficient at identifying these shifts and offering new credentials for them.

In this paper, we introduce Alumnus, a decentralised protocol on the Solana blockchain that creates and awards NTT (non-transferrable token) credentials for completion of online tests. Tests are designed and maintained organically by a community of members who already hold that credential. Having contributed to a community, members are able to launch their own credentials, encouraging the formation of new decentralised communities. This behaviour is incentivised by ALUM, an SPL token native to the Alumnus protocol.

The Alumnus protocol has three major advantages over all existing credential issuing institutions.

Firstly, new credentials are frictionlessly available. The ability of community members to make their own credentials resolves the inability of traditional testing services to respond to shifts in what skills are considered valuable. Empowering community members in this way creates a more efficient ecosystem where credentials can be offered in accordance with real-time demand and supply, as opposed to the more stagnant model of traditional alternatives.

Secondly, the price of each credential is non-volatile. Once set by its founders, the price of an Alumnus credential is fixed by the protocol which guarantees price stability. This is in contrast to traditional organisations like universities and online courses, which consistently inflate tuition and prices respectively. Therefore, a user is given complete flexibility to take the credential at their own pace, once they feel confident in the material they have learned.

Finally, the unique transparency of the blockchain enables each credential to be prestigious. Users can verifiably see who already owns and designs the credential they are considering. If a user can verify on

chain that a credential in Artificial Intelligence is held and maintained by the most skilled AI researchers in the world, the prestige of that credential is self-evident.

This decentralised system for the frictionless accessibility to credentials will enable anyone to prove what they know regardless of location or socio-economic background.

2 The Alumnus Protocol 

The Alumnus network’s fundamental function is to accessibly verify knowledge by awarding non transferable NTT credentials to those who pass community designed tests. The basic mechanism is as follows:

1. A user chooses to sit a test for a given credential at any point, paying in ALUM to access a test. 2. The user’s answers are graded either automatically or by the alumni community and if the user meets the admission criteria, they are given their result.

3. Once proctors have validated the user’s attempt, an NTT verifying the result is immediately sent to the user’s address as well as a unique one-time Discord link to join this specific credential’s community.

4. Upon joining the alumni community for this credential, the user can now propose/vote on new test questions to defend the question bank of the credential,

5. Anytime an individual interacts with the platform by proposing/voting on questions, they earn varying amounts of ALUM tokens.

6. If an individual accrues a sufficient number of ALUM tokens, they can spend these to start a new credential and its associated question bank

7. The founder of a new credential receives a fixed percentage cut of incoming ALUM payments for their new credential. This incentivises the creation of new decentralised credentials.

2.1 Principal Agents 

The principal actors in the Alumnus network are:

1. Users/exam takers seeking certification

2. Already certified members of existing communities

3. Exam proctors

4. The founders of new NTT credentials

The ALUM token has three primary sources of demand:

Exam takers pay in ALUM to sit a new test.

Proctors stake ALUM to validate exam attempts.

Founders pay in ALUM to start a new credentials

Exam takers and founders pay ALUM to the community treasury such that credential owners can be remunerated for valuable contributions. Exam takers pay ALUM to the founder so they are rewarded for their risk in starting a new credential and creating the initial question bank. Founders and exam takers pay the Alumnus Network a maintenance fee so that server costs can be paid for test administration. Finally, exam takers pay ALUM to a decentralised pool of proctors to verify their attempt. The diagram below represents the flow of ALUM between these stakeholders.

2.2 Exam Takers 

An exam taker first identifies an NTT which certifies the skill they wish to demonstrate competence in. Examples may include a statistics NTT or a deep learning NTT. They must then pay in ALUM for the

right to take a community-determined test. If they pass (with passing standards also determined by community consensus, to be described later) and their exam is approved by the proctors (described below) they are awarded an NTT credential. For example, if a student passes the linear algebra test and proctors verify their exam, they are minted the linear algebra NTT credential. These tokens cannot be transferable since they must be obtainable only by completing the community-assigned credential test to retain their value. If the exam is not successful, no NTT is issued and the payment is retained by the network.

2.2.1 The Credential Test 

For every credential C, there is a question bank Q(C) associated with it, of size N. When a user requests to sit a certification test for C, a random sample of size m of questions is chosen from the bank to generate a new test. The test expires within t hours, and a minimum number of questions r where 1 <= r <= m must be answered correctly to pass.

To deter cheaters, it is necessary that m << N. If this condition holds, then the only way to successfully cheat is to get knowledge of a significant proportion of the large question bank. This is made difficult by the fact that the only way to do so is to ask those who have already taken the test, or repeatedly take the test and pay multiple times. For those who have passed the test, there is a natural incentive to not give information to prevent devaluing their own credential. For those who haven’t passed yet, the hurdle of having to pay for the test many times to be able to see enough of the questions creates a disincentive to cheat.

2.3 Decentralised Proctoring 

To maintain the value of communities, a proctoring mechanism is essential to maintain exam integrity. The Alumnus protocol uses a decentralised pool of proctors. Anyone may become a proctor by staking ALUM tokens with the protocol.

During an exam, a user is asked for permission to activate their webcam. Once the user submits their answers, recording is stopped. From this exam recording, j segments of length τ seconds are spliced and the remainder of the video is deleted. These j segments are then available for viewing by the decentralised pool of proctors, who view the video segments and confirm that no unusual behaviour occurred. Each exam is shown to P proctors with a simple majority vote deciding if the attempt was valid. For their time verifying these short exam clips, proctors are rewarded ALUM tokens. Proctors

who disagree with consensus will be penalised part or all of their stake. Once viewed, these clips are deleted.

The verification of attempts after the fact allows users to test whenever they want and not be constrained by the availability of a proctor, reducing friction.

2.4 Credential Community Governance 

Once a user has obtained their NTT, they are able to participate in community governance for that particular credential. The primary function of this governance is to defend the question bank of the credential. There are three actions that a credential holder can take: propose changes to the question bank, vote on proposed changes, and verify proposed changes.

At any time, a user with NTT credential C and a minimum of k ALUM tokens can propose to add a question to the question bank Q(C). There are then two stages to the verification process:

1. Each holder of credential C can vote on whether the question should be added - contingent on some quorum of votes. Members are awarded ALUM tokens for participating in the vote. If the question passes the vote, the original user who proposed the question is rewarded with ALUM tokens. If the quorum is not reached within some time period, the proposal is discarded.

2. A minimum of p holders of credential C must then independently solve the question to ensure the question is written correctly. If this threshold is reached, all p question solvers and the original question proposer are awarded ALUM tokens. If this threshold is not reached in t/m hours following the vote, the proposal is discarded.

The only ALUM available to the community for remunerating contributions comes directly from the founder’s initial payment and entrance fees for each exam taker. This ensures that a rogue community cannot drain the supply of ALUM in the entire network by making unnecessary updates to a question bank.

The community can vote on changing the number of questions m that are randomly sampled from Q(C) for a test, the test expiry time t and the passing threshold r. Finally, the size of the question bank is lower-bounded to ensure it remains sufficiently large to prevent leaked questions affecting the test’s

validity. For a question bank Q(C), we enforce that N >= s*m where m is the size of the test and s is some scaling factor consistent across all communities.

2.4 Decentralised Grading 

Initially, test answers will be automatically checked versus a rubric to make sure that they are correct. However, this ultimately restricts the nature of the type of exams that can be given because some exams have different testing requirements that are more subjective. To resolve this, communities are empowered to select certain members to act as a decentralised admissions committee for their credential. This committee’s members can grade submissions manually if the community votes for it and award admission accordingly. Graders are remunerated in ALUM from the community’s ALUM pool according to a consensus grading protocol that each community can design for themselves. This gives them full autonomy over the admissions criteria while also providing the option for immediate verification.

2.5 Founding New Credentials with ALUM Tokens 

ALUM tokens are needed to create new credentials in the Alumnus protocol. For example, by participating in the linear algebra community and saving up ALUM tokens, a user could start a differential equations credential by supplying a new question bank. Different people may also pool together their ALUM balances to found new communities together.

The founders must specify the variables m (size of test), r (passing threshold) and t (test expiry time) as described above. Given this, they must create the initial question bank for credential, Q(C*), with a minimum of s*m questions.

Initially, it is impossible to verify that the questions the founder proposes to seed Q(C*) are all correctly written. However, it is possible to see the existing credentials the founder has earned and their history of community contributions. Founders who have created previous communities or have a long track record of community involvement are more likely to create strong initial question banks. This creates trust in the new credential.

Since ALUM tokens can be traded, it is possible to start a new credential even if a user hasn’t been able to earn enough ALUM via contributing to communities.

2.6 Rollout Strategy 

There will be three phases to protocol development, starting from centralised control and progressively decentralizing to community ownership.

In the first phase of protocol development, Alumnus will be under centralised control. There will be no ALUM token while we seed the members of initial communities, and therefore all credentials will be free. Initially, Alumnus will control all functions associated with founders/community members/proctors: creation of new credentials, maintenance of question banks/test sizes, and proctoring.

In the second phase, Alumnus will introduce the ALUM token. Users will pay in ALUM to access exams and committed community members will be awarded ALUM according to the question bank updating protocol. Alumnus will relinquish maintenance of question banks to communities at this stage, but maintain root control over creation of new credentials. The proctoring phase will be deployed to the community in this stage also.

In the final phase, Alumnus will no longer control any functions of the protocol as the community will own the network. Community members will be able to create new credentials with the option of decentralised grading and stakeholders will dictate the future direction of the Alumnus protocol.


List of Variables 

N - size of the question bank

m - number of randomly sampled questions for a test

s - scaling factor that determines minimum size of question bank by law N >= s*m p - number of credential holders who must solve question correctly in verification process t - maximum time allowed to complete the test

r - minimum number of questions to answer correctly to pass the test (1<= r <= m) j - the number of segments each exam video is spliced into

τ - the length in seconds of each spliced video segment

P - the number of proctors that validate an exam attempt