What Changed in My Bitcoin Setup

Detecting Unnoticed Changes in Custody Over Time

This memo is published by CustodyStress, an independent Bitcoin custody stress test that produces reference documents for individuals, families, and professionals.

The Baseline Problem

A holder looks at their bitcoin custody and wonders: what changed in my bitcoin setup? Something feels different. Maybe a device looks unfamiliar. Maybe a file was moved. Maybe a memory conflicts with what the holder sees in front of them. The question emerges, but the answer remains unclear. Without a record of what existed before, change cannot be identified with certainty.

This page examines how change detection fails when baseline documentation is absent. Holders cannot compare current state to original state if original state was never recorded. The question "what changed" assumes a reference point that may not exist. In its absence, the holder faces uncertainty that cannot be resolved through examination of current conditions alone.

---

The Baseline Problem

Detecting change requires comparison. The current state must be measured against a prior state. If no prior state was recorded, comparison becomes impossible. The holder can see what exists now, but cannot determine what existed before. Current observation provides no window into the past.

Most custody setups never establish a formal baseline. The holder creates the setup, perhaps tests it, and moves on. They do not photograph the arrangement. They do not write down every detail of the configuration. They do not record firmware versions, software settings, or storage locations with enough precision to enable later comparison. The moment passes, and with it goes the reference point.

Memory becomes the informal baseline, but memory is unreliable for this purpose. Holders remember general impressions. They remember that they set something up, that it seemed to work, that they felt confident at the time. These memories lack the specificity needed for change detection. "I think the hardware wallet was in the desk drawer" is not the same as "The Ledger Nano S with serial number X was in the second drawer from the top, right side, in a gray box labeled 'crypto.'"

Without documented specificity, the holder is left comparing current state to a fuzzy recollection. Fuzzy recollections accommodate many different current states. The holder cannot distinguish between "nothing changed" and "something changed but my memory was never precise enough to detect it."

---

Types of Undetectable Change

Different kinds of changes escape notice in different ways. Some changes are invisible because they occur in components the holder does not regularly examine. Others are invisible because the change is subtle enough to fall within the range of normal variation. Still others are invisible because the holder has no reference for what the unchanged state looked like.

Configuration changes happen inside software without visible external sign. A wallet application updates its settings. A hardware wallet firmware version advances. A passphrase option that was off is now on, or vice versa. The device looks the same from the outside. The change lives in digital state that requires active inspection to observe.

Physical changes accumulate gradually. A label fades. A box moves slightly. Dust covers a device that was once clean. Each individual change is small. The cumulative effect may be large, but no single transition triggered notice. The holder adapts to the new state without recognizing it as new.

Environmental changes affect the context around the custody rather than the custody itself. A roommate moved out and had access to the storage area. A cleaning service now enters the home office. A family member discovered where backups were kept. These changes do not alter the bitcoin setup directly, but they alter the security assumptions the holder operates under. The holder may not track these environmental factors at all.

---

The Reconstruction Attempt

Faced with uncertainty, holders often attempt to reconstruct original state from memory and inference. They think back to when they set things up. They try to remember where they put things and how they configured them. This reconstruction is speculative. It fills gaps with assumptions that may or may not reflect what actually happened.

Reconstruction confidence often exceeds reconstruction accuracy. The holder builds a mental picture of the original setup. The picture becomes more vivid with repeated consideration. Eventually the holder treats the reconstruction as memory, even though it was assembled from fragments and guesses. The distinction between what was remembered and what was inferred disappears.

Conflicting evidence disrupts reconstructed narratives. The holder believes the seed phrase was written on white paper. They find it on yellow paper. Now they face a puzzle: did they misremember the paper color, or did someone copy the phrase onto different paper? The reconstruction cannot accommodate the conflict without revising itself. Each revision introduces new uncertainty about which elements of the reconstruction are accurate.

Other people involved in the original setup may have different reconstructions. A spouse remembers the conversation differently. An advisor recalls giving different instructions. These alternative versions are not necessarily wrong, but they cannot all be correct. The holder must decide whose reconstruction to trust, often without any way to verify any of them.

---

Implicit Versus Explicit State

Some aspects of a custody setup exist only in implicit form. They were never written down because they seemed obvious. They existed in the holder's understanding of how things were arranged. When the holder later asks what changed, these implicit elements are especially vulnerable to distortion.

The location of backup materials may have been implicit. The holder "knew" where the seed phrase was stored, but that knowledge lived only in their head. When they go to retrieve it and find it somewhere unexpected, they cannot determine whether the phrase moved or their memory was always imprecise. The implicit state left no trace that could be compared.

Security assumptions often remain implicit. The holder assumed certain people could not access certain areas. They assumed certain devices were never connected to the internet. They assumed certain information was never photographed or copied. These assumptions were part of the custody design, but they were never documented. Years later, the holder cannot verify that the assumptions held throughout.

Explicit state can still be incomplete. A holder who wrote down their seed phrase documented that element explicitly. But they may not have documented which derivation path they used. They may not have noted whether a passphrase was employed. The explicit documentation captures some aspects while implicit assumptions cover others. The mix of explicit and implicit creates uneven visibility into what might have changed.

---

Time-Based Drift

Change accumulates continuously over time. Each day introduces small possibilities for drift. A single day rarely matters. A thousand days matter enormously. The holder who asks what changed after several years faces a fundamentally different question than one who asks after several days.

Longer intervals allow more changes to accumulate. More software updates occur. More environmental factors shift. More memory degradation takes place. More people may have encountered the setup. The total space of possible changes expands with time. At some point, the space becomes too large to enumerate.

Shorter intervals offer less drift but may also offer less motivation to check. A holder who checked their setup yesterday feels no urgency to check today. This reasonable approach leaves small changes unexamined. The small changes compound. Eventually the holder looks and finds enough accumulated change to trigger concern, but by then too many individual changes have merged into a general state of uncertainty.

Time-based drift is not linear in its effects. Some changes cancel out. Others reinforce each other. A firmware update might fix a previous update's problem. A moved backup might be moved back. The net effect of many changes over time is not simply the sum of individual changes. Interaction effects create outcomes that no single change would have produced alone.

---

Evidence of Change Without Identification

Holders sometimes detect that something changed without being able to identify what. A sense of wrongness persists. The setup looks different somehow. A feeling of unfamiliarity attaches to something that was once familiar. This intuition may be accurate, but it provides no actionable information.

Physical arrangement might trigger the sensation. Items are not quite where they were expected. The holder cannot specify what moved, only that something seems out of place. The feeling could indicate real tampering. It could also indicate normal memory drift. The holder cannot distinguish between these explanations from the sensation alone.

Device behavior might seem subtly different. A hardware wallet takes longer to start up. A software interface looks slightly altered. The holder notices but cannot say whether the change is significant. Software evolves. Devices age. Some changes are cosmetic. Some changes matter. Without understanding what changed, the holder cannot assess severity.

Relationships and access patterns might feel off. The holder thinks someone is acting strangely around the topic of bitcoin. They suspect someone may have seen something they were not supposed to see. These suspicions rest on behavioral interpretation rather than direct evidence. They may reveal genuine security events or may reflect unrelated social dynamics. The holder is left with suspicion rather than knowledge.

---

The Forensic Limit

Professional investigation cannot fully overcome the baseline problem. Forensic analysts can examine current state in detail. They can identify inconsistencies within that state. What they cannot do is reconstruct a past state that was never documented. Their expertise helps interpret evidence, but it cannot create evidence that does not exist.

Digital forensics reveals what is present now and sometimes what was present recently. File metadata may show modification dates. Device logs may capture recent activity. But these records have limits. Metadata can be altered. Logs can be cleared. Older history disappears as newer activity overwrites it. The forensic window extends backward only so far.

Physical forensics can identify certain kinds of tampering. Seals that were broken leave traces. Materials that were handled may retain evidence. But not all changes leave physical traces. Photographing a seed phrase does not disturb it. Reading a document does not alter it. The most dangerous changes may be the ones that forensics cannot detect.

Ultimately, forensic analysis confronts the same problem the holder faces: without a baseline, comparison is limited. The analyst can describe current state with precision. They can identify anomalies within that state. They cannot identify changes from a prior state that was never recorded.

---

Living With Unknown Change

Holders who cannot determine what changed must decide how to proceed under uncertainty. The uncertainty does not resolve itself. It simply persists as a background condition. The holder continues to use their custody setup without knowing if it has been compromised in some way they cannot see.

Some holders respond by treating unknown change as insignificant. They reason that if they cannot identify what changed, the change probably does not matter. This reasoning may be correct. It may also be wrong. The reasoning itself provides no way to distinguish between the two outcomes. It is a choice to accept uncertainty rather than a resolution of it.

Other holders respond by treating unknown change as significant. They assume something has gone wrong and act accordingly. This may mean abandoning a setup that was actually fine. It may mean expending resources on unnecessary precautions. But it may also mean catching a real problem before it causes loss. The cost of caution exists either way.

The choice between these responses depends on factors outside the custody system itself: the holder's risk tolerance, the value at stake, the cost of precaution, the ability to recover from loss. The custody system provides no guidance. It simply sits in whatever state it sits in, changed or unchanged, detectable or not.

---

Outcome

The question "what changed in my bitcoin setup" exposes a structural limitation. Change detection requires baseline documentation that most holders do not create. Without a detailed record of original state, current observations cannot reveal what differs from the past. The holder sees what exists now but cannot see what existed then.

Reconstruction from memory fills the gap imperfectly. Memory lacks the precision needed for accurate comparison. Implicit assumptions were never documented at all. Time allows countless small changes to accumulate beyond enumeration. The space of possible changes eventually exceeds any investigation's ability to catalog.

Holders left with unidentifiable change must proceed under uncertainty. They cannot determine whether the setup remains in its original state. They can only decide how to act given that they do not know. This uncertainty is inherent to custody systems that were not documented at baseline, and no later examination can fully compensate for that missing record.

---

System Context

Examining Bitcoin Custody Under Stress

Bitcoin Foundation Endowment

Simplest Secure Bitcoin Setup

← Return to CustodyStress