Full chapter text
The Cycle of Digital Truth-Testing
Claim → source → original material → context → provenance → independent confirmation → counter-evidence → revision
Popularity ≠ truth • visual resemblance ≠ authenticity • source marker ≠ truth • every truth-claim remains testable
Problem
The question of truth in the digital world is more difficult than it first appears. In a printed page, oral testimony or direct
observation, source and context were often visible to some extent; in digital media, the same material can be copied thousands
of times, its title can be changed, an image can be cropped, its date can disappear, audio and video can be synthesised, and every
trace of the original source can vanish. Thus, ‘I saw this’ is no longer by itself sufficient grounds for ‘this is true’.
A second difficulty is the speed of circulation. On digital platforms a false or misleading claim can reach millions before it is
corrected. The cause is not merely human carelessness; psychological factors such as surprise, anger, fear, identity, group loyalty
and urgency also influence the decision to share. Truth often requires patience; systems of circulation may reward immediate
reaction.
A third difficulty is the transformation of evidence. An old photograph may be shared as a new event; a real video may be
attached to the wrong place or time; a true sentence may become misleading in incomplete context; satire may be read as fact; a
synthetically generated scene may be treated as an authentic record. Falsehood, therefore, is not merely a ‘false sentence’; the
wrong context for true material can also produce epistemic harm.
A fourth difficulty concerns the social structure of evidence. In digital life we cannot personally verify the countless claims we
encounter each day. We acquire knowledge through news organisations, scientific bodies, local witnesses, archives, search tools,
communities, experts, friends and platforms. This dependence is not a weakness; it is a normal condition of modern knowledge.
The question is whom to trust, for what reasons, to what extent, and in the presence of what defeating evidence.
A fifth difficulty arises from the prestige of visual evidence. The popular belief that ‘pictures do not lie’ is especially risky in an
age of digital alteration and synthetic media. Sometimes an image is real but its meaning is wrong; sometimes an image is
artificial but the event is real; sometimes an image is unverified yet emotionally powerful. Truth-testing requires not only the
eye, but also source history and independent confirmation.
A sixth difficulty concerns digital memory. Material may disappear and later reappear; older versions may be altered; claims
may change over time. A screenshot can be useful evidence, but its source, time, complete context and possibility of alteration
must also be examined. For archival truth, version, date, persistent identifier, original copy and change history become
important.
A seventh difficulty is that digital platforms create a new relationship between truth and attention. Material that is more visible
is not therefore more true; material that is less visible is not therefore false. Visibility is not itself an epistemic category.
Popularity, number of shares, number of followers or number of reactions cannot take the place of evidence.
Core Proposition
The central proposition of this chapter is that the digital world does not change the nature of truth, but it profoundly changes the
conditions under which truth-claims are tested. States of affairs, events, documents, bodies, places and historical sequences
resist digital description independently of it; yet our knowledge of them reaches us through chains of sources, platforms, copies,
editing, translation, communities and technical mediation. Truthfulness therefore requires both realism and source criticism.
First principle of digital truth: separate the material from the claim. A photograph may be real while the claim written about it is
false; a sentence may be factually correct yet misleading in that context; a document may be authentic while its interpretation is
biased. The unit of truth-testing is therefore not merely the ‘file’, but file + claim + context + source.
Second principle: the reputation of a source may be an initial indicator, not final proof. A reputable institution can also make
mistakes; an unknown local witness may also provide true information. Reliability is built through a record of performance,
transparency, correction practices, disclosure of sources, expertise and willingness to accept defeating evidence.
Third principle: provenance provides history, not truth. Information about where digital material was created, who altered it,
through which device or system it passed, and what signatures or credentials are attached to it is extremely useful. But ‘source
authenticated’ does not mean ‘claim true’. An authenticated person may still be wrong; an authentic photograph may still be
used to support a false conclusion.
Fourth principle: truth-testing should be multi-evidential. Original sources, independent eyewitnesses, geolocation, temporal
consistency, documentary records, expert analysis, counter-evidence, physical plausibility, previously published material and
version history should be combined as necessary. It is risky to base an entire conclusion on a single indicator.
Fifth principle: correction is not merely a failure of knowledge; it is also a strength of knowledge. A digital institution that
acknowledges mistakes, preserves earlier versions, clearly marks changes and explains the reasons for corrections can increase
its epistemic reliability. Silent edits, unnoticed headline changes or removal of sources can instead increase distrust.
Sixth principle: the defence of truth requires both freedom and accountability. The power to remove false claims can itself be
abused; yet it is also wrong to grant organised deception, forgery, impersonation and harassment epistemic neutrality in the
name of ‘freedom to say anything’. The Parallel perspective requires process, appeal, transparency and explicit evidential
criteria.
Principal Arguments
First argument — a digital copy is not equivalent to the original in every relevant sense. At the bit level a copy of the same file
may be identical, but its social meaning can change with source, time, title, place and chain of sharing. Truth-testing must ask
not only ‘what is this file?’ but also ‘why is this file here?’
Second argument — context is part of a truth-claim. When a photograph of an old flood is shared as a new flood, the image is not
false; the claim is false. Attaching true material to the wrong place, date, person or cause produces a false conclusion.
Third argument — it is useful to distinguish misinformation, deliberate disinformation and harmful sharing of true information,
or mal-information. In the first, falsehood can spread without deceptive intent; in the second, the intention to deceive matters; in
the third, the material may be true but is used privately or out of context to cause harm. Policy and ethical responses should not
be identical in all three cases.
Fourth argument — intention matters ethically, but truth-value does not depend upon intention. A person sharing misleading
material may be innocent, yet the claim remains false. Conversely, material shared with deceptive intent may turn out by chance
to be true. Moral fault and factual truth therefore require different criteria.
Fifth argument — popularity is not evidence. Millions of shares, many comments, endorsement by a famous person or
prominent placement by a platform do not make a claim true. Social approval shows only that a claim circulated; its relation to
reality requires independent examination.
Sixth argument — search results are not a complete map of knowledge. Search tools rank results according to available pages,
indexing, language, freshness, relevance and many technical signals. What appears first is not therefore ‘the truest’. Digital
literacy requires distinguishing ranking from evidence.
गजेन्द्र ठाकु र
Seventh argument — an epistemic bubble and an echo chamber are different. In the first, important opposing sources may be
unintentionally absent; in the second, outside sources are systematically treated as untrustworthy. The first may be addressed by
wider source exposure; the second may require rebuilding structures of trust.
Eighth argument — doubt is not always a virtue. Saying ‘everything is false’ is not truth-testing but epistemic defeat. Excessive
distrust places expertise, journalism, science and public institutions on the same level. Reasonable doubt asks for reasons;
universal doubt abolishes the need for reasons.
Ninth argument — visual resemblance is not proof of authenticity. Synthetic media may look highly realistic, while authentic
footage may appear unreliable because of poor quality. Visual verification may require provenance, original source, time, place,
light and shadow, audio consistency, other evidence and the prior history of the material.
Tenth argument — deepfakes create a truth crisis not only because false media can be manufactured, but also because genuine
media can more easily be dismissed as ‘fake’. Future truth-testing therefore cannot depend only on technologies for detecting
fakes; secure provenance for authentic material is also needed.
Eleventh argument — provenance is an epistemic aid, not an arbiter of truth. Standards such as C2PA can provide a technical
mechanism for verifying the creation and alteration history of content, but they do not themselves determine whether the claim
made by the content is correct. This distinction must remain explicit.
Twelfth argument — a history of corrections is as important as the present form of a claim. If a news page was changed, what
was the earlier headline? If a statistic was updated, why did the earlier number change? If a post was removed, why was it
removed? Version history captures the temporal dimension of truth.
Thirteenth argument — the truthfulness of a quotation is not merely word-for-word identity. The speaker’s full sentence, the
context of the question, time, irony, conditional wording and later clarification can affect meaning. A ten-word excerpt may be
literally accurate yet misleading.
Fourteenth argument — data do not speak for themselves. Digital tables, graphs, percentages and maps depend upon
measurement methods, selection, base rates, time windows and category definitions. Calling something ‘data-driven’ does not
guarantee truth; the origin of the data and the perspective of analysis must be examined.
Fifteenth argument — an anonymous source is not automatically unreliable, but the burden of verification increases. In cases of
persecution, war, corruption or internal disclosure, concealment of identity may be ethically justified. In such situations, the
verifying methods of the mediating institution, multi-source confirmation and checks for conflicts of interest become especially
important.
Sixteenth argument — expert disagreement is not the collapse of knowledge. The digital world makes disagreements among
experts immediately visible. This does not mean that ‘there is no truth’; rather, the quality of evidence, type of question,
uncertainty, method and level of consensus must be distinguished.
Seventeenth argument — the cost of verification is unequal. Falsehood may be easy to create while refutation is difficult. A
fabricated image can be produced in seconds; checking its location, time, source and alteration history may take hours.
Knowledge institutions should respond to this asymmetry by developing precautions, provenance systems and rapid correction
mechanisms.
Eighteenth argument — language translation can increase truth-risk. Automatic translation can mishandle technical terms,
irony, negation, cultural signals or legal language. The original text, alternative translations and subject-matter expertise may be
necessary.
Nineteenth argument — digital records are neither ‘immortal’ nor ‘ephemeral’; both possibilities coexist. Some material survives
through endless copies, while other important material disappears through link rot, platform closure, account deletion or format
change. Historical truth requires active preservation.
Twentieth argument — truth-testing requires social virtues. Intellectual humility, fairness toward sources, patience in reading
counter-evidence, willingness to accept correction, restraint against sensationalism, and the discipline of ‘check first, share later’
are no less important than technical tools.
Pūrvapakṣa
First Pūrvapakṣa: truth has become impossible in the digital age because anything can be fabricated. This conclusion is
excessive. Forgery has increased, but so have tools of verification — original files, timestamps, location data, independent
evidence, archived versions, public data, expert examination and provenance. Difficulty has increased; truth has not
disappeared.
Second Pūrvapakṣa: if every piece of material depends on context, then truth has become relative. Context-dependence and
truth-relativism must be distinguished. The claim ‘it rained’ requires a place and time; this does not mean that rain becomes true
or false at will. Correct context specifies the conditions of the truth-claim.
Third Pūrvapakṣa: if a trustworthy institution says it, no further checking is necessary. Expertise and institutional reputation can
justify initial trust, but no institution is infallible. The type of source, direct evidence, conflicts of interest, correction history and
independent confirmation should be examined as necessary.
Fourth Pūrvapakṣa: every citizen should verify everything personally. This is impossible under the modern division of
knowledge. We depend on physicians, scientists, engineers, journalists, translators, archivists and eyewitnesses. The goal is not
complete self-sufficiency but reliable dependence and accountable chains of evidence.
Fifth Pūrvapakṣa: technical verification tools are the final solution. Image analysis, signal analysis, provenance and synthetic-
media detection are helpful, but every tool has limits. Technical evidence must be read together with context, source and
independent confirmation.
Sixth Pūrvapakṣa: remove false content and the problem is solved. The power of removal can itself become an instrument of
political, commercial or institutional abuse. Removal may be necessary in some cases; but without transparent rules, appeal,
archiving, proportionality and disclosure of reasons, the defence of truth can become censorship.
Seventh Pūrvapakṣa: more openness always increases truth. Open data and archives are useful, but privacy, security, personal
dignity and interpretive capacity also matter. Raw data without context can produce false conclusions. Epistemic openness
requires interpretive responsibility.
Eighth Pūrvapakṣa: digital media are bad, printed media are trustworthy. Forgery, rumour, propaganda and bias existed before
the digital age as well. Digitality changes the scale, speed and form of some risks, but printed form is not itself proof of truth. The
same evidential criteria should apply across media.
Uttarapakṣa
First response — let the criteria of truth be medium-independent while the method of verification remains medium-sensitive.
Whether a claim fits reality is the stable question; but digital evidence requires attention to file history, copies, metadata,
provenance and versioning.
Second response — separate source, material, claim and interpretation into four levels. The source may be genuine, the material
authentic, the claim false, and the interpretation biased. Keeping these four levels distinct prevents leaps such as ‘this is a real
photograph, therefore the story is true’.
Third response — trust should be graded and corrigible. Instead of an immediate true/false binary, categories such as strongly
supported, probable, disputed, unverified and refuted can be useful. The category can change when new evidence appears.
गजेन्द्र ठाकु रक समानान्तर दर्शन — खण्ड २
Fourth response — provenance and verification standards should be public. Where was the material created, who altered it,
when, by what device, and is the signature valid? Such information makes civic verification easier. But the standard itself does
not certify the truth of an idea; this limit should be stated explicitly.
Fifth response — make corrections visible. Where possible, when a false post is removed, or before removal, record the reason,
the original form, the date of correction and the changed claim. Erasing history weakens future research and institutional
accountability.
Sixth response — the language of refutation should be evidential rather than insulting. In an echo chamber, outside sources may
already be considered untrustworthy; merely saying ‘you are wrong’ can therefore backfire. Showing the claim’s source chain,
clear counter-evidence and the method of verification is more productive.
Seventh response — digital literacy is not merely training in tools; it is epistemological education. Students should ask: what is
the claim? what is the source? where is the original? what is the date? what is the context? what independent confirmation
exists? what counter-evidence exists? what correction has been made? These questions remain useful even when the phone or
platform changes.
Eighth response — the defence of truth should be decentralised. No single platform, government, institution or ‘fact-checker’
should receive complete epistemic sovereignty. Mutual checking among journalists, researchers, archivists, local communities,
courts, scientific institutions, independent verifiers and citizens creates a safer arrangement.
Indian Dialogue
Indian philosophical traditions are not predecessors of digital media, but their developed discussions of pramāṇa, testimony,
inference, error and the trustworthy speaker can enter into dialogue with digital truth-testing at the level of problems. The
purpose of this dialogue is not to discover an ‘ancient philosophy of the internet’, but to clarify the criteria of knowledge-claims.
The Nyāya discussion of verbal testimony raises an especially useful problem: why is a speaker trustworthy? How is the
meaning of a statement determined? What happens to earlier belief when defeating evidence appears? In the digital world, the
distance between original speaker, forwarder, platform and quotation increases; the question ‘who said it?’ therefore expands
into ‘what is the source chain of this statement?’
The analysis of inference offers a second parallel point for digital verification. The classical example of inferring fire from smoke
has a modern analogue in inferring a hidden cause from a sign on a screen. But without examining pervasion, alternative
causes, exceptions and defeaters, the inference remains weak. No single visual feature is final proof that something is ‘synthetic’
or ‘real’.
Mīmāṃsā and traditions of linguistic analysis open questions about sentence, context, speaker intention, textual continuity and
determination of meaning. Digital quotation culture offers an important caution here: removing a statement from its context
may preserve its literal wording while changing its meaning.
Buddhist epistemology, through distinctions between perception and conceptual construction, raises questions about confidence
in what is seen. What appears to us is affected by concepts, expectations and naming. But it would be anachronistic to treat this
dialogue as a predecessor of modern image science; its usefulness lies only in clarifying the problem.
The Jain tradition of anekānta reminds us of plurality of viewpoints, but it is wrong to read this as ‘all claims are equally true’.
The parallel answer for digital truth is that perspectives can be many while events are not arbitrary. Several relevant
descriptions of the same event may be possible, but contradictory factual claims require independent evidence.
A simple slogan such as ‘accept only perception’ in the name of Cārvāka is not adequate for the digital age. Most modern
knowledge lies beyond personal direct observation. Dependence on scientific instruments, remote witnesses, documents,
statistics and expertise is unavoidable. Thus, while acknowledging the importance of perception, we need an epistemology of
networks of evidence.
Mithila’s Parallel Perspective
Mithila’s Parallel perspective will test digital truth not on the basis of local pride or the assumption that ‘our own source is
always correct’, but through the discipline of Pūrvapakṣa–Uttarapakṣa and evidential testing. Local news, history, genealogy,
manuscripts, land documents, flood images, literary quotations or linguistic claims all face the same questions: what is the
original source, where did the copy come from, what is the date, what are the textual variants, and what independent
confirmation exists?
A digital manuscript archive is an especially important example. A scan is not the original manuscript; it is a digital
representation of the original object. Colour correction, cropping, page sequence, missing pages, metadata, collection number
and date of copying should all be recorded. Researchers should clearly distinguish between ‘viewed the image’ and ‘examined
the original manuscript’.
Version identification is necessary in digital quotations from Maithili literature. Unless it is clear which printed edition, which
page, which online copy and which edited text are being used, later research may be built on the wrong text. Source integrity is a
cultural responsibility in the preservation of knowledge in local languages.
The risk of reusing old images is especially high in digital material about floods and disasters. To verify images of a river, bridge,
village, embankment or road, date, location, weather, local eyewitnesses, other independent images and official or local records
should be combined. Merely saying ‘this is a photograph from our area’ is not sufficient.
When local oral history enters digital recording, a double responsibility arises: preserve the witness’s voice, but do not treat the
statement automatically as fact. The speaker’s location, time, limits of memory, alternative evidence, consent, privacy and
editing history should be recorded. Oral testimony deserves respect; it is not exempt from criticism.
Eight Criteria for Digital Truth-Testing
Clarify the claim → find the original source → check date/location → read the full context → inspect provenance → seek
independent confirmation → test counter-evidence/defeaters → state the level of certainty of the conclusion
No single image, post, institution or platform is final — combine multiple levels of evidence
Contemporary Applications
In disasters and breaking news, the first rule of digital truth-testing is not to let the pressure of speed substitute for evidence.
During floods, earthquakes, violence, accidents or public crises, old images, wrong locations, unverified death tolls and cropped
videos can spread rapidly. Official sources are useful, but they should be combined with local eyewitnesses, independent
journalists, hospitals, weather or geographical data, and independent chains of visual verification.
In public health, the headline of ‘one study’ does not immediately become a treatment truth. The type of study, sample, control,
peer review, uncertainty, later replication and expert consensus must be examined. On digital platforms, preliminary research
and established medical guidance can appear in the same format; the reader should be shown the difference in levels of
evidence.
In elections and public policy, verification of data, quotations and visual material is important for democratic trust. Whatever
the political side making a claim, the same criteria should apply: original speech, full quotation, official data, basis of calculation,
time frame, counter-evidence and independent checking. Fact-checking itself cannot be assumed free from selective bias; its
methods and sources should therefore be transparent.
In courts and digital evidence, when screenshots, messages, location information, audio or video are presented, chain of custody,
original device, time, alteration, collection method and expert examination can be decisive. Digital files are easily copied;
therefore the presence of a file must be distinguished from a reliable history of that file.
गजेन्द्र ठाकु र
In scientific communication, preprints, press releases, social-media summaries and final research papers have different
epistemic status. A headline or one graph is not a substitute for the full study. Making the original method, limitations, data, peer
review and later revisions available together strengthens scientific integrity.
With synthetic images, audio and text, asking only ‘was this made by AI?’ is insufficient. Artificially generated imagery can be
legitimate artistic or educational material; a real image can be used in misleading propaganda. Four questions matter: what is
the provenance of the material? what is the claim? what signals were given to the audience? what independent evidence exists
about the real event or person?
In group messages and local rumours, a trusted acquaintance often forwards false material unknowingly. Social relationship is
not a guarantee of truth. Along with ‘who sent it?’, ordinary civic verification should ask ‘from whom did they receive it, where is
the original source, what is the date, and is there local confirmation?’
In digital historiography, web pages, social media, e-journals, digital photographs and online archives are useful but mutable
sources. Researchers should record access date, archived copy, version, author or publisher, original identifier and, where
available, a persistent archival reference. Preserving today’s sources is necessary for the history of the future.
In education and research, students should not depend only on search results or summaries. They need practice in reading
primary sources, peer-reviewed secondary sources, bibliographies, accurate quotations, editions and opposing literature. Digital
convenience is not a substitute for reading sources; it is a means of reaching them.
When using AI-generated answers, fluent prose without sources requires special caution. An answer may look confident yet
contain false facts, invented references or conflated sources. Users should open the original source, verify quotations, check
dates and independently confirm important claims. The broader question of accountability for artificial intelligence will be
treated separately in Chapter 89.
Digital provenance standards are a useful new application. If a verifiable history of content creation, editing and signatures is
available, the question ‘where did this file come from?’ can be answered more strongly. But such credentials do not themselves
determine the truth of a news report, the validity of a political claim or the interpretation of an event shown in an image; both
the usefulness and the limits of the standard should be taught.
Chapter Conclusion
The digital world is not the end of truth; it is a new condition of truth-testing. Falsehood is easier to create, copying is rapid,
context is easy to remove, visual synthesis can look convincing, and source chains may become obscure. But the answer to these
challenges is not universal scepticism; it is more disciplined multi-evidential judgement.
Digital truth requires four pillars: accept the resistance of reality; practise source criticism; preserve provenance and version
history; and keep open pathways for independent counter-evidence and correction. Without all four, ‘fact’ can easily turn into
reputation, popularity or technical glamour.
Parallel Philosophy does not treat truth here as the private property of any centre, but neither does it make truth arbitrary.
Claims by the state, platforms, experts, local communities, journalists, researchers and citizens are all equally open to evidential
testing. Marginal voices must be heard; their evidence must also be examined. Mainstream sources are useful; they too remain
open to criticism.
Parallel maxim of Chapter 85: ‘In the digital world, truth is protected not by the shine of content, but by the chain of sources,
context, independent confirmation and the courage to revise.’