Scope

The truthchain system aims to be a catalog of every fact or claim about reality, including the justification. By fact/claim we mean:

• things that happened
• things that were said
• things that were observed
• ...anything with a date and a place.

This covers all science and all history including current events. All types of observations are representable: firsthand, secondhand, and beyond.

Artifacts are part of this system (text, media, books, web pages, and any type of content. The uploading of an artifact is an event.

Authorship is a critical piece of this system. All events must be associated with the author who claimed to observe it.

• an occurrence was observed by someone.
• the words were heard by someone.
• the web page was scraped by someone.
• the artifact was uploaded by someone.

Without authorship we cannot trace the provenance of claims, which is the root problem of the current internet and social media.

Truth

Ironically, even though the name of this project is "truthchain", truth is actually never provable. Instead, everything is a claim (hearsay) and we must determine the reliability of each claim, and in turn their conclusions.

As a person, your strongest proof is when you directly observe something. Everything else is hearsay to you (things you have heard). If you want to prove your direct observation to another person, you can't, because your claim is hearsay to them. All hearsay must be vetted (who said it, and are they lying, and are they competent?). After vetting, there always remains a measure of doubt. And since there is doubt, nothing can be ever be absolutely proven. Instead, we deal with measures of reliability.

So, regarding truth, the only real truth is that either you saw something or that you heard someone tell it to you, and if it's the latter then the truth is only that you heard it. Similarly, in a computer system, its direct observation is only that it saw a claim. This is explained in further detail on other pages.

Bad Facts / False Claims

The term "fact" (broadly) means "something that is claimed". This can be a concrete claim or a derived conclusion. Here we'll use the word "claim" but below we also refer to conclusions as facts. Note: this article does not imply that "facts are true" (see the "Truth" section above).

A claim may end up being wrong in only two ways: by malice or by mistake. To evaluate both, we look at the author and the circumstances. The truthchain project aims to make this evaluation visible and objective.

For malice, we measure the probably that the author is intentionally lying.

• Is there a motive to lie?
• Is the person known to be dishonest?

These are the only two reasons a person would intentionally lie. Any claim to either of these is itself a claim that must be substantiated (a reliable claim). If there is no reason to ascribe malice, we move on.

For mistakenness, we evaluate the competence of the author as a witness, and the circumstances of the claimed event. For example:

• How close was the witness?
• How good was the eyesight of the witness?
• How gullible or competent was the witness?
• How precise were the instruments?

Depending on the claim, you can ask all types of investigative questions to uncover deficiencies in the witness's claim.

Evaluating malice and mistakenness is how all knowledge is vetted by anyone with healthy skepticism, even scientists and scholars.

Note: if there is no reasonable suspicion of either malice or mistakenness, it doesn't automatically make the claim true, it just leaves it in the realm of "we don't know". Reliability is earned through corroboration, where multiple reliable people make the same claim.

Truthchain aims to codify the rules of this evaluation, and develop automated and social ways for people to contribute claims and do investigation.

Calculating reliability is a hard problem, but if it can be done, it is done in the way described above, operating on the fundamental data type of a claim containing authorship.

Bad Logic

Claims are nothing without logic, because the vast majority of knowledge are conclusions based on claims. When claims are assembled in a statement with logic, it yields a conclusion. This statement is called an "argument".

A conclusion may then be included in another argument. In fact claims aren't the only elements of an argument; an argument may have claims, assumptions, and other conclusions. If an argument uses a conclusion from another argument, it depends on that other argument. So conclusions may depend on other conclusions. So knowledge forms a graph (specifically a DAG where the leaves are either claims or assumptions).

Logic is the easy part. There are formal rules of logic and well-known fallacies. Bad logic is much easier to identify than bad facts.

The technicals of how to detect bad logic will be an interesting challenge involving NLP or other ML/AI technologies, potentially vetted or voted on socially by humans.

Social aspect & trust

As they say in philosophy, "at some point you need to know (i.e. assume) something". You may have noticed our necessity on reliable authors, for corroborating claims. But how does the truthchain system know who is reliable or not? It can't, or rather it shouldn't. These assumptions must be made either by individual users or by gateways or aggregators.

The whole system — all of the claims and counterclaims and votes — form a network of deterministic logic whereby the reliability scores of claims or facts depend on whom you choose to trust. You may theorize ways of making this user friendly, or promote lists of generally recognized reliable sources.

The fact that trust must be assumed is not as bad as you may think. The truthchain system allows for the tree of knowledge to be explorable. So even though a gateway or aggregator may answer "yes, fact ABC is true", it can also explain its justification by allowing the user to explore the dependent arguments and their elements and the reliability scores for each element, and the justification for each of those (turtles all the way down). So the user can explore the tree to see exactly why fact "ABC" is regarded as true.

Censorship

Decentralization