A recurring topic in this blog is my speculation of an alternative form of government that is run by computer algorithms using data instead of humans using public approval. My concept involves a government that defaults to not interfering in people’s lives at all unless the population demands action due to some urgency agreed by a supermajority of the population. The expression of urgency is democratic, but the triggered policies are outside of any human approval or veto, let alone a democratic approval. I conceive of a particular from where the public accepts this condition because there is a democratic process to approve the algorithms used, the priority of objectives to optimize, and the types of data allowed to the algorithms. I further refine this specific concept by classifying data between bright data representing recent observation data from trusted sensors and dark data representing human theories and science. In my concept, this government prioritizes bright data over dark data and this is made possible by the extensive, exhaustive, and intrusive data collection. The proposition is that given sufficient observations about the real world, the algorithm should be able to rederive the science of human theories but also could be expected to discover new truths humans have not yet discovered, or perhaps humans are incapable of understanding due to the complexity and the multitude of factors involved.
To collect unbiased data, the default state of government is no government at all. People are allowed to act according to their wishes as long as it does not raise widespread alarm. As a result, the data collected about people, individually and in groups as a whole, will reflect the current human condition and what to
To collect unbiased data, the default state of government is no government at all. People are allowed to act according to their wishes as long as it does not raise widespread alarm. As a result, the data collected about people, individually and in groups as a whole, will inform the algorithms of what the current population can do and can tolerate. For these discussions, I speculate on a distant converged state after a few generations where the population as fully accustomed to this government as we are with our government. In particular, the population will have memories of the last time the government imposed brief periods of authoritarian rule. This memory will instill caution about triggering government rule because the impact on people’s lives are out of anyone’s control, randomly affect different groups, some who will benefit but perhaps many more will suffer.
The government does not act at all unless triggered by widespread expression of some urgency. This expression is democratic in nature except that it must reach a higher threshold than simple majority. Urgency needs a supermajority consensus, such as at least two thirds of the population. In addition, this government’s policies have very short term expiration dates. The policies expire quickly in order for the population to re-assert an urgency to trigger a fresh policy, but even if this were to happen the second policy is under no obligation to be consistent with the prior policy. In this sense, the subsequent policies are more frightening than the first because the policies will push the populations in different directions at different times.
In context of recent events, an expression of urgency would have been alarm at a new viral pandemic at the time it was recognized around March in 2020. Personally, I do not think even then the level of fear rose to the supermajority level until after our governments started their restriction policies. In my conceived government, there would not be any action without supermajority request, and thus my conceived government would not have caused the urgency that our actual governments caused.
For sake of argument, I will assume that the public would have on their own expressed alarm at the situation at sufficient numbers. This would trigger the algorithm to create a policy in response to the alarm. When this happens and according to my concept, the algorithm will consider all available data about the current situation and consider possible policies in order to select the policy that optimizes the predetermined priorities the population, maximizing beneficial opportunities while minimizing hazards. I specifically allow this government to come up with a policy that has nothing to do with the current crisis at all and there is no obligation on the algorithm to convince the population of the appropriateness of the policies.
As mentioned above, I am imagining a population living under a similar government for generations and they learned to expect this incomprehensibility. In fact, they learned to fear this incomprehensible result. This fear encourages people to avoid triggering a policy unless in the most dire circumstances. The resulting policies will hurt most people and likely will not ameliorate the issue that is causing the alarm in the first place.
Now that population triggered the policy-making process, the algorithms will consider all available data to pick the best policy based on the population’s prior determination of goals and of things to avoid. The data is segregated into bright data of observations and dark data of human theories and science. The algorithm will use dark data to fill in gaps in the observations, but the assumption is that those gaps are rare due to the extensive data collection available. This demotion of science is deliberate because we expect the observations to support the prior knowledge, but also may discover some new knowledge to exploit.
Among the data available to the algorithm is the data point about the expression of urgency in the first place. The urgency itself is bright data. I presume there is a trusted way to collect the public sentiment to confirm the supermajority consensus. On the other hand, the cause of that urgency is dark data. The people are expressing some concern based on their interpretation about what is happening. That interpretation may be wrong, and it certainly is not complete.
As a result, when triggered the algorithm will independently derive what the true emergency is. It may figure out that the emergency is not in any human’s consciousness. The data that the algorithm trusts is the fact that there is an emergency. It figures out for itself what the emergency is.
In this recent experience, I imagine the urgency being triggered by large numbers of people dying in hospitals. The patients show unusual symptoms and the disease progresses in an unusual manner. In addition, there are varying experiences about the disease in different parts of the world, in different cities, and on things like cruise ships. I recall at the time that there was a large variation of how this disease impacted people. There were reports of people dying suddenly on the streets when just prior they were feeling fine enough to be outside. There were other reports where a large portion of people in close proximity getting the disease and needing medical assistance, while other cases of similar close proximity resulted in only a few infections and even fewer needing attention.
To human policy makers, this fact pattern escalated the alarm because this disease had so many ways to harm people. The potential fault in their thinking is that the same pathogen was causing all of these different results. More alarming to me is how consistent all the various governments were in their responses. I recognize there was central coordination, but there seemed to be a lot of contrary evidence undermining their interpretations or the justification of their policies. There were various experts offering compelling counter arguments and alternative approaches but they were unsuccessful in changing the decisions made from the start.
In times of crisis, human decisionmakers rely heavily on previously prepared operational plans that rely on predictions of speculative scenarios. In this case, the operational plans conceived of an extrapolation of the understanding of viruses and pandemics of a generic nature. When confronted of evidence that fit the scenario, the decisionmakers had little choice but to follow the matching plan. Implicitly, this was the acceptance that the scenario’s underlying assumptions are actually true. In this case, the truth is not the result of applying a scientific method, but instead it is the implied result that follows from selecting an operational plan. The proof of this being a viral pandemic comes from the fact that we responded with a plan based on a viral pandemic.
I am convinced there was bad science behind the decisions made in this particular scenario. The science did not occur at all. Instead what happened were a cascade of operational plans triggering each other as a high level declaration of there being a need to respond as if there were a viral pandemic. I agree that this is necessary for human government. Given the speed and dire consequences of a rapidly contagious disease, there is no time for normal human science. There is no time for the usual rigorous controlled studies of scientific methods, and certainly no time for the ensuing peer review and argumentation for a new explanation. Human decision makers have little choice but to react on the possibility that this may be one of the catastrophes we previously predicted might happen. The possibility is the proof.
At the beginning of this episode, a machine algorithm considering the same observations while not taking the human interpretation too seriously may have concluded something differently. In the specific case of my imagined government, the algorithm does not to rely on previously agreed upon operational plans. In contrast to human decision making, the machine algorithm can speedily come to a new conclusion after processing all of the recently observed data. In my imagined government, the algorithm has no obligation to persuade any human of what it discovers in the data. As a result, this kind of government can respond as quickly as human governments following a prepared plan. The difference is that the government by algorithm is responding to actual current events while what we experienced was a response to a speculation that a prior prediction has come true.
From early on it became apparent that there were a lot of variations of experiences across different populations. The viral pandemic explanation should have shown a lot more consistency. It was just one virus genome. In biological terms, a virus is very simple so that that it is both specific to the human species but also general in equal effects on everyone within the species. Every population should have the same criteria for susceptibility of infection and of dire complications. Yet, we observed a wide variation across populations that defied a purely biological explanation.
The data could suggest that humans were contributing the problems, and in particular the variations. There were clearly political contributions in the form of enforcing certain policies for how to deploy and prioritize healthcare resources allocated to patients. Different policies in different areas lead to certain areas exposing vulnerable people to the virus. Variations in health care practices led to differences in how cases were counted and treated. Certain practices made out of ignorance or out of provider’s fears led to deteriorating patient conditions and eventual death. For example, the early preference for full ventilation as the best way to avoid spreading virus-infected aerosols from the patient may have lead to death due to the damage done by the ventilators. Also, the lack of adequate and frequently replaceable personal protective equipment led to extreme caution in terms of assisting patients, leading to more neglect and less intervention than what normally would happen.
Inevitable human politics was also playing a large role in the outcomes. In addition to the obvious opportunities for accusing opponents of incompetence or for promoting their own competence, there were broader alignments of competing ideologies. The objective of trying to best respond to a virus became subverted to a more pressing objective of obedience to the authority of science approved by the government.
Almost immediately after the determination of virus caused pandemic, there was an announcement that something labeled as vaccines are ready for testing. When this happened, the global objective became universal and verified vaccination of the entire human population. This led to the discouragement or even banning of developing therapeutic approaches. At this point, there was something more political occurring. In particular, this effectively was a declaration of war against the anti-vaccination movement that has been previously growing much to the annoyance of governments everywhere. The anti-vaccination movement predominantly objected to the mandatory childhood vaccination schedules. The current situation provided government with the counter attack of mandating a vaccination schedule on adults as well as the children. I imagine the rationale is that vaccinating every adult would convince them that vaccines are not as dangerous because they themselves did not suffer any long term effects.
Going back to the algorithm considering the policy to pursue in response to a trigger of urgency, the algorithm would consider all of the current data. The algorithm would have data about medical practices exacerbating the numbers associated with the pandemic. It would also have data about the growing contentious divide between vaccine promoters and refusers. There is more going on than just a simple virus.
As I continue to point out about government by data, the algorithm is under no obligation to directly address the issue that caused the alarm that triggered the need for a policy. The public’s expression of alarm is bright data, it is observed fact. The reason for the public’s alarm is dark data, the public is responding to scientific theories about what is happening. In my conception of government, the algorithm places more confidence in observations than human science. This government would use human science only to fill in gaps in observational data, and then only when those gaps are needed to better evaluate a promising policy.
In this case, the algorithm has observational information that can counter the dark data of scientific explanation of death by virus. We currently largely ignore the abundant data of adverse consequences of the application of modern western medicine. For decades there has been an acknowledgement that medical practices are the cause of many patients’ ailments and even may be the leading cause of death when properly reported. In the current pandemic there is much evidence that many illnesses and deaths had such iatrogenic or nosocomial causes.
The excess mortality and hospitalizations of the past year equally may be blamed on these medical causes as on a virus. Our human democratic governments have been incapable of addressing the problem of iatrogenesis. There is a human rationalization that doing something is better than doing nothing and that the risks outweigh the rewards. Applied to mortality, this argument has democratic strength because the survivors can exclaim their recoveries from the same practices that killed others who no longer are able to vote.
This bias to accept risks of human actions over risks of nature appeared in the recent experience. The first example was the willingness to sacrifice the future opportunities of the young largely unsusceptible population in order to avoid the embarrassment of losing vulnerable people to a natural virus. The second example is the willingness to accept fatal consequences of a vaccine on otherwise healthy people for the perceived greater benefit of again not being embarrassed by losing life to a natural virus. Implicit in the obedience to science is an assumption that it is more moral to die as a result of the application of science than it is to die from natural causes. It appears that a natural death is worth many deaths by the application of science.
Politically, we ended up in this situation as a consequence of history. First the enlightenment project itself needed science to replace religion. Although religion involves supernatural explanations, religion acknowledged the necessity of the natural world. Science provides rational explanations of nature, but it leads to a project to escape the consequences of living in the natural world.
The iatrogenic problem is a real problem that is causing an annual death toll likely exceeding the annual death toll attributed to this virus. If 500,000 deaths to this virus is a reason for shame, then why isn’t the same true for 780,000 deaths due to consequences of modern medical practices. The problem is more stark when considering many arguments that medical malpractice and negligence may account for the majority of the 500,000 deaths attributed to the virus. Unlike the virus, these medical-caused deaths occur every single year and will continue to occur indefinitely.
A government by data could consider the observations of iatrogenic complications and deaths. The public’s fear of a virus could grant this government permission to impose some new authoritarian policy that would do something, but that something would exploit the opportunity to improve the future prospects based on all observations of the current world. Such a government would be free to decide to tackle the problem of iatrogenesis instead of the problem of the virus. Fixing the overextension of medicine may ultimately benefit more people than overreacting to a virus that is not as threatening as the population perceived.