Cryptocurrency for project Apogeios. Value of the currency system grows as the quality of certain SW stack grows even better, the ecosystem grows stronger and the inhabitants of Apogeios internalizes more sense of quality.
- 02AM, 03/18/2018 Shanghai time, woke up. Thanks to dearest Budha. All of a sudden, the ideas just swamped in from nowhere.
- 05PM, 05/04/2018 Shanghai time. T - 30 days qcoin on chain.
- 10AM, 06/03/2018 Shanghai time. T + 00 days Initial code merged.
There are two major challenges to lead Apogeios inhabitants through the transformation.
- Awareness of quality and what exactly quality means for a genearl SW stack and a good SW ecosystem. In most cases, the inhabitants skip necessary pondering of quality not because they do not care or they are ignorant. Instead, they do not know exactly what quality means and what possible measurements for quality improvements.
- Encouragement of extra efforts (always in risk of delaying of workload at hand) for better quality. Everyone rushes for getting things done faster for shorter term targets while the flaws usually fire back sooner and later and will incur more work days to fix them.
The transformation does not nececessily have to be any connections to blockchain based cryptocurrency which can be overkill, but the popularity of blockchain and all hypos built around it could give a nudge to the transformation. And beyond that blockchain based cryptocurrency does bring foundamental benefits to resolve the challenges. It offers a quantiative metrics of quality improvement efforts which can be recognized in a specific concensus algorithm. The concensus algorithm plays a most wanted vision - quality recognition is better to be decentralized and self-organized regardless a potentially practical application in Apogeios that we need a hybrid approach in the beginning.
Crytocurrency encourages risk taking of extra efforts for better quality. Not just the number increase in personal balance account is a psychological reward but also the potential possibility of cashing out through standard on the Earth reward system. Benefits or penalities of anomymity nature of cryptocurrency in this application is arguable but should not be a big concern either way. On the other hand, it opens an opporutnity to encourage those who endorses final schedule and work assignments to allow more risk taking because the protocol can be designed to reduce their balance otherwise.
Besides quality improvement efforts would be rewarded as a balance increase, unintentional consequences of compromising quality from undisciplined operation could result in balance deduction. How much the system should apply this would require a thoughtful consideration. Apparently it should not hurt the positive encouragement especially in the early phase when the system rolls out.
The decentralized ledger plays as a place to store all quality efforts. It helps every inhabitants to view what's going on in the society. There are various types of activities aiming at quality improvement. These activities happens everywhere in the stack. The underlying blockchain mandates a broadcast of these activities because if you do not do, the activities would not be appreciated, be rectified and stored in the chain. This is soemthing like ISFS and it's actually the Modal, so it makes it easy to develop a Viewer to zoom in/out the evovlement of Apogeios quality.
Another subtle benefit resides in the essence of any currency. Currency is all about trust. In Apogeios application, the trust is a common sense (consensus) of SW quality and the recognition/appreciation of others on your efforts. More inhabitants understand (trust) quality and want to spend more energy in quality efforts, bigger value the currency system will hold. In other words, this is essentially the direct mapping from total value of the cryptocurrency economy to the sense of quality in Apogeios society.
Last but not least, traceability if you know what it means.
GNOSIS - An Ethereum based platform for prediction market. “Our mission is to build a truly impartial exchange for information aggregation to quantify the future. A permissionless and decentralized platform built on Ethereum, Gnosis is the easiest way to aggregate relevant information from both human and AI agents into one number.”
Prediction market, in essence, aggregates information about the expected outcome of a future event. Instead of depending on a pre-selected polling panel to make decision from potentially biased and subjective interpreation of information, GNOSIS enables EVERYONE to participate in the decision process by contributing what they know and getting incentives from how their information help the final outcome from the decision.
There are a few similarties between GNOSIS and QCOIN,
- Both believes in the vision of crowdsourced wisdom.
- Both are based on blockchain, the underlying infrastructure to provide an univeral, decentralized, permissionless and trustless platform.
However, there are key differences,
- GNOSIS targets a generic platform for prediction market. QCOIN is very much focused on SW quality.
- GNOSIS core objectives are to build a forecasting tool (universal), offer customized information search (like Google) and setup standard for predictive assets. QCOIN, on the other hand, wants to track and quantify all SW quality activities, use cryptoeconomy to verify/reinforce various existing SW quality metrics/perceptions, and eventually help all Apogieos inhabitants internalize sense of quality.
- GNOSIS evaluates decision outcome in a seemingly easier way (?). QCOIN has the major challenge in evaluting SW quality and by how much and then use the differences to calibrate the system (think about SGD). How to keep inhabitants' interests in playing with QCOIN is another big challenge.
Literally all engineering activities and associated decisions have more or less impact on quality. Pointers to these activities and decisions need to be put on the blockchain. This is an obvious connection/bridge between the cryptoeconomy and the Earth. Let's see how these engineering activities can be mapped to the economic behaviors in the cryptoeconmy A few examples of these engineering activities,
-
Design review
-
Code review/inspection
-
Dev ULT review
Out of question the importance of Dev ULT. However it's incorrect to measure the quality of Dev ULT ONLY by code coverage. Intuitively, this is just not right - the quality of art of programming cannot be judged by some stupid cold tools. Even setting a hard level of coverage would be inducing undesired behaviors. Martin Fowler explained in readable language that test coverage is simply for finding untested code rather than as a measurement of quality. The AssertionFree testing is just hilarious about 100% coverage but testing nothing. Another example of 100% coverage but not telling the truth of quality is that the code logic itself is incomplete so a 100% coverage only takes care of the existing code. In the same blog, Martin talks about two conditions of enough testing:
* You rarely get bugs that escape into production AND * You are rarely hesitant to change some code for fear it will cause production bugs.Said that, code coverage is still a good reference. A very low ULT coverage is very concerning. But note the subtle difference between expecting a high level of coverage and requiring one. This is exactly like when PM requests 98% PR and they would just get a number around that.
For Dev ULT which is white box test, developer should know the best (if not, the team should) the amount of testing necessary depends on a number of factors, and the developer is expected to know those factors better than others – it’s his/her code after all. There is no single, simple, answer, and the developer is expected to be smart enough to handle the truth and work with that.
-
Smart validation review
-
Validation plan review
Some of these activities involve automatic tools to measure certain aspect of quality (e.g., code coverage, complexity). This will possibly help a quickly converged concensus, however there are many cases it cannot and therefore the review of a team consisting of experienced members is required.
There are many other activities that reflect the sense of quality. Below are some examples. Simple results can be used to check the quality automatically.
-
Continuous integration as check-ins of smaller granularity
-
Accuracy of documentation
-
......
There are more complicated situations beyond atomic activities described above. One example is that differnt group of inhabitants have different and sometimes contraray perspectives driving these activities in different order and differetn priority. Clearly the impact on quality would be different. How to map this difference is TBD.
There are a few considerations in this design of consensus protocol (in other words, rules to enforce in the crypto economy).
-
PoS or PoW. PoW is certainly out of question. PoS, on the other hand, requires thoughful consideration to ensure economic incentives and penalites are effective.
-
Miners. Who, how, how many incentives and penalties? This is where art and science mixed together.
Clearly miners, in the case of PoS, validators vote on the next block. Natural enough, the validator who has the biggest size of deposit get the biggest weight. Validators are rewarded for their service to the chain. The inhabitant(s) initiating the quality efforts would be joining the validator group by default.
Fun part here is about the bet. Using code review as one example, if the validators bet on the effort (as promosing for better quality) and put it in a block, they will get incentives later when the code change eventually pass tests; otherwise they get penalized losing some of their stakes (to those who bet otherwise as one option). This is almost like a smart contract of post mortem. It does not require to duplicate the existing code review efforts - those are still off chain data, however it fills in the gap in the past that nobody would look back at why a code review fails gating some easy failures.
In the same code review case, it is arguable that penalizing validators for failing regression tests would demotivate validators to look at the quality metrics that could not be measured by tests. Actually this is not a zero sum game. The protocol can be designed to support both. For example, if the initiator's first submission of code/design change fails automatic tool check on quality, corresponding balance could be decreased. If later on the approved design proved to be not good enough, all validators who bet on the efforts would get penalized.
-
Finality is in gernal not a concern of this cyptocurrency design. The individual or group of inhabitant to conduct the efforts can choose to be anonymous or not. All quality efforts take days or weeks or months to complete. Final ownership will be claimed only when the efforts are completed, and that's when finality is supposed to achieve. However, odd is after a period of time the completed quality efforts is proved to be of not that useful (does not worth that much rewards) or in the worst case an incorrect one. It's probably not a good choice to reverse the finality already made before, because that would essentially mean a hard fork from where the finality is written in the chain. There are various options to handle the situation. One can be automatically add a transaction to transfer the balance of those inhabitants to VOID - like a write-off. Another one can be requiring more stake or reducing the weight in the next round of concensus.
-
Attacks are out of question. This is a game afterall.
-
Confidentiality. Like regular cryptocurrency, the address of each inhabitant is a hashed public key created from a private key only the inhabitant knows. A blockchain viewer can only list account balance but does not know which inhabitant actually owns the account. This is probably mostly wanted because it could lead to discouragement when everyone sees everyone else's account - this is an unecessary competition.
-
Interoperability with existing tools/process. In the best case, the cryptocurrency system should just run in the background and be transparent as much as possible. Considering there are already many systems running in daily life of every inhabitant, adding one more would not entirely build a momentum (of course someone could argue that the marginal cost is very low so it does not matter).
It's a hetergeneous environment consisting of various system each inhabitant lives and interacts everyday. Name a few, JIRA, HSD, VS, Git, Gerrit, Code collabrator, Sharepoint, Eclipse... REST is a common API making it easier for various system to communicate. However not all these systems provide REST API or extensibility for smart contract like programs running to trigger consensus protocol. This means it leaves no other options but the inhabitant has to trigger the process explicitly.
Moreover, to make it more complex, one quality improvement typically involves multiple actions and interacts with multiple systems. It's arguable each action/interaction could stand alone. Flexiblity must be allowed in the design to pack multiple actions/interactions into a mega one for concensus.
-
Psychological safety
This is an interesting experiment to pratice some rigid logic/science in management art. Psychological safety could be a major concern for all inhabitants. Inhabitants could easily connect the amount of cryptocurrency in their balance account to their performance on the Earth. On the other hand, if inhabitants take this easy as a game w/o concerns of real on the Earth penalities and rewards, it makes it easy for the quality sense interlization.
There is argument about a direct mapping from on-chain identity to off-chain real identity on Earth is more effective for promoting sense of quality. The thinking claims authorities on Earth can better recognize those who have more balance and discipline those who have less. This essentially discourages the innate motivation/pride of being better. And this is against the confidentiality.
Protecting psychological safety is of promount priority for an engineering team. Regardless how the crytocurrency system is developed, regular management activities conducted on the Earth is happening day in and day out. What's the marginal benefit of creating another evolving crytocurrency system to push further the management activity by providing the most value of quantiative quality behaviors? It would not be necessary. Instead, the system should be designed in a pull/leader mindset to nurture and cultivate the innate self-esteem and self-pride in the inhabitants. A basic assumption is that every inhabitant has his/her self-esteem and self-pride which could just hibernate - and the acccount balance (even published in an anonymous way) can wake them up.
Comparing to hundreds and even thousands of inhabitants, the management team is at a much smaller size - actually the more self organized the inhabitants are, the less managers would be required. So leveraging tranditional management activities to educate the management team to be more quality oriented is possible - and it's not just about the numbers it's about the maturity as well.
Besides, there are two chances to first start with identity confidentiality and then to switch to real identity if it does not work at the first place. If the cryptocurrency system starts with real identity, it's not going to work to switch back to confidentiality because the trust is broken.
-
Inflation. In genearl, we want to see the total value of the crptocurrency system increase linearly to the population of inhabitants. It does not have to be similar to the knowledge explosion people see on the Earth, because at least internal quality efforts would be getting harder and harder, and contribution from internalization would be approaching to zero. However the growth of ecosystem could be explosive, which offset the other two contributors.
-
Wire money. Inhabitants get incentives from doing extra/better work about quality, and receive pelnaties for doing counter things. Wiring moneny among them, on the other hand, does not carry explicit meaning. Could it be just something like a master coaching an apprentice passing knowledge and now in this case it's the cryptocurrency?
-
Explicit/implicit impact
There are various types of activities/efforts affecting the quality. Some are explicit - impacts are obvious, and these impacts are mostly negative. For example, a code review only cares about coding standard, or a code change breaks complexity check, or a check-in includes multiple indepedent features or bug fixes. These activities may or may not incur incentives or pelnaties depending on how aggresive the system is designed. It is very aggressive if the system automatically write-off some balance as long as the code change fails complexity check no matter the change has been raised for code review - this clearly will requires the inhabitant to check by himself/herself before submitting, and to some extent indicates the inhabitant has more sense of quality. It is very conservative if the system does not even penalize a check-in including multiple fixes as long as it does not incur any regression.
The challenge gets bigger when the activity's impact on quality is implicit. This can be the most likely case in the beginning because that's where the society starts to ramp up on quality. There are only few inhabitants called as quality guru who knows more about quality than most others. They infuence others and increases the probability of a success bet on block. However they may not get involved as validator in all cases so in the case of absence the remaining validators actually put on bet blindly until later the impact emerges. It's also possible that guru might not be able to articulate quality impact as others for some cases. In both situations, a mechanism is required to adjust the amount of incentives/penalities for simliar cases in the future. This is probably an automatic system suggestion about the amount based on historical data.
-
Economy healthy metrics
Parameters of consensus protocol might change if the cryto economy does not run as expected. It requires more research on healthy metrics of an economic system. And even if the metrics are known and there are tools to evaluate the metrics, the correlation to consensus protocol parameters is hard to identify. A BP like approach might be too costly.
The seperation of cryptoeconomy from the things on Earth is actually a good thing. This opens an opporutnity to identify the discrepencies. For example, if the total value of crytoeconomy grows but SW stack quality drops, something is wrong in cryptoeconomy, probably because of incorrect incentivies and/or penalities on various quality activities.
Metrics of the SW stack quality must be formed and automatic agent must be developed to check these metrics on a regular basis. These metrics (indicators) can come from different level. They can be chosen from the high level SW quality indicator set. They can also be micro indicators about execution efficiency, e.g., build time, CI, dev branch regression etc.
-
Keep interests
Feel it's necessary to think through how to keep inhabitants' interests in playing with the crypto currency. A currency nobody cares loses its foundation of existence. In the discussion of phychological safety, I talked about the reluntance of reflecting the amount of qcoin account to any concrete recognition/pelnaty on the Earth - because this is simply not fun. And essentially the cash-out is also based on the interests in the account balance.
A practical incentive system must exist to encourage inhabitants continously invest energy in the currency with or without real benefits from the Earth. According to Limon's hierachy of social drivers ( (1) utility & tools, (2) content & media, (3) community, and (4) vanity), vanity is good. So how to create a sustaible gaming mechanism to faciliate the vanity is the key to success.
A few initial thoughts,
- A more frequent positive feedback loop
- A mechanism that everyone else can thrumb up or down certain activities - and get incentives as return.
- Reguarly posted rank table
-
Apparently this must be a private blockchain as all quality efforts recorded in the chain are confined in the context of Apogeios.
-
It appears that Ethereum is the platform of choice.
-
The implementation must be OS agonostic.
-
Notification. Each inhabitant should receive a daily report of his/her balance. A NPC like agent can be implemented to scan through the full blockchain and accumulate efforts related to certain (anonymous) inhabitant and send over the balance. However because the real identity of the inhabitant is encrypted, the NPC agent can only post all balances in a public billboard.
-
Must note that the intention is not to move the properties onto the blockchain. Instead, each "transcation" in the chain includes a pointer linking to the corresponding physical records in whichever systems.
-
JIRA support is critical since all major activities could have been tracked in JIRA. What about develop a Chrome extension which could monitor JIRA activities and act as a qcoin agent? Note the agent is now running in a sandbox which has no access to local host storage and/or local OS services so the operations on blockchain might be a challenge.