What does a file carved from unallocated space lack that a file recovered through its MFT record keeps?
answer
- content without context
- the name never lived in the bytes
- header magic in, footer or length out
- assumes contiguity, breaks on fragmentation
- internal EXIF time is not a filesystem time
basics
~20 sA carved file is content with no context: no filename, no path, no owner, no filesystem timestamps. An intact MFT record keeps all of those bound to the data, so you can say where the file lived and when.
solid answer
~50 sCarving searches unallocated space for known format headers -- for example a JPEG's FF D8 FF opening, a PDF's %PDF- marker, a ZIP-family container's 50 4B 03 04 -- and reads forward to a footer or a length taken from the format itself. What comes out is bytes, and only bytes. There is no filename, no parent directory, no MACB timestamps, no security descriptor, so you cannot say the file lived in the folder under investigation or when it got there. Recovery through a still-intact but freed MFT record is different in kind: the record binds the content to a name, a parent reference, timestamps and a size, so the resulting claim has provenance. Carving also assumes contiguity, so a fragmented file yields a truncated or corrupt object, and header matching produces false hits on embedded thumbnails and on data that merely happens to start with the right bytes.
go deeper
Know that carving finds files by their format's header bytes in unallocated space, and that the result has no filename attached to it.
Explain the contiguity assumption, why fragmentation and partial overwrite corrupt output, and which metadata is lost because it lived in the MFT record rather than in the file.
Show you word findings to match the recovery method, and that you strengthen a pathless carve by hashing against a known copy rather than implying a location you cannot support.
Own the expectation-setting with whoever commissioned the examination: what a carve can and cannot answer, before the effort is spent chasing content that would not settle the question anyway.
## Two different recoveries When data has been deleted, there are two fundamentally different ways to get it back, and interviewers ask this because candidates routinely present the weaker one as if it were the stronger. **Record-based recovery.** The MFT record has been freed but not reused. It still describes the file: its name, the reference of its parent directory, its $STANDARD_INFORMATION and $FILE_NAME timestamps, its size, its security descriptor, and the cluster runs holding its data. Following those runs gives content *bound to* identity. You can say: this named file lived at this path, was this large, was created then and deleted then. **Carving.** The record is gone, or was never findable, so you search raw unallocated space for structure that the file format itself provides. Every common format opens with a recognisable magic value: JPEG with FF D8 FF, PDF with the ASCII %PDF-, GIF with GIF87a or GIF89a, and the entire ZIP family -- which includes modern Office documents, since a .docx or .xlsx is a ZIP container -- with 50 4B 03 04. Some formats give you an end marker (JPEG's FF D9, PDF's %%EOF) and others give you a length field in the header, and the carver reads forward accordingly. ## What carving cannot give you Everything that lived in the filesystem rather than in the file: - **No filename.** The name lived in the MFT record and the parent index, not in the bytes. - **No path.** You cannot place the object in a folder, which in an insider case is precisely the claim under dispute. - **No filesystem timestamps.** No created, modified, accessed or MFT-changed time. You may find timestamps *inside* the content -- EXIF capture time in a photograph, a document's internal creation metadata -- but those are author-controlled fields describing the document, not the filesystem's record of when the file sat on this volume. - **No owner or ACL.** Nothing ties it to an account. - **No deletion time.** Nothing says when it stopped being allocated. So the honest sentence about a carved object is: "these bytes were physically present in unallocated space on this volume". That is real, and it is much less than people assume. ## The failure modes of carving itself - **Fragmentation.** Simple carvers assume the file occupies contiguous clusters. If the file was fragmented, reading forward from the header runs into unrelated data, and the output is truncated, corrupt, or silently contaminated with someone else's bytes -- the most dangerous outcome, because the object still opens. - **Partial overwrite.** If some clusters were reallocated and rewritten, the carve yields a mixture. A JPEG that renders its top third and then turns to noise is the classic picture of this. - **False positives.** Any run of bytes beginning with a magic value looks like a file. Thumbnails embedded in larger documents, format headers inside archives, and coincidental byte sequences all produce spurious hits, so a carve of a large volume returns thousands of candidates that must be triaged. - **No end marker.** Formats without a footer or a length field are carved to a fixed maximum size, producing objects padded with unrelated trailing data. ## Using both honestly in a report A carved object still has value: hash it and compare against a known copy, and you have shown that *this specific content* was on the volume. Match its internal metadata against a document you already hold and the link tightens. But if the question you were asked is "did the design folder contain the drawings", a carved drawing with no path does not answer it -- it answers a related, weaker question. Say which question you answered. When you present carved output, describe how it was produced. "Recovered from the folder" is a claim about location that carving cannot support; "recovered from unallocated space on the same volume, not attributable to a directory" is what actually happened, and the difference is exactly the kind of overreach a competent challenge will find.
- A carved photograph has an EXIF timestamp. Can you use it as the file's creation time on this laptop?No. EXIF is metadata written by the camera or editing software into the file's own bytes, describing the image. It says nothing about when the file arrived on this volume, and it is trivially editable. Filesystem timestamps live in the MFT record, which carving does not recover, so a carved object simply has no filesystem times.
- How would you strengthen a claim about a carved file when you have no path for it?Anchor it to something outside the carve. Hash it and match against a known-good copy from a server, a backup or version control, which shows the specific content existed. Cross-reference its name and reference number in the change journal if a matching name is there. And where the object came out of the volume can sometimes be correlated with surviving records, but say plainly when it cannot.
Record-based recovery is finding a document still in its labelled folder. Carving is finding loose pages in a recycling bin: you can read them, but nothing tells you which cabinet they came out of.
saying these in an interview costs you the question
- Reports a carved file as recovered from a specific folder
- Treats internal EXIF or document metadata as filesystem timestamps
- Assumes carving output is complete and uncorrupted
- Ignores fragmentation as a source of contaminated output
- Thinks a carver knows the original filename