Knowledge arks – CONSTRAINT N°1: COMPLETENESS (6/9)

SUMMARY:

By cataloging all items that constitute intellectual heritage around the world, we can plan for conservation efforts and establish priorities for preservation.

Heritage institutions scattered across the globe already help preserve our knowledge, but global institutions that centralize all knowledge in a single location could facilitate this preservation through complementary means.

Initiatives to preserve ambitious portions of our heritage are on the rise, but international organizations, such as UNESCO, could facilitate the implementation of comprehensive “ark of knowledge” projects.

Due to certain practical limitations, it may be appropriate to select or structure the knowledge stored in a universal library rather than centralize all of our existing knowledge, until solutions emerge.

Selection biases and neglect can pose risks in the creation of a selective knowledge ark. Even if selective, a knowledge ark should not be limited solely to established scientific knowledge but should also encompass philosophy, the arts, culture, and more.

New information storage technologies could facilitate the emergence of knowledge arks, and these arks could in turn help test these technologies.

Preservation focused on information—particularly in digital form—would not require as many resources as preservation focused on physical objects (which requires numerous simultaneous conservation methods), while facilitating the large-scale centralization of knowledge in a single location

Copyright restrictions and the legal framework governing cultural heritage institutions can complicate the establishment of a “comprehensive” knowledge repository, unless certain legal adjustments are made.

Cataloging is the first step in preservation

For any knowledge ark project, the first major challenge is ensuring completeness. If we want to ensure that our knowledge survives for centuries to come, all the information contained in the documents selected for preservation must be complementary, and nothing must be forgotten, either by mistake or deliberate omission. This step already presents its own set of difficulties, since our knowledge is scattered across the globe, in various forms and in a wide variety of locations. Beyond the problem of ensuring that everything can be preserved if our knowledge is not centralized at one specific place, it can be especially difficult to keep track of it, trace all the information, and verify that nothing has disappeared. This difficulty is of course amplified by the considerable diversity as well as the number of items that have been accumulated everywhere (books, articles, sound tracks, films, websites, etc.), sometimes in multiple copies, sometimes even in “fakes.”

Therefore, the first step in preservation is cataloging. The purpose of cataloging is both to index everything that can be indexed, i.e., to identify all items that have something to do with our knowledge, and to sort through them in order to identify the rarest items, i.e., those that are most likely to disappear quickly if nothing is done. Cataloging is therefore an opportunity to classify knowledge by level of vulnerability, thus, identifying what has to be preserved in priority. As a result, all kinds of ancient manuscripts, unique art works, or archaeological remains, for example, can be tagged as “high priority” items with a view to creating rapidly more durable copies, in digital format if necessary. Simply put, the sooner an item of knowledge is identified as rare, the sooner it can be duplicated and preserved for the long term. Cataloging therefore makes it possible to limit blind spots in heritage preservation and ultimately guarantee the existence of sustainable versions of all kinds of items, not based on their supposed or actual “essential” nature, but rather on their uniqueness.

For efficiency, the cataloging of our knowledge must follow the following steps:

*Exploration: we search for all items that could constitute the ark of knowledge and prepare an inventory of the heritage to be preserved. For example, we investigate writers from a given geographical area.

*Identification: we take stock of the evidence proving the existence of a particular item and give it a unique name. In the example, this is the title and identification number of a document written by a given author.

*Location: we identify and track the exact location of the item at all times. In the example, we regularly check that a particular document is still in the library or private collection where it was originally found.

*Indexing: we describe the nature of the item (descriptive cataloging), the nature of its content (subject cataloging), rank it in relation to other items, and classify it. In the example, we determine that the document in question is a book, that it deals with art history, and that it belongs to the category of anthropological essays.

*Authentication: we ensure that the item is genuine and not a forgery. In the example, we certify that the book is an original or an authorized copy.

*Condition: We assess the degree of protection required for the item. In the example, we determine the ideal climatic conditions for preservation based on the type of paper and ink used.

*Metadata creation: multiple data items are associated with each item, including those identified in the previous steps (unique name, location, nature, context, etc.), to facilitate future searches related to the item. In the example, a separate but unique file is created that is inextricably linked to the book to which the file refers, and as much information as possible about the book itself is entered: summary, themes, publication date, author’s name, language, etc.

*Duplicate search: we identify whether the item is a variant or a copy of another, if we have already identified it elsewhere. In the example, we try to determine the number of copies produced, whether sold or not.

*Copy: if only a few copies of the item exist, or even just one, a copy is created, either physical or virtual, which must be stored in another location. In the example, a digital copy of the book is created so that the essence of the book will survive if the original book is destroyed.

*Protection: we ensure that at least one of the locations where a copy of the item is stored is well protected from external threats and censorship. In the example, we ensure that the original copy is kept in a heritage institution (public library or other) and that the digital duplicate is stored on an accessible and secure server.

For this last point, we begin to go beyond simple cataloging and enter into the preservation of the item itself. But the priority given to an item in preservation depends considerably on certain information identified during cataloging. If we identify early enough that an item is too fragile, we can plan preservation methods more quickly and, if necessary, take the initiative to create digital copies—such as 3D scans or digitized versions—even before the cataloging process is complete, to ensure that the item’s legacy is not lost. This is particularly true for items that are not written, visual, audio, or audiovisual documents, such as immovable heritage (buildings, fossils, fragments of historical documents, etc.) or intangible heritage (representative arts, crafts, traditions, etc.).

Consequently, the protection of this category of item is obviously a priority. Massive monuments, such as the Great Wall of China or the Pyramids of Giza, original works of art, such as Da Vinci’s Mona Lisa or the Venus de Milo, and even cities such as Venice or Teotihuacan, and natural spaces such as the Amazon rainforest or the Arctic ice cap, all constitute heritage that is given priority attention because it cannot be replaced except by simulacra. But these simulacra are more likely to reach our descendants than the items they simulate. In descending order of priority come items of high historical and cultural value (holy books, the American Declaration of Independence, Antoine de Saint-Exupéry’s The Little Prince, the Apollo 11 archives, etc.), items representing a very rare type of item (Aztec codices or Inca quipus, for example), or items with a comparative value that is more interesting than other items of the same type in terms of rarity (Fabergé eggs for example), completeness (a complete dinosaur skeleton is more interesting than a single isolated femur from the same species; a complete manuscript is more interesting than scattered fragments), integrity (an item in good condition will be easier to protect than an identical item that is badly damaged), or quality.

It should be noted that the examples of items cited above are more sources of knowledge than knowledge itself, but as we saw earlier, the preservation of science also requires the preservation of everything that can enrich it, including philosophy and the arts, as well as culture in the broadest sense. The cataloging of everything that can be cataloged can also serve other complementary preservation objectives: ensuring that the general public is aware of the existence of a given source of information and has access to it, preparing a bank of quotations and possible sources for future study, helping those seeking specific source through rigorous classification, or facilitating centralization of established knowledge in specific locations.

Depending on the preservation strategy, information may or may not need to be centralized

A secondary objective of a knowledge ark would be to centralize this knowledge. Since the whole point of a large epistemic memory is to store as much knowledge as possible, it is clear that centralization would facilitate this. Historically, many civilizations quickly understood the need to collect as many documents as possible, notably through the first antic libraries, including the famous libraries of Alexandria and Pergamon, which aimed to gather all the knowledge and literature of the Greek world. Or through cabinets of curiosities, private collections, and other proto-archives that may have been assembled by the intellectual elites of yesterday. Nowadays, this centralizing role is mainly held by what are known as heritage institutions, i.e., all kinds of organizations that are most often (but not exclusively) public, whose goal is to preserve as many items as possible. There are several major collection sites, depending on the nature of the preserved items:

*Libraries: these sites primarily collect written documents, books, manuscripts, parchments, and any other conceivable writing medium. Libraries can be public, private, school or academic, and may sometimes include additional archives containing completely different information media (magnetic tapes, discs, photos, etc.).

*Museums: these places tend to collect objects that are not strictly speaking information carriers, but rather objects of scientific, philosophical, historical, and artistic interest, such as: works of art (visual arts, sculptures, ruins, etc.), realia (everyday objects such as textiles, currency, tools, etc.), technologies (vehicles, instruments, etc.) and other artifacts used in scientific research (fossils, stuffed animals, astronomical photographs, prototypes, formaldehyde bottles, chemical samples, seeds, etc.). Museums are most often organized around a particular theme or field of search.

*Archives: these places bring together eclectic media and most often constitute the material memories of specific organizations, such as companies, public institutions, schools, universities, research centers, associations, cultural centers, administrative centers, etc. Although the information stored in archives is usually written, as in libraries, there are also institutions that specialize in preserving specific types of media: sound or music recordings, films and audiovisual items, photographs, digital data, optical media (CDs, DVDs, etc.), magnetic media (magnetic tapes, cassettes, etc.), metals, special papers, etc.

*Data centers: these places specialize in the preservation of digital data, ranging from raw data to more sophisticated information, such as software; computer files, communications or websites, which may themselves constitute multimedia media (videos, articles, music, blog posts, etc.). These facilities are designed to centralize information that can be stored on mass storage media.

It should be noted that the main objective of a heritage institution remains the centralization of the items themselves, since it is these items that must be maintained in good condition, whereas a knowledge ark prioritizes the centralization of information and aims to preserve knowledge, regardless of the physical media, and does not necessarily aim to maintain the physical integrity of the latter, although this is equally desirable.

Of course, much of the heritage cannot be gathered in one place, if only because of the immovable objects we mentioned earlier. More prosaically, it would in any case be very difficult to gather most of the existing artifacts in one place, for obvious reasons of space and logistics. But in any case, if we centralized all of our assets in one place, the risk of a disaster wiping them out would be considerably greater. For safety reasons, it is much more desirable to distribute humanity’s heritage across the globe to ensure that at least part of it survives in the long term. On the other hand, centralization is entirely possible, and even desirable, when it comes to storing items that can be easily duplicated or digitized. We may not be able to make copies of every painting in every museum, but we can take high-quality photographs of them, accompanied by metadata (title, date, location, medium, texture, etc.). In other words, while it may be futile to create a global museum, it is entirely feasible to create a global library. This global “library” may only contain copies or digital copies, or even consist solely of a digital library. Thus, the destruction of such a library would not pose a problem as long as other traditional forms of archives remained intact. Furthermore, a digital library, even a global one, would be easier to duplicate in multiple copies across the planet, which would greatly increase the chances of survival of the stored information, while ensuring that as many different regions around the globe as possible have access to this information. Another function of centralization is also to simplify access to our entire heritage as much as possible, while facilitating research: if an item is missing from library A but present in another library B, while another item is missing from A but present in B, this obviously complicates the parallel study of these two items and their possible interconnection. A heritage that is too scattered is an obstacle to comparative research. The advantage of effective centralization of information is, beyond facilitating its preservation, ensuring every possible source of information produced by humanity is identified and available, mapping our knowledge, and, more generally, facilitating the organization of this information.

Moreover, a network of “global libraries” would perfectly complement existing decentralized networks; that is: they would not replace existing libraries. We could even assign different missions to each of them: traditional libraries would mainly aim to preserve items as items (manuscripts, books, films, parchments, etc.), as vulnerable objects of value, while global libraries would aim more to preserve knowledge itself. In traditional libraries, every effort is made to protect priceless and sometimes irreplaceable physical objects: this is direct preservation. In this case, decentralization is the best strategy for promoting the survival of items over time. In a global library, every effort is made to store as much information as possible, and protecting this information involves converting it into a medium that is optimal in terms of both wear resistance and storage capacity: this is indirect preservation. In this case, centralization is the best strategy for ensuring the survival of information. The traditional library exists to ensure that copies of Gutenberg’s Bible will always exist in the future, while the global library will exist to ensure that the contents of such a Bible will always be available, and that we know that it existed in such and such forms in the past. Once again, these two systems complement each other. Both networks share the same goal of preserving our cultural heritage and can both constitute variants of a large epistemic memory. One of these networks would simply be decentralized to ensure the preservation of material heritage, while the other would be completely centralized to ensure the completeness of the heritage content to be preserved. In addition, there are also “items” that do not exist in material form, namely intangible heritage: know-how, practices, customs, oral histories, languages, etc. These “items” cannot be preserved in the traditional way, and only indirect preservation is possible. These cases are therefore even more suitable for centralization in a global memory than material heritage.

Pre-existing organizations would facilitate the creation of a knowledge ark

To sum up, in order to catalog, classify, and centralize knowledge, there are already many institutions, whether private, public, or other. Most of these organizations are local, meaning they are capable of managing information found in a given city, region or country. One difficulty in centralizing knowledge may also lie in the multiplicity of national organizations, which may want to retain exclusive rights to preserve certain items that are of importance in the countries concerned. However, there are international organizations with the capacity to carry out ambitious indexing and preservation programs on a global scale. One example is UNESCO, which has already carried out several (decentralized) cultural memory projects. Among the most notable are: the creation of a network of geoparks (to preserve important nature reserves and thus the world’s natural heritage), the UNESCO collection of representative works (to translate each country’s most emblematic works into multiple languages), the creation of World Heritage lists, both tangible (particularly for monuments, historic centers, museums, and archaeological sites) and intangible (particularly archives, oral traditions, and specific cultural practices), the creation of a World Digital Library, and the “Memory of the World” project, which is a kind of global archive.

These last two projects in particular focus on the preservation of a few representative works, but would undoubtedly constitute a very good foundation for a knowledge ark project, provided that these projects are extended to include the preservation of a much wider variety and greater number of items. An “expanded memory of the world” could thus help us to preserve knowledge and intellectual heritage, rather than limiting ourselves exclusively to the preservation of a restricted selection of works based on national criteria. That said, in the context of an international project supported by several different countries, it would probably be much easier to obtain the necessary funds than with separate initiatives. It would also be easier to establish a global treaty to reform the scientific publishing circuit or even to ensure that different copyright jurisdictions authorize the copying of a work if it is dedicated exclusively to the long-term preservation of knowledge. Such a project would be much more difficult to implement if it came from a specific private or national organization than if it were initiated by an UN-related agency such as UNESCO. Nevertheless, UNESCO is undoubtedly the organization that currently has the most power to initiate a truly ambitious knowledge ark project, while more modest private and non-governmental initiatives are multiplying around the world (Arch Mission Foundation, Long Now Foundation, Memory of Mankind, Internet Archive, etc.) and that the first steps toward “universal library” projects are beginning to take shape: these include, to name just a few, the Wikimedia Foundation; the European Library; Project Gutenberg; the “Global Memory Net”; the “Digital Public Library of America”; and the “Chinese Text Project.”. This means that the world is probably ready for a fully-fledged knowledge ark.

A knowledge ark must strike the right balance in terms of completeness

Another important aspect of epistemic memory is the amount of information that needs to be stored. Until now, we have more or less assumed that in order to ensure the preservation of as much of our knowledge as possible, we need to be as exhaustive as possible: preserving as many items as possible, creating copies, even “virtual” ones, anything with even the slightest scientific or cultural potential, however tenuous. Admittedly, centralizing all our heritage is very difficult if we stuck to preserving physical media (books, films, tapes, artifacts, etc.), but the creation of a kind of global library was less problematic if it consisted of a digital information bank. However, even in this case, there are obvious practical limitations to the creation of an exhaustive collection of information. But above all, even if we manage to overcome all these obstacles and create a magnificent universal library capable of standing the test of time, how can we ensure that our descendants will be able to navigate the vast ocean of information thus accumulated? In other words, shouldn’t we select the information we want to preserve? There are three possible answers to this question, each of which outlines three broad approaches to what a knowledge ark should be:

*The solution of raw, exhaustive storage: epistemic memory revolves around accumulating as much information as possible, without distinguishing between different types of information. The major drawback of such an approach, as has been noted, is the complexity of the research required for anyone seeking to navigate such a mountain of information. The slightest search for information would be like looking for a needle in a haystack, and our descendants might have a very hard time rebuilding a new foundation of knowledge. It would also be more difficult to distinguish between truth and falsehood if every single idea expressed by anyone were preserved the same way. But—and this is precisely where the great advantage of a non-selective epistemic memory lies—by ensuring that every single idea is preserved, we also increase the likelihood that good ideas that have been unfortunately overlooked will endure. Above all, we avoid much of the ideological bias that can arise during a selection process: if all hypotheses and all philosophical positions are preserved—even the most flawed ones—this reduces the risk that our descendants will be manipulated by the information they might encounter. It should also be reminded that we cannot necessarily realize at a given moment everything that may have a significant impact later on. Many works of art or scientific discoveries that are considered important today sometimes went unnoticed in their time, even falling into total oblivion, and it is therefore not extravagant to imagine that current archives possess items gone unnoticed, that will be identified as important in the future. In the specific case of scientific publications, the term “drawers’ effect” refers to the tendency to overlook or set aside scientific articles with inconclusive findings, even though these articles are, paradoxically, crucial for assessing the current state of knowledge on a given subject. The rule of absolute exhaustiveness therefore has the merit of eliminating a great deal of selection bias, and for our descendants, this would result in a fairly representative picture of our civilization and our science.

*The solution of non-exhaustive selection: epistemic memory revolves around giving priority to information considered “relevant” for preservation. The main advantage of a selective information library is that it makes it easier for future generations to analyze this information. The advantage of a limited amount of information, apart from the fact that it requires less energy from our descendants, is that it offers them a concise overview of our knowledge. This solution also has the advantage of being more feasible with current technologies, pending the emergence of new information medium with significantly more advanced storage. Among the most relevant information that could be stored in a selective archive, priority should be given to preserving the most comprehensive sources (encyclopedias, dictionaries, atlases, etc.), the most significant sources in historical and cultural terms (memories of important events, artistic masterpieces, the most popular works, historical legal treaties and documents, etc.), the most scientifically rigorous sources (serious scientific publications and meta-studies), the most representative sources for each existing media category, and the most representative sources for each era and geographical area on Earth. While these criteria do not eliminate bias, they do allow for a synthesis of our heritage, which can already serve as a fairly rich foundation for future generations. Another possible criterion for selection is to prioritize the preservation of anything that is vulnerable or rare, because, let us remember, certain items (fossils, ancient manuscripts, archaeological sites, endangered species, languages, customs…) will be irretrievably lost once destroyed, whereas theoretically our more general knowledge (mathematics, physics, astronomy…) might not disappear irreversibly if lost. A good ark of knowledge would not necessarily be an ark of our most essential knowledge, but, counterintuitively, an ark of our most specific knowledge—that is, also the most ephemeral.

*The solution of comprehensive yet hierarchical storage: epistemic memory is based on accumulating as much information as possible, but this information is organized hierarchically in relation to one another. This approach represents a compromise between the two previous solutions, since it retains the advantage of “forgetting nothing” while also making what is “more important” more accessible. In this way, our most valuable knowledge would be more accessible to our descendants than our more “unfinished” information. It would still be like crossing a vast mountain range for them, but the range would be crisscrossed by a network of marked and mapped trails. In concrete terms, this could mean the following: we could easily envisage a global archive in which the most important information is stored on medium A (easy to access, easy to decode, and not too detailed, like clay tablets), which would serve as a kind of “introduction” to our civilization. Then, deeper within this archive, we could find medium B, containing a much larger amount of information, on a more difficult to access media (like optical discs), and more of an encyclopedic nature. Then medium C even further in, containing even more information, and so on. Thus, if each of these repositories were to contain audiovisual archives, Repository A would contain a selection of photographs, Repository B would contain a collection of thousands of films and documentaries as well as thousands of historical video documents, and Repository C would be the equivalent of a universal audiovisual archive containing as many documents as possible (including forgotten films, reports, clips, advertisements, TV shows) and their associated metadata. Ultimately, the greatest challenge of such an archive would be to determine the appropriate level of detail for the information and to decide which types of information to prioritize. If there were to be a “Chapter 1” for a knowledge repository, then this chapter could include, for example, an index, a guide to interpreting our materials, “Rosetta Stone”-style glossaries, and so on.

All of these approaches to knowledge arks are schools of thought that complement one another perfectly. The diversity of these projects would provide a better guarantee that at least one of them will survive in the very long term.

There are many relevant criteria for selecting intellectual heritage

In the case of a knowledge ark project that involves selecting or prioritizing information, the list of criteria used for these operations will be a crucial decision. Indeed, the major difficulty with selection, apart from the risk of relegating crucial information to the background, is that it also provides a potentially biased view of our civilization to future generations. This question arose, for example, at NASA when creating the Voyager Golden Record in 1977, a disc containing a selection of images, music, and words in several languages. For example, the decision was made not to show “negative” images (scenes of war, tragic historical events, etc.) of the human race, and therefore to include photographs of planets, cities, scenes of everyday life, vehicles, and works of art, among other things. This approach may raise questions because it does not paint a fully representative picture of our industrial civilization, which accomplishments were matched only by its atrocities. One could even add that preserving knowledge of these atrocities could be useful for future generations, who would benefit from being aware of the risks they face if they rediscover a particular technology (nuclear energy, greenhouse gas-emitting machines, information technology, organic chemistry, etc.) or if they adopt combination of policies similar to those that may have led to tragic events in the past. Thus, among the knowledge that would fall under the category of history, tragic events would probably be among the priority knowledge in the selection process.

When it comes to scientific knowledge, it might be tempting to prioritize the most groundbreaking scientific articles (for example, those on quantum physics, natural selection, the theory of general relativity, Gödel’s incompleteness theorem, etc.). However, as noted, some publications did not have an immediate impact upon their release, which implies that the selection of scientific works should not be overly restrictive, so as not to allow major forgotten articles to slip through the cracks of preservation efforts. Furthermore, there are many fields that are still poorly understood by scientists, or that cannot really be approached through a scientific method (metaphysics, ethics, ideologies, aesthetics, etc.), even though they are major areas of human thought that offer a wealth of avenues for reflection on the meaning of existence, the origin of the world, good and evil, the pursuit of happiness, beauty, and many other more or less universal existential questions. In other words, even if it means preserving scientific knowledge that does not all have the same degree of reliability, it would be worthwhile to also seek to preserve more philosophical or speculative works, whose insights can complement those of the sciences.

In addition, there is also a very specific category of works that could be given special attention: encyclopedias. An encyclopedia is a literary work consisting of a multitude of articles covering as many topics, concepts, or people as possible. Each article does not provide exhaustive knowledge on the subject it covers, but despite its concise nature, it can contain most of the key information about the subject covered (definitions, main characteristics, classification, relationships, etc.) and has the advantage of being much more accessible to the general public than a scientific publication or philosophical essay. Therefore, if an encyclopedia is successful, it can provide an excellent summary of all our knowledge, whether they are ideas, questions, established facts, or doctrines. As there are several encyclopedias, sometimes with different editorial philosophies (encyclopedias by experts such as Encyclopedia Britannica, encyclopedias based on the collective knowledge of the general public such as Wikipedia, encyclopedias specializing in a specific field such as the Stanford Encyclopedia of Philosophy, etc.), the risk of errors within an article can be offset by the existence of multiple alternative articles on the same subject. While encyclopedias do not in themselves constitute an exhaustive epistemic memory, they provide a sufficiently comprehensive overview of our knowledge, especially when accompanied by other reference works such as atlases, maps, dictionaries, data tables, and graphs. In short, where an exhaustive epistemic memory would be an important source of information for idiographic studies, a selective and encyclopedic epistemic memory would be a useful source of information for nomothetic studies.

Finally, when it comes to cultural heritage in the broadest sense, both intangible and tangible, immovable and movable, there are again many possible selection criteria. We have already mentioned a few of them: rarity, state of preservation, replicability, representativeness of a specific category of items, among other things. In the case of works of art, we can also consider masterpieces, i.e., works of art considered to be key pieces in their field (genre, period, culture, artistic movement, author, style, etc.). Of course, we encounter the same problem as with scientific publications, namely that the historical importance of a work can vary over time, and therefore selecting works can cause potentially major works to fall into oblivion. The problem is further exacerbated by the fact that the aesthetic criteria used to define a “masterpiece” are much more arbitrary than those used to define the scientific rigor of an article. We can at least limit the problem by seeking to create a selection in which as many artistic movements, genres, nationalities, eras, etc. as possible are represented. The same applies to heritage that does not strictly speaking fall within the realm of the arts: any documentation of lifestyles, techniques, trades, customs, languages, and many other aspects of culture must not leave out any people, any era, or any social milieu. Incidentally, we should not neglect the preservation of anything that might be considered non-artistic, since, on the one hand, art is only one part of culture and, on the other hand, art is a concept with rather vague boundaries that have fluctuated throughout time: certain activities considered artistic at one time were no longer so at another, or became so again later. The fact remains that, even if selective, an ark of knowledge is all the more relevant when it spans a wide variety of diverse fields. It is also conceivable to bring together a multitude of thematic repositories (artistic, philosophical, scientific, historical, etc.)—or even local ones (regional, ethnic, and/or national)—to form such an ark.

Technical solutions enable us to store as much knowledge as possible

Whether we decide to store a selective amount of information or not, the quantity will likely be enormous. Therefore, we will need to find a suitable technological solution, i.e., an information medium with maximum storage capacity. Before the industrial era, storage methods were very limited: any material that could be written on would do (leather, bark, turtle shells, stone, clay tablets, etc.). The most widely used medium was, of course, paper, due to its practicality (it is light and thin, which limits bulk) and its relative ease of duplication, especially after the invention of printing. In fact, the development of science is closely correlated with the advent of this latest innovation. However, despite its advantages, paper, and by extension books, are not insignificant in terms of volume, and the accumulation of paper has imposed a major constraint: the need to construct buildings dedicated to the storage and preservation of books, whose size and number have had to increase tenfold over time. In the mid-20th century, some estimates at the time indicated that the size of libraries would have to double every 16 years if we wanted to continue to store all the books and other writings produced in the world. Even then, it was clear that miniaturization of information media would be necessary to avoid libraries around the world from being overwhelmed by books. Microfilm and microform photographs were the first avenues explored for concentrating the information contained in a book into a smaller, less cumbersome medium. But the digital revolution provided a temporary solution to the problem with the advent of computer memory, whose storage capacities were far superior to those of previous media. Furthermore, the pace of innovation in computing is able to keep up with the growth of information. While computer processing power has doubled every two years since the mid-20th century (Moore’s Law), storage capacity has doubled at the same rate, due to the miniaturization of transistors and their proliferation within a single computer. Storage capacity therefore follows another law, Kryder’s Law, which is very similar to Moore’s Law, i.e., it increases exponentially over time. As a result, the amount of information produced on Earth in 2023 was 120 zettabytes, or a thousand billion of billion bytes, which is equivalent to a stack of books 16,200 billion km long (for comparison, a light-year is equivalent to a distance of 9,461 billion km). It is clear that all the libraries on Earth would be incapable of storing such a mass of information. However, this volume remains so high that it still requires thousands of data centers (themselves containing thousands of servers) to store all the information available on the planet. Furthermore, these data centers are used for a wide range of tasks (social media, communications, stock exchanges, AI, etc.) and consume vast amounts of energy, which thus represents the new upper limit to the centralization of information. Consequently, even today, the creation of a comprehensive epistemic memory does not seem feasible with existing technical solutions, and any “ark of knowledge” project is doomed to be selective or decentralized—that is, to consist of multiple storage media scattered across the globe.

That said, technological progress continues, and there is still hope that the miniaturization of computer components will become so advanced that it will reach the molecular level. There are two particularly promising ways to achieve dramatic progress in miniaturization: nanotechnology on the one hand, and storage on artificial DNA strands on the other. As with traditional media, information is stored in the form of computer bits, which can take the value “1” or “0,” and these “1”s and “0”s were linked to electronic components that had a given size. But with nanotechnology and DNA, these “1”s and “0”s could now be the size of a molecule or even an atom. To give you an idea, a DNA molecule has a diameter of 2 nm (1 nm = 0.000000001 m) and each basic “piece” of the molecule (nucleotide) is 0.33 nm long. If we take the 16,200 billion km long stack of books and assume that each nucleotide pair can hold 2 bits (0.25 bytes), we would need a molecule 43,000 km long (more than 10% of the distance between the Earth and the Moon). This distance may still seem very large, but each human chromosome has 220 million nucleotide pairs spread over a length of 7 cm, and humans have 24 chromosomes per cell. In theory, we could store all the information produced on Earth in 2024 in strands of DNA contained in a volume smaller than a human body. DNA storage technology is still in its infancy, and there are still a few technical obstacles preventing its widespread use for everyday activities (notably slow write and read speeds), but teams of researchers are now able to store the entire contents of the English Wikipedia (2 gigabytes) in strands of DNA, which is already sufficient to envisage the existence of a new category of data center, with a very small volume, and whose purpose would not be to replace existing data centers in order to keep the Internet running, but rather to create a sort of “black box” of the Internet at a given point in time—a back box that would not be accessed and that would updated regularly. In other words, DNA storage would be a satisfactory solution for digital archiving purposes. By deciding to store only the content of existing encyclopedias (not just Wikipedia), we could already create a gigantic epistemic memory by using a ridiculously small amount of space, without requiring considerable energy and investment upstream. Every library, archive, museum, or other institution could easily create its own black box by using a medium that would fit in the palm of one’s hand. If we were to combine all the DNA archives from every cultural heritage institution, we would end up with a kind of universal library—not particularly large, but exceptionally rich in information—which would be a good starting point for comprehensive “ark of knowledge” project, even a decentralized one. It is worth noting that since 2024, there has already been a “lunar library” aboard the Lunar Odyssey probe, containing about twenty books and 10,000 images encoded in strands of DNA, as well as other information stored on various media—a project spearheaded by the Arch Mission Foundation.

An “Ark of Knowledge” project would have technological benefits

In addition to DNA-based technologies and nanotechnologies, other mass data storage solutions are currently under development. This is the case, for example, with 5D optical discs, whose very existence is based on the promise of exceptional longevity—an important criterion for an archive of knowledge, a topic we will revisit later. Other technologies include holographic storage, as well as the use of advanced materials (PCM, glass, or ceramics) on which micro-engravings are made. Some knowledge ark projects, in fact, utilize some of these emerging technologies, such as the “Rosetta” project by The Long Now Foundation or the high-profile example of the 5D optical disc containing a copy of Asimov’s Foundation series, aboard a car launched into space in 2018. Many of these technologies are still in the early stages of development, and the market for new storage technologies is still quite small at the time of writing. DNA appears to be the most promising technology, given that numerous companies are interested and that the total amount of investment (R&D, startups, sequencing, equipment, etc.) ranges from tens to hundreds of millions of dollars. A drop in the bucket compared to other technologies such as AI, nuclear power, or space exploration, but also a sum far greater than what the few research centers are investing in 5D optical discs.

The fact remains that data storage is a major challenge for today’s society: let us not forget that our industrial civilization now depends on information and knowledge. The physical limitations imposed by data centers (space requirements, energy consumption, maintenance, etc.), combined with the exponential growth of digital data, will create an increasingly significant burden in the years to come. Consequently, it is inevitable that the industry will take an interest in all of this, which may also mean that investors face little risk in supporting projects with high symbolic value that will serve as the standard-bearers for these innovations. A “knowledge ark” would serve as both an opportunity to test these new technologies in a “spectacular” way while paving the way for other, more concrete applications: improvements in data centers and cloud computing, innovations in photonics, optics, and biotechnology, as well as applications in the space sector, defense, and more. In other words, a “knowledge ark” would serve both to highlight these new technologies and to demonstrate their feasibility prior to their potential widespread adoption, all at a relatively affordable cost. Finally, archiving could likely be one of the first use cases for new information storage technologies, for the simple reason that one of the main current limitations of these technologies is their write speed—and especially their read speed—which do not yet match that of traditional digital media. Since certain archives are not necessarily intended for regular consultation (to facilitate preservation, in particular), the “slowness” of these new technologies would not constitute a major drawback for this use. Knowledge archives could very well become the vanguard of 5D, holographic, nanometric, or genetic optical media before these technologies mature, while the fact that applications are already possible could help fund research and development in these fields.

Our knowledge is spread across too many different media

New information storage technologies offer a golden opportunity to build a massive, more or less centralized repository of knowledge. But another advantage of these media lies in the fact that the maintenance required to preserve them is minimal. One of the greatest challenges in preserving cultural heritage today is that not only is the quantity of items to be preserved considerable, but also that these items come in a wide variety of forms, which in turn requires equally varied preservation conditions (temperature, security, light exposure, contamination, etc.). New storage technologies would provide an opportunity to establish a new kind of archive, requiring little or no human intervention and minimal protection against external threats, all while utilizing a “universal” medium tailored for any type of information: text, audio, audiovisual, digital files, 3D models, photographs… A single standardized medium could therefore be used for a knowledge repository, eliminating the drawbacks of preservation using current methods. Today, the wide variety of information media also implies a wide variety of institutions dedicated to their preservation, while the indexing of preserved items constitutes a major challenge in itself. This variety can manifest itself at several levels:

*Fundamental level: heritage items, i.e., the things we seek to preserve. There are two types: duplicable items and non-duplicable items. The latter category includes, on the one hand, physical objects that cannot be reproduced or replaced if destroyed, such as architectural heritage (monuments, villages, cities, major structures, etc.), archaeological heritage (archaeological or paleontological sites and their contents: ruins, fossils, time capsules, traces, etc.) and natural heritage (landscapes, living species, interesting geological formations, and even planets and stars). This is also known as “immovable” heritage. On the other hand, there is heritage that can be described as intangible, which includes all traditions and social phenomena that do not involve a material medium per se (at most, an ephemeral physical signal): the performing arts (theater, dance, rituals, etc.), oral traditions (stories, legends, founding myths, languages, etc.), culinary traditions, music, and many other social phenomena (trades, sports, leisure activities, behaviors, values, beliefs, etc.), or even natural phenomena that have been identified. It is understandable that for this category of non-duplicable items, long-term preservation is difficult, as the items in question are bulky, intangible, or unique. This is where we see the value of a knowledge repository dedicated to preserving information rather than directly preserving heritage: all non-duplicable items become virtually duplicable, while also becoming reproducible in the process, in the form of 3D models, photographs, video footage, computer simulations, audio recordings, and so on.

*Secondary level: heritage objects. Among all tangible and reproducible items, there are two main categories: information media and information sources. The first category includes everything used by humans to “write,” to convert information (images, text, sound, etc.) into something that can last for a certain amount of time: photography for images, books for text, sound recordings for sound, etc. In some cases, these media are part of artistic heritage: paintings, sculptures, musical scores, films, etc. They therefore straddle the line between objects that can be duplicated (reproducible in absolute terms) and those that cannot be duplicated (unique and inestimable). Some objects are not designed to contain information, but can still be preserved because they are interesting artifacts in more ways than one. These objects are known as realia: currency, tools, inventions, herbariums, clocks, vehicles, pottery, furniture, chemical or culinary substances, seeds, clothing, and many other objects. Here, the long-term preservation of objects is not limited by non-reproducibility, but rather by the fragility of these objects, which can vary considerably depending on their nature, as well as by the wide variety of methods that must be implemented to restore these objects. As for copying, this is limited by the risk that the recreated object will not be completely identical to its model.

*Tertiary level: information. Information can be classified according to the type of medium (book, disc, film, etc.), but it can also be classified according to the content of the information itself (scientific, artistic, philosophical, religious, political, mundane, etc.) or even according to the form of the information (image, volume, movement, text, speech, music, etc.). Beyond the information medium itself, which can be preserved over the long term, there is the information itself. The challenge of an approach to preservation based on information rather than the object is therefore to safeguard an intellectual element, independently of the medium. The philosophy behind such an approach is to preserve knowledge itself, and more broadly data, ideas, words, etc., to accept the fleeting nature of most of the tools we use today to keep track of everything we do. Furthermore, there are elements of heritage that are only of interest in the form of information: notably literature (it is not the books themselves but the writings that are appreciated for their artistic or intellectual value) and digital arts (3D animation, website design, video games, etc.). Here, the limitation of copying lies primarily in the quality of the stored information, especially when it is copied numerous times and passed from source to source.

This brings us back to the challenge of cataloging and centralization, for which strategies vary according to these three levels. If what we seek to preserve above all else are items, the smallest elements of our universe or our civilization that have been examined by our species, then we must be able to “virtualize” all items that cannot be reproduced identically in the future, converting them into information. If it is objects in particular that interest us, then preservation must be adapted to each type of object and medium, and our epistemic memory suddenly takes on a very protean appearance, making it impossible to standardize preservation methods. If it is information that interests us, then, on the contrary, our epistemic memory will be able to use media that are as standardized as possible in order to facilitate preservation (using preservation methods that will not become technically obsolete) and take into account the nature of the information itself (both content and form). These forms of preservation are complementary: information-oriented preservation may reach its limits if the information to be preserved is copied a large number of times without “remastering,” hence the value of combining the preservation of information with the preservation of the physical items containing that information. Conversely, item-oriented preservation has the drawback that, in the event of an item’s loss, that loss is irreversible unless the item has been virtually reproduced. Knowledge archives and other “universal” libraries will therefore always need to coexist alongside traditional heritage institutions.

Digitization makes it possible to create a massive repository of knowledge

The fact remains that for today’s heritage institutions, fully preserving whatever can be preserved remains an immense challenge, given the vast diversity of media we have just reviewed: it requires mobilizing different skills, different methods, and different facilities; it requires adapting to the materials used; it requires maintaining the ability to decode certain types of items (turntables for vinyl records, computers for websites), not to mention the ability to decipher all kinds of languages and writing systems. Furthermore, in the very long term, this diversity of media may pose a serious obstacle for our descendants (assuming our media are still available in the distant future). The Knowledge Ark therefore aims to circumvent this problem through digitization, using a standardized format.

Digitization consists of “converting” the content of any physical medium into a “virtual” version of it: for example, we can obtain a digital photograph from a printed photograph, create a 3D model of an object that has been scanned, convert the content of a book into text files or web pages, etc. The practice of digital archiving serves several purposes. On the one hand, it allows us to create a “backup” of each item of heritage that has been digitized, thus preserving the object in a certain way even after its destruction. On the other hand, it also facilitates the preservation of intangible heritage as well as immovable heritage, as mentioned earlier. We cannot protect a cathedral or a ritual dance, but we can at least easily preserve an archive of them. Before the advent of computers, this was only possible in a very fragmented way. We could make engravings or photographs of monuments, we could write descriptive accounts of the traditions observed in a particular culture, we could make models or artistic representations. However, on the one hand, all these methods are very varied (So we’re back to the problem of the tedious task of preserving such a wide variety of media), and on the other hand, they did not necessarily allow for a very faithful transcription of the items they represented. A photograph does not always allow us to know the layout of a building down to the last brick; it may be blurry, have poor resolution, or be in black and white, and a photograph can usually only preserve information in visible light, not in other areas of the electromagnetic spectrum. Similarly, a detailed report may overlook certain details, contain subjective comments, and be less valuable than a recording, for example using a digital camera. When it comes to preservation of more “invisible” things (systems of values, definitions, meanings, beliefs, etc.), the difficulty will always be the same, whether digital or not. But thanks to digital technology, we can at least store much more data, so we can afford to be much more precise and rigorous, without any technical limitations forcing us to be concise. What’s more, by being able to store radically different types of information (visual, textual, audiovisual, audio, etc.) on the same medium and in the same format, we can create a centralized information repository in a single location. Our descendants will still need to decipher digital information, but they won’t need to grapple with a multitude of different technologies to do so.

As a result, with all the advantages of digital storage, mass digitization projects have multiplied in recent years (Google Books, Project Gutenberg, etc.): databases and servers are now dedicated to preserving millions of books, millions of pieces of music, hundreds of thousands of films, and many other iconographic and audiovisual archives. Initiatives such as Google Street View enable us to preserve the memory of many geographical locations on Earth (streets, building exteriors, landscapes, etc.), and thus part of our immovable heritage. We are also beginning to create “virtual museums,” where we can find digitized content from traditional museums, and there are a growing number of initiatives to scan as many artifacts, archaeological finds, and more or less ephemeral objects as possible. There are even digitization projects focused on preserving minority cultures, or sometimes even those in danger of disappearing, projects such as “Gi-gikinomaage-min” for Native Americans or “SAADA” for Asian Americans, with all the challenges that this entails in terms of adequately cataloging everything that can represent a culture (speeches, ideas, folklore, living histories, lifestyles, etc.). In short, digitization has led to the emergence of a “mass memory” that is much more comprehensive than any other more traditional form of memory, facilitating the preservation of the cultural heritage of many civilizations, as well as the preservation of all cumulated knowledge across the world.

The main obstacle to completeness remains a legal one

To address the issue of centralizing all our knowledge, there remains a major problem we have not yet discussed: copyright. While certain encyclopedias like Wikipedia and certain elements of our cultural heritage (particularly the oldest ones) are not affected, most scientific and non-scientific works are subject to copyright. In practical terms, this means that the works in question cannot be copied, shared, or distributed. While this is not necessarily a problem for selective knowledge repositories—those whose philosophy is to preserve a synthesis of our knowledge rather than comprehensiveness—for repositories aiming to be more complete, it becomes an obstacle for various reasons:

*In the case of a “knowledge ark” project that takes the form of a private collection, the patron establishing such a project would, at best, need to seek permission from the rights holders to keep a copy of each work in their collection, and at worst, would have to pay to obtain that copy. Since the goal here is to preserve as many works as possible, this can quickly cause the investment required to create this “ark of knowledge” to skyrocket. It should be noted that once the copies have been obtained, they can be digitized without issue; they simply cannot be shared and distributed to the general public. However, since this is not the objective of a knowledge ark, it does not really constitute a drawback in and of itself. Nevertheless, a collection that is both private and exhaustive seems hardly viable, and a patron who sets out to collect as many copies of works as possible would quickly find themselves forced to scale back their ambition and settle for assembling a selection of works. This brings us back to the issue of choosing a selection criterion, discussed earlier

*In the case of a “knowledge ark” project that would take the form of a private archive, the situation would be much the same as in the previous example: the archive would necessarily have to pay to obtain copies of copyrighted works. It should also be noted that a legal entity (company, association, NGO, etc.) has a different legal status than cultural heritage institutions and individuals, notably the inability to reproduce and digitize copyrighted works. Here again, for an archive project initiated by a private company, achieving comprehensiveness may prove impossible.

*In the case of a knowledge ark project that would qualify as a heritage institution, the rules regarding copyright are very different. In fact, in most countries, the law requires that copyrighted works be deposited with the national library or another equivalent institution, depending on the nature of the work. This means that libraries, museums, public archives, and other such institutions do not necessarily have to pay for all the items they collect. Furthermore, heritage institutions enjoy “privileges” such as the right to reproduce and digitize materials, provided that these actions are justified by preservation. At first glance, a legally recognized library that would transform into a “universal” library collecting all the world’s knowledge would suddenly seem more realistic. However, there are still two major obstacles. The first is that copyright jurisdiction is not entirely the same from one country to another. This means, in particular, that for a heritage institution, it is sometimes theoretically possible to collect virtually all the works from the country in which it is located at little to no cost, but this is not always the case, and, more importantly, it implies that for works from other countries, authorizations and payments are still required. The second obstacle is that heritage institutions are obligated to do everything in their power to preserve the works in their possession, which, paradoxically, can create a number of constraints for the creation of a universal library. In particular, a heritage institution is required to store its collection in an accessible location and must be able to retrieve its contents at any time. And this applies equally to original items and to copies—whether digital or not. But as a result, this accessibility constraint means that an “ark” project initiated by a heritage institution cannot be sent into space, nor can it be buried underground for a very long period, which is, however, entirely feasible for a private initiative. In summary, an ultimate national library is conceivable, but it would not be universal and would not have much leeway regarding very long-term preservation methods.

For all these reasons, most of the knowledge archiving projects that have emerged in recent years (Lunar Library, Memory of Mankind, Knowledge Vault, etc.) do not yet aspire to become “ultimate” libraries, and are content to centralize books, encyclopedia articles, and royalty-free images, possibly supplemented by a handful of paid works, for lack of a better option. For a comprehensive knowledge ark project to be feasible, there are three possible solutions. The first is to slightly modify copyright law to create a special status for “knowledge arks,” which would still impose obligations on them (authorization from rights holders, guarantees regarding preservation, prohibition on distribution…) but also greater privileges than heritage institutions, such as an expanded right to digital copying or the ability to make preserved works accessible only to distant future generations. A knowledge archive would thus have the right to create digital copies of any work, without mandatory payment, and the right to make these digital copies irrecoverable, provided that this applies only to digital copies, and not to physical copies or even directly to the original items. Another possible strategy to circumvent the copyright issue would be for rights holders worldwide to no longer be required to send a copy of their works only to the cultural heritage institutions of the country in question, but to also send them to a kind of international body tasked with cataloging the world’s intellectual and cultural heritage (perhaps overseen by UNESCO, for example). Since it would be physically impossible for such an organization to collect all the works from every country, it would “settle” for collecting digital copies that cannot be distributed to the general public. Such an organization would thus serve as a simple international extension of national heritage institutions, and even if the countries involved would still have the right to “retrieve” their digital copies for whatever reason, the organization’s collection would remain fairly comprehensive at all times. One could even consider provisions preventing countries from retrieving their digital copies in certain very specific situations (such as a coup d’état or civil war). And provided that this global digital collection is located in a neutral territory, relatively far from population centers and conflict zones, it could very well last for a very long time. Finally, a third possible strategy, though less likely, would be for a billionaire to decide to create his own private collection of digitized copies of non-public-domain works and to store his collection in an isolated location, still without sharing the contents of his collection with the rest of the world.