Media archaeology investigates how media devices, formats, interfaces and infrastructures have shaped communication across time, including paths that were abandoned, overlooked or later reinvented. Its gold standard is a historical explanation that connects a particular technical arrangement with evidence of how people encountered and used it, while keeping reconstruction, observation and inference distinguishable.
An old computer, a projected image, a recorded voice or an interactive artwork can each reveal choices that a simple timeline of inventions misses. This guide develops a practical way to investigate those choices: identify the object and its dependencies, examine what its operation makes possible, compare competing explanations, and test what survives when the context changes. All unnamed cases and numerical examples are original teaching scenarios, not accounts of actual collections or experiments.
Start from the Gold Standard collection or its Education guide if your goal is to connect understanding with independent practice. The How X Works hub provides further routes into mechanisms. Here the central task is historical inquiry; Digital Preservation develops the separate responsibility of keeping digital material usable over time.
Choose your media archaeology route
Chapters 1–3: Understand the subject and frame a question
Chapters 4–6: Establish evidence and test constraints
Chapters 7–9: Investigate interfaces and reconstruction
Chapters 10–12: Explain circulation and repair weak claims
Chapter index
Understand the subject and frame a question · Chapters 1–3
1. Begin with a question that an object can help answer
Establish evidence and test constraints · Chapters 4–6
4. Establish provenance before assigning a date or purpose
Investigate interfaces and reconstruction · Chapters 7–9
7. Read interfaces as arrangements of action and attention
Explain circulation and repair weak claims · Chapters 10–12
10. Follow media through work, institutions and circulation
Transfer the method and continue learning · Chapters 13–14
CHAPTER 1 OF 14 · Understand the subject and frame a question
1. Begin with a question that an object can help answer
A device becomes historical evidence when a question gives its features significance. A small screen might interest a designer studying layout, a historian studying access to information, or an artist studying the relationship between gesture and response. Measuring the screen alone will not answer all three questions. Each inquiry needs a different connection between material evidence and interpretation.
Media archaeology has several intellectual and practical traditions. Michal Šimůnek’s overview of the field describes its attention to material and social arrangements, alternative histories and challenges to linear accounts of progress. The working method in this guide is a teaching framework for that inquiry, not an official disciplinary checklist or accreditation scheme.
Turn curiosity into an inspectable claim
Suppose an unfamiliar writing program displays only a few lines at once. “People used to write differently” is too broad to test. A more useful question is whether this particular interface encouraged users to revise locally rather than survey a whole document. That question points towards screen layout, navigation commands, manuals, surviving drafts and accounts from people who used the program.
Now formulate alternatives. Perhaps the display mattered less than a workplace deadline. Perhaps users printed drafts frequently and revised on paper. Perhaps the apparent limitation could be changed through a setting you have not discovered. A good question keeps those possibilities open until evidence separates them.
Write down the proposed mechanism: limited visible text may increase the effort needed to compare distant passages. That mechanism is plausible, but it is not yet a finding about historical writers. A present-day trial could demonstrate the effort required under specified conditions; historical documents would still be needed to show how people responded at the time.
The first useful result is therefore a research boundary. Name the system, the activity, the users or setting, and the period under discussion. Then state the evidence you would need to change your mind. This makes a small inquiry stronger than a sweeping narrative supported mainly by the charm of an old machine.
CHAPTER 2 OF 14 · Understand the subject and frame a question
2. Follow several layers of the medium
What looks like one object often depends on several distinguishable layers. A recorded interview involves a physical carrier, an inscription or digital representation, a playback arrangement, a listening setting and a chain of custody. An interactive work adds rules, input devices and changing states. A catalogue entry describes these things but is another historical object with its own authors and revisions.
For a practical investigation, separate six layers: carrier, representation, operating environment, interface, use and circulation. These are prompts for asking questions, not claims that every medium has identical architecture. A printed pamphlet and a networked game will distribute their dependencies differently.
Distinguish an inscription from its presentation
Imagine receiving a disk labelled “school exhibition”. The label may describe the event, the files, an intended destination or a previous use of the disk. An image file on it might have been created later than the label. The application used to open that image may substitute colours or fonts. A photograph of the resulting screen adds the camera’s settings and framing.
Describe exactly which layer supports each assertion. “The label names an exhibition” is a direct observation. “These files were shown at that exhibition” requires a link between label, files and event. “Visitors saw this exact appearance” requires further evidence about the original presentation. Each additional claim crosses a boundary that needs support.
The Library of Congress explains in its format relationships guide why a format name alone can be insufficient: wrappers, encoded content, versions and related specifications matter. In an investigation, record the specific representation when it is known rather than treating a familiar extension as a complete technical description.
For non-digital material, make equally careful distinctions. A projection slide is not the projected image at a particular size, brightness and distance. A printed page is not every reading of that page. The point is to retain the chain between evidence and experience, so that a claim about one does not silently become a claim about the other.
CHAPTER 3 OF 14 · Understand the subject and frame a question
3. Replace the ladder of progress with comparisons
It is tempting to arrange media into a ladder: the newer device has greater capacity, so it represents an inevitable improvement. Capacity can improve while other qualities change in less straightforward ways. A tool may become easier to distribute but harder to repair locally. An interface may offer more options while concealing the process behind a single command.
The University of Colorado Boulder’s Media Archaeology Lab offers a concrete example of studying functioning media through hands-on research and creative practice. Its collection includes optical, audio and computing technologies. Such encounters can expose alternatives that disappear when history is written only backwards from the current dominant device.
Compare tasks rather than prestige
Consider two fictional systems for distributing a community bulletin. System A produces paper copies that can be annotated and passed around without electricity. System B updates a shared digital page quickly and supports searching. Neither description establishes a universal winner. The answer depends on access, update frequency, reading circumstances, printing costs and the authority to alter the text.
Make a comparison around one task: correcting an incorrect venue. The digital page may change quickly, but someone holding an earlier screenshot may still receive the old information. Paper copies may be difficult to recall, but a dated correction can leave a visible record. Investigate which versions circulated and how readers identified the current one.
Avoid reversing the ladder into nostalgia. An awkward control is not automatically more thoughtful, and an obsolete format is not automatically more authentic. Ask which possibilities the arrangement supported, which costs it imposed, and for whom. The same constraint can be a creative resource for one user and an exclusion for another.
A credible alternative history does not need to claim that an abandoned system should have won. It needs to show that the path taken was contingent on decisions and circumstances. Distinguish a technically possible alternative from an alternative that historical participants could actually afford, access, maintain or imagine using.
CHAPTER 4 OF 14 · Establish evidence and test constraints
4. Establish provenance before assigning a date or purpose
Provenance concerns the history through which an object and the information about it reached you. Begin with acquisition records, labels, accompanying documents, known modifications and the source of each description. An apparently precise date can hide an uncertain inference. A catalogue may repeat a donor’s recollection without identifying it as such.
The specialist guide to metadata provenance explains why the lineage of a descriptive statement differs from the custody of the object itself. In a media inquiry, preserve both. Knowing who held a disk does not automatically establish who supplied its title or date.
Worked case: three dates, three possible events
A fictional collection contains a cassette with “1994” handwritten on its case, an accession record dated 2002 and an audio file exported in 2024. A researcher describes it as “a 2024 recording of a 1994 event”. That wording assumes more than the evidence establishes.
The accession date supports a claim about entry into that collection, assuming the record is accurate and linked to this item. The export date concerns a digital derivative. The handwritten year is evidence of a label, but its author and referent remain uncertain. It could mark recording, copying, cataloguing or an event mentioned in the audio.
The repaired description is narrower: “Cassette labelled 1994, accessioned in 2002, represented here by a 2024 digital export; original recording date unconfirmed.” A programme naming the speakers and event, or a documented transfer log identifying an earlier master, could strengthen the chronology. Until then, uncertainty belongs beside the date.
Keep an assertion record in plain language: claim, supporting source, source creator, date of assessment and confidence or unresolved problem. This is especially useful when new evidence changes an attribution. Retaining the earlier statement as a superseded interpretation helps future researchers understand why catalogues disagree.
Chronology also limits causal claims. A later device cannot explain an earlier production decision merely because the surviving copy now depends on it. Separate the history of creation from the histories of transfer, restoration and present access.
CHAPTER 5 OF 14 · Establish evidence and test constraints
5. Treat operation as an experiment with limits
Operating a historical system can reveal details that a photograph leaves hidden: waiting, resistance, feedback, sequence, error recovery and the need to remember commands. Those observations are valuable because an interface is partly an arrangement of actions through time. They become stronger when the trial has a stated question and a record of its conditions.
Operation today does not transport the researcher into an earlier world. A restored component, unfamiliar user, modern display or different room can alter the encounter. The appropriate claim is usually about what this surviving or reconstructed setup does under the documented conditions. Claims about earlier users require additional evidence.
Design a small repeatable trial
For a fictional text editor, choose a task such as changing a phrase on page one and checking its counterpart on page five. Record the starting document, application version, display arrangement, navigation commands, instructions given and result. Repeat the task from the same starting state so that a saved cursor position does not make the second trial artificially easier.
Observe where the user hesitates. Is the command undiscoverable, the response delayed, or the instruction misunderstood? Separate system behaviour from the operator’s learning. A novice’s first unsuccessful attempt cannot establish that experienced users found the system unusable.
Where possible, compare the observation with a manual, a contemporary demonstration and a user account. Agreement can support a narrow reconstruction. Disagreement is productive: a manual may describe intended operation, a demonstration may omit delays, and an interview may compress years of changing practice into one memory.
Do not operate fragile or unfamiliar equipment merely to complete an exercise. A working collection may have supervised access and handling procedures; follow them. For a beginner, published documentation and an authorised demonstration can support useful inquiry without electrical repair, opening sealed components or risking a unique carrier.
Finish the trial with two statements: what the setup demonstrated, and what it could not establish. For example, it may reveal that comparing distant passages requires several navigation actions. It cannot, by itself, prove that historical writers produced shorter arguments because of that feature.
CHAPTER 6 OF 14 · Establish evidence and test constraints
6. Calculate a constraint without making it destiny
Numerical models can show why a media arrangement created pressure towards certain choices. They become misleading when an illustrative assumption is mistaken for a measured specification or when a technical limit is treated as a complete explanation of culture. State the model first, calculate it, then identify the historical evidence still missing.
Worked calculation: image size and transfer time
Assume an uncompressed raster image is 800 pixels wide and 600 pixels high, with three eight-bit colour channels per pixel. Ignore headers, metadata, row padding and compression. Its pixel payload is 800 × 600 × 3 = 1,440,000 bytes. At eight bits per byte, that is 11,520,000 bits.
Assume a constant payload transfer rate of 28,800 bits per second. The ideal transfer time is 11,520,000 ÷ 28,800 = 400 seconds, or 6 minutes 40 seconds. This is an intentionally simplified payload model, not a claim about a particular modem, network or historical download. Connection overhead, retransmission, contention and compression can change the observed time.
Now compare a 640 × 480 image using one eight-bit palette index per pixel. Ignoring the palette table and other overhead, its pixel payload is 640 × 480 = 307,200 bytes. The same model gives 307,200 × 8 ÷ 28,800 ≈ 85.3 seconds. The first payload is 4.6875 times larger.
Interpret what the answer means
The comparison establishes a possible incentive to reduce dimensions or representation size. It does not prove that a historical designer chose indexed colour because of transmission cost. Display capabilities, authoring software, aesthetic preference and distribution media could also matter. A production note discussing download time would connect the calculation to an actual decision.
Changing more than one variable requires care. This example changes both dimensions and colour representation, so it cannot isolate the contribution of either. Holding dimensions at 800 × 600 while changing from three bytes to one byte per pixel would yield a threefold payload difference under the same assumptions.
Use the calculation to ask better archival questions: was the material transmitted, carried on physical media, or installed locally? Was delay visible to the user? Did an access copy differ from the production master? A number becomes historical explanation only when it is connected with the situation in which it mattered.
CHAPTER 7 OF 14 · Investigate interfaces and reconstruction
7. Read interfaces as arrangements of action and attention
An interface distributes what is visible, what can be changed, what must be remembered and what counts as a successful action. A labelled button, a command line and a physical dial each organise the relationship between intention and response differently. Describing those arrangements is more informative than calling one interface primitive and another intuitive.
Begin with a sequence. What must the person know before starting? What action changes the state? How is the change indicated? Can it be reversed? Which errors are recoverable? What happens when the user pauses or leaves? These questions connect surface appearance to operation without assuming that everyone approaches the interface in the same way.
Worked case: saving a draft
Two fictional writing systems offer different saving behaviour. In A, the user selects Save and chooses a filename. In B, changes appear to save continuously, while named versions require a separate action. Both can protect work, but they make different aspects of responsibility visible.
A researcher interested in authorship might investigate when writers considered a draft finished. A researcher interested in collaboration might examine who could restore an earlier version. A researcher interested in labour might study the time spent naming, copying and transporting files. The same saving mechanism supports several historical questions.
Do not infer practice from the button alone. A workplace may require paper approvals even when automatic saving exists. Users may keep their own copies because they distrust the system. Training notes, naming conventions and actual document histories can reveal these practices better than a promotional screenshot.
Accessibility is also a historical question about particular users and circumstances. A small visual target, an unlabeled control or a command requiring simultaneous key presses may affect different people differently. Record the barrier you can demonstrate; do not use a single present-day test to speak for all past users.
When comparing interfaces, specify the task and the participant’s experience. Faster completion by one trained person is evidence about that comparison, not proof of universal superiority. A persuasive account explains how the interface shaped an activity while leaving room for learned workarounds, institutional rules and individual variation.
CHAPTER 8 OF 14 · Investigate interfaces and reconstruction
8. Separate emulation, migration and documentation
Several methods can make historical media available, but they preserve different kinds of evidence. Format migration changes a representation so that another system can use it. Emulation recreates aspects of an environment needed to run historical software. Documentation records features or performances through descriptions, screenshots, video, diagrams or testimony.
The Digital Preservation Coalition’s preservation-action guidance distinguishes format migration from moving data between storage media and discusses emulation alongside other approaches. Its useful caution is that obsolescence is contextual: a system can remain usable in one institution after another has stopped supporting it.
Choose according to the historical question
Imagine an interactive poem in which clicking a word changes the next available sentence. A transcript can preserve visible wording but may omit the relationship among choices. A screen recording can document one route through the work. An emulated version may permit new interactions, but its response timing and display still need assessment.
For a question about vocabulary, the transcript may be highly useful. For a question about choice and anticipation, the interactive behaviour matters. For a question about exhibition design, photographs, room plans and installation records may be essential. No single access version answers every question equally well.
The CLIR overview of emulation as a preservation method introduces the practical relationship between historical software and contemporary environments. It supports treating emulation as a documented method of access, rather than a promise that the entire original experience has returned unchanged.
When using an emulation, identify the historical object, the environment being recreated, the emulator and its configuration, and the input and output equipment used for the session. Record relevant departures from the earlier setup. A different controller or altered display aspect ratio may be immaterial for one question and decisive for another.
Use How Emulation Works for Digital Preservation for the dependency chain. Keep the original object and documented derivatives distinguishable. A successful launch shows that something runs; the next chapter asks what evidence would justify a stronger claim about its behaviour.
CHAPTER 9 OF 14 · Investigate interfaces and reconstruction
9. Test the properties on which your interpretation depends
Preservation and historical interpretation meet when a claim depends on a property that may have changed. If you argue that a game demanded rapid decisions, timing matters. If you study a visual composition, spacing and colour relationships may matter. If you study textual revisions, exact wording and version order may matter more than the display casing.
The significant-properties guide helps articulate that choice. State what must remain sufficiently stable for your particular question and how you will compare it. A vague declaration that an object has been faithfully preserved is less useful than a list of tested behaviours and known differences.
Worked calculation: a timing discrepancy
In a fictional comparison, an authorised reference recording shows a demonstration sequence completing in 120 seconds. A reconstructed environment completes the same sequence in 100 seconds. Assume the input sequence and endpoints are comparable. The duration has fallen by 20 ÷ 120 ≈ 16.7 percent. The execution rate, measured as completed sequences per unit time, is 120 ÷ 100 = 1.2 times the reference rate, or 20 percent faster.
These percentages answer different questions. Saying “20 percent shorter” would be incorrect. Before attributing the discrepancy to an emulator, check whether the reference video was edited, whether loading time was included, whether input timing matched, and whether the reconstructed software version is the same.
Research by Mark Guttenbrunner and Andreas Rauber on evaluating an emulation environment specifically examines rendering and timing characteristics. The general lesson for this teaching case is to test the behaviour relevant to the claim, rather than infer fidelity from an apparently correct screenshot.
A discrepancy does not make the reconstructed object useless. It changes which claims it can support. You may still inspect menu structure while withholding conclusions about reaction time. Record that boundary so that a later researcher does not unknowingly reuse a demonstration for a purpose it was never tested to support.
CHAPTER 10 OF 14 · Explain circulation and repair weak claims
10. Follow media through work, institutions and circulation
Technical operation is only part of a medium’s history. Someone buys equipment, supplies materials, trains operators, authorises publication, repairs faults and decides what to retain. The cost and organisation of those activities help explain why technically possible practices were adopted in some places and absent in others.
Consider a fictional local newsletter. A printing device may support several layouts, but available staff time determines which is practical. A distribution rule may exclude material that the device could reproduce perfectly. A shortage of consumables may matter more than maximum speed. Evidence about procurement, staffing, schedules and distribution connects the machine with the institution using it.
Distinguish availability from use
An advertisement establishes that a product was offered under particular terms. A purchase record establishes acquisition. Neither alone proves routine use. A worn control, service invoice, operator log and surviving output might jointly support that further claim. Each source contributes a different part of the account.
Likewise, a capability described in a manual may have been optional, expensive or rarely used. A researcher should not treat every advertised feature as a daily experience. Search for workarounds: handwritten annotations, copied templates, shared instructions or improvised adapters can reveal how people made a system fit their circumstances.
The existing Music History guide illustrates how recordings and instruments connect with institutions and circulation. Sound Studies extends the inquiry into listening and social meaning. Media archaeology contributes questions about the material and operational arrangements through which those practices became possible.
When writing the explanation, name agents where evidence permits: operators, designers, distributors, maintenance workers and audiences. Avoid making “technology” the grammatical actor that causes everything by itself. A more precise account says which feature changed which task, under which organisational conditions, with which documented consequences.
The goal is neither to dissolve machines into social context nor to explain society from specifications alone. It is to trace the connection closely enough that another reader can inspect where the explanation is well supported and where an additional source is needed.
CHAPTER 11 OF 14 · Explain circulation and repair weak claims
11. Account for survival, absence and unequal visibility
A collection is a surviving selection. Some objects were expensive enough to keep, some were collected because they became famous, and some remained because nobody emptied a cupboard. Others were discarded, damaged, recycled or never documented. The proportion visible today can differ sharply from the proportion once in use.
Worked case: a collection that appears to show dominance
A fictional archive holds 80 examples of system A and 20 of system B. A researcher concludes that A had 80 percent of the historical market. That conclusion has no defensible denominator. The archive measures surviving accessions, not all historical purchases or users.
Suppose, purely to demonstrate the problem, that 20 percent of A units and 5 percent of B units entered this archive. Dividing observed counts by those assumed inclusion rates gives 80 ÷ 0.20 = 400 A units and 20 ÷ 0.05 = 400 B units. Under that hypothetical model, equal underlying numbers produce a four-to-one archival ratio.
The calculation does not recover the real population because the inclusion rates are invented. It proves a logical point: the collection counts alone cannot identify historical prevalence. To estimate prevalence, seek independent sales, distribution, ownership or usage evidence and examine the coverage of those records too.
Missing evidence can still shape an inquiry. If a collection lacks manuals, ask whether they were stored separately or excluded from acquisition. If users are absent from the catalogue, seek oral histories or organisational records. Do not convert an unsuccessful search into proof that a practice never existed.
Private correspondence and personal storage can contain material outside the purpose for which access was granted. Keep investigation within collection permissions and agreed research scope. An old carrier is not automatically an unrestricted source. A responsible account can describe the limits of available evidence without exposing unrelated personal material.
Finally, distinguish absence from deliberate silence. A missing recording may reflect technical failure, disposal, selection or a decision not to record. Choose among those explanations only when supporting evidence exists. An honest gap is more informative than a dramatic story constructed to fill it.
CHAPTER 12 OF 14 · Explain circulation and repair weak claims
12. Diagnose a weak explanation and repair the exact step
Historical arguments often fail at an identifiable transition rather than everywhere at once. A source may be genuine while the conclusion is too broad. A reconstruction may be useful while its description overstates fidelity. Finding the exact transition allows a proportionate repair.
The object looks old, so its contents must be old
The error is transferring a date from carrier appearance to recorded content. Repair it by distinguishing carrier manufacture, inscription, copying, accession and current access. If those events cannot be dated, use a range or an explicit unknown. Avoid giving a single year the burden of several different histories.
The feature exists, so it caused the practice
The missing step is evidence of use and consequence. Identify a plausible mechanism, then seek records connecting the feature with actual decisions. If only the mechanism is established, write that the feature made a practice possible or imposed a constraint, rather than claiming it determined what everyone did.
It runs, so the experience is identical
The error is treating operational success as complete equivalence. Check the properties needed for your claim: timing, inputs, display, sound, sequence or state. Describe untested differences. A modern access version can be valuable without being presented as a complete recreation of an earlier encounter.
Several websites repeat the claim, so it has several sources
Trace the references. Ten descriptions may derive from one catalogue entry. They are ten repetitions, not ten independent observations. Locate the earliest identifiable assertion and the evidence behind it. If the trail stops at an unattributed caption, make that limitation visible rather than strengthening the wording through repetition.
A surprising resemblance proves direct ancestry
Similarity can suggest a research question, but transmission needs a route. Look for contact, shared personnel, documentation, citations, demonstrations or access to an earlier design. Independent responses to similar constraints are another possibility. Keep resemblance, influence and direct derivation as different claims.
To review your own draft, underline every sentence containing “therefore”, “because”, “first”, “everyone” or “identical”. Those words are not forbidden; they identify places where the evidence must carry particular weight. Add the missing link, narrow the sentence or explain the uncertainty. The repair should improve the argument, not merely add cautious adjectives.
CHAPTER 13 OF 14 · Transfer the method and continue learning
13. Transfer the method to a contemporary platform
The method should remain useful when the object no longer looks historical. Consider a fictional museum that wants to study a recently discontinued messaging service. It has screenshots, exported text files, an installation package and interviews with six former users. Its proposed exhibition says that the service made conversations disappear after one day and therefore encouraged spontaneous speech.
Try the analysis before reading the answer
Identify which evidence could establish the disappearance rule, which could establish what users believed, and which could support the claim about speech. Then decide what the installation package would need before it could reproduce the relevant behaviour. Finally, write an exhibition statement that remains defensible if the original server cannot be recreated.
A worked answer
Screenshots document selected visible states, provided their provenance is known. Exported files may preserve wording while omitting timing, reactions, deleted material or interface cues. Interviews describe remembered experiences and interpretations; they do not automatically establish a universal technical rule. Contemporary documentation and controlled observations could help determine whether disappearance depended on settings, account type or version.
The installation package alone may not contain the server, account state, authentication service or remote configuration that controlled behaviour. A functioning local interface could therefore differ substantially from the service as used. The research record should identify which dependencies survive and which are unavailable, rather than label any successful launch a full restoration.
The spontaneity claim needs another evidential step. Users might report feeling less permanent, but that does not establish that all messages were more spontaneous. Compare accounts and surviving practices, including screenshots or copying that could defeat expected disappearance. If the evidence supports only the six interviewees’ perceptions, keep the statement at that scale.
A defensible exhibition label might read: “These six interviewees recalled treating some conversations as temporary. The preserved screenshots show selected interface states, while exports retain part of the wording. We have not reconstructed the server behaviour, and the relationship between perceived ephemerality and speech remains an interpretation supported by these accounts.”
The transfer succeeds when you recognise the familiar structure inside a different setting: object, dependencies, operation, use, circulation and evidence. The old-computer vocabulary can disappear while the reasoning remains. That is the educational value of the method.
CHAPTER 14 OF 14 · Transfer the method and continue learning
14. Build a small inquiry and choose the next route
Choose one accessible media object with a manageable question. A documented projection device, an authorised software demonstration or a public collection’s digitised object can be enough. Begin with observation and source comparison; you do not need a rare collection or a technical restoration project to learn the discipline.
Produce a short research record containing the object’s identity, your question, the source trail, a diagram or description of its dependencies, the observations made and the claim you can defend. Include one competing explanation and one piece of evidence that would distinguish it from your preferred account. Let someone else trace a key assertion back to its source.
Is media archaeology simply the history of old computers?
No. Computing provides important examples, but the inquiry can include optical devices, recording systems, print, projection and other media arrangements. The useful boundary is the question about mediation and historical possibilities, not a minimum age or a particular category of collectible hardware.
Does the object have to work?
No. A non-working object can preserve design, wear, modification and material evidence. Manuals, output samples and other documentation may support an account of operation. State which claims rest on direct testing and which depend on documentary reconstruction. Do not treat a non-working survivor as evidence that it never worked effectively.
Can a replica answer a historical question?
Yes, when its construction and limits are explicit. A replica can test whether a proposed arrangement is feasible or illustrate a mechanism. It cannot by itself establish that the arrangement was actually built or used in the past. Keep feasibility and historical occurrence separate.
Is an emulator enough to preserve an interactive work?
It may preserve important functionality, but the answer depends on the work and the question. Inputs, timing, external services, installation conditions and documentation may also matter. Use the tested-property approach in Chapter 9 and the existing emulation and digital-preservation guides for deeper technical investigation.
How do I know when a small project is complete?
You can explain the object’s relevance to the question, show the evidence behind the main claim, identify a plausible alternative and state the remaining uncertainty. Completion does not mean discovering everything about the device. It means finishing a bounded argument that another person can inspect and, if necessary, correct.
Primary and specialist reading routes
For a working collection and experimental practice, explore the Media Archaeology Lab’s own website. For format identification and sustainability questions, use the Library of Congress digital formats resource. For preservation methods, continue with the Digital Preservation Coalition handbook and the CLIR emulation report. These sources support specific methods; they do not turn an illustrative case into a documented historical finding.
Within eduKate, follow metadata provenance when a description needs an evidence trail, significant properties when access changes the object, and Sound Studies when listening is central. Return to the How X Works hub to investigate another mechanism, or the Education guide to test whether your method transfers to a new task.