Information in the dark

Information in the light

There are two sorts of information:

  1. Information we have and
  2. Information we don’t have.

Within the traditional perspective on information, this might be one of the essential properties of information:

To be had

or not to be had.

There is however an alternative view, and it leads to a more operable distinction.

There are two sorts of information:

  1. Information in the light and
  2. Information in the dark.

Like the previous distinction, this alternative one doesn’t involve intrinsic properties of information. Both distinctions concern where the information stands with respect to us.


in the light

or in the dark

  1. Information in the light lies in the light of knowledge we already have 
  2. Information in the dark is overshadowed by information we already have.

Information in the light is wedged tight in between things we already know (TWAK) . We access it by racking our brains long and hard on TWAK and the hypotheses formulated in terms of TWAK … until hopefully … puff: out comes a new piece of information or knowledge.

Taking a theoretical computer science approach to address fundamental biological research questions, I found out there is another way of producing new information. And I really mean “new information”, not just new “point of view” or “new intuition”: new information as in a brand new math theorem with applications in genetics for instance.

The other way of uncovering information is to look for it in “the dark” i.e., outside of the scope of the light shone by existing knowledge. But not too far out. Only right where the light ends.

Information in the dark is information that isn’t to be found in the light of the answers we already have to the questions we’ve already raised. There are other questions than those that have already been raised (by science, by humankind…): questions we haven’t yet thought to raise because until now, it didn’t cross our minds their answers could be anything else than obvious. And further than that: because the very fact that they might not be obvious shakes the foundations of some totally functional knowledge.


N.B. :

Dark information is something different from François Jacob’s “nighttime science”.

Nighttime science is an informal notion referring to the process of making science, and dealing with scientific ideas that aren’t ready for official publication (daytime science).

Information in the dark is information. Or at least, it is potential information until it is taken out of the dark and formulated as actual information.

Information in the dark is resistant to scientific exploration. You could be the most acute researcher equipped with infinitely precise measuring instruments, information in the dark would still not be accessible to you.

One of the most important lessons I learned practicing research was that information can obscure information. It’s not just misconceptions that obscure information.

Information is a matter of point of view, in science as well. Public discourse will take that as an avowal of relativistic weakness of science. But I argue the opposite. There is no relativism and it is a sign of science’s great strength.

In science, models are defined that formalise a certain point of view on reality. A model captures aspects of a piece of reality that are deemed worthy of interest. Other aspects of reality are disregarded by the model. This is necessary and a great strength of science: the whole of the world isn’t regarded at once, most of it is deliberately ignored. This allows extreme focus on only a manageable part of the world. Conditions are carefully optimised to allow us to manage even when starting from a complex global situation that seems unmanageable.

In addition to this, science has an extraordinary characteristic: points of view in science are formalised into models. Unlike in the rest of the world, we have concrete handles on the point of views we take on the world. The definitions of the models we work with are the record of how we look at reality, which parts of it we concentrate on and which we ignore.

Some truths which are visible from one point of view, are not from another. There is no scientific information that isn’t based on a great narrowing down of the field of vision. All scientific information is guaranteed to continue making sense under the condition that we continue ignoring what is outside of the field of vision that allowed to derive the information. This is noteworthy: science provides guarantees.

Sometimes scientific understanding progresses by changing perspectives on something. We are used to the sensational historical examples of changes of perspective like when we switched from the geocentric model to the heliocentric model. But changes of perspectives can be more modest and frequent. The notion of information in the dark provides a way to substantiate what happens in a change of scientific perspective and learn to provoke changes of perspective on demand.

Information in the dark is information that is inaccessible until we change perspectives and give up on a part of the information that substantiates and is substantiated by the current perspective.

There are gateways to information in the dark that can be systematically exploited. Typical gateways are basic assumptions, especially the tacit parts of basic assumptions.

The precise formulation of a basic assumption can be very informative. It can hint at an associated assumption that is more tacit and taken for granted. For instance: “Simultaneity in the expression of genes is fortuitous. So synchronicity in the mathematical model is unrealistic“. This assumption confuses “simultaneity” from the realm of biology with a mathematical property called “synchronicity” from the realm of a formal system. It tacitly assumes the formal system to have the expressive power to account for a notion of time flow that matches the biological time flow in which “simultaneity” takes its meaning. It seems overwhelmingly natural that something called “synchronicity” in the maths represent something called “simultaneity” in the biology… Faced with any imperfections of the model, we may take solace in the idea that a model is a simplification of reality anyway. Plus, our preoccupation in dealing with the complexity and variety of existing biological knowledge is a legitimate priority.

This can easily make us miss that the obvious too needs dealing with. The obvious is a goldmine for new information.

→This relates to this mysterious relation we seldom discuss and yet to which we owe most sciences: “the information relation“.

Uncovering dark information requires considering carefully how one particular object (the model) informs on another object (the reality): under what assumptions, and using what obvious yet unsupported analogies.

Most models that are used in science today have a history. Throughout their years or decades of existence, they have been evolving prior to our encounter with them. They have been looked at through different scientists’ eyes. Understanding of what they represent and how they work has shifted. When a scientist inherits a model as an object of study, she inherits an accumulation of associated assumptions, some explicit, some tacit (exception made of formal models studied as pure mathematical objects). If it is not checked, there is no certainty that the set of assumptions is consistent and compatible with the model. Not checking comes so easily. And yet inconsistency and incompatibility are difficult to avoid and they are good news:  they represent a chance to identify some assumptions that can be dropped, and look differently at the model and what it models.

Gateway assumptions are easy to confuse with knowledge — things we think we know because we have never seem them from a perspective where we could imagine them to be different. The only way we could see them differently would be through a different perspective. The difficulty usually lies in us being locked in our present perspective by the things we see with it. But in science, our ascetic preoccupation with making information and assumptions explicit are a huge advantage.


There are two ways to uncover information:

  1. One is purely constructive: build on answers we already have / uncover information from the light
  2. The other starts more destructively: challenge what we think we know / change the spotlight