Latency / Bandwidth / Transfer-Time Calculator mechanism

Model scope

The existing honest label is directionally correct, but the instrument should be labeled as an idealized serialization and bandwidth-delay model unless a precise RTT contribution model is added. The exact ambiguity requiring resolution is whether transfer_time means size/bandwidth only or total time including RTT, and, if RTT is included, how many RTTs are to be added.

• t=size/bandwidth is the standard serialization time, assuming size and bandwidth use consistent units, but it does not include the claimed round-trip latency contribution.
• The specification does not define how RTT contributes to total transfer time: adding one RTT is not universally correct, because the number of RTTs depends on protocol behavior, handshakes, acknowledgements, slow start, and transfer size.
• BDP=bandwidth×RTT is standard when bandwidth is in bits per second and RTT is in seconds; the result is an amount of data, normally reported in bits or converted to bytes.
• effective_throughput=min(bandwidth,window/RTT) is the standard idealized window-limited model, but only when bandwidth and window/RTT are converted to the same rate units.

Units and conventions

• Size must be converted to bits when bandwidth is in bits per second; byte-based sizes require multiplication by 8.
• Bandwidth and effective throughput are assumed to be reported in bits per second unless explicitly converted to bytes per second.
• RTT is assumed to be in seconds for both BDP and window/RTT calculations; millisecond inputs must be divided by 1000.
• TCP window size is assumed to be a byte quantity; it must be converted to bits before comparing window/RTT with a bit-per-second bandwidth, or the result must be reported in bytes per second.
• SI prefixes and binary prefixes must remain distinct: MB=1,000,000 bytes, MiB=1,048,576 bytes, while Mbps means 1,000,000 bits per second.

Inspectable vector

One of 4 source-bound review vectors:

{
  "inputs": {
    "file_size": 10,
    "file_size_unit": "MB",
    "bandwidth": 100,
    "bandwidth_unit": "Mbps",
    "RTT": 80,
    "RTT_unit": "ms",
    "TCP_window": 1,
    "TCP_window_unit": "MB"
  },
  "expected_outputs": {
    "transfer_time_s": 0.8,
    "BDP_bits": 8000000,
    "BDP_bytes": 1000000,
    "effective_throughput_bps": 100000000
  },
  "working": "10 MB = 10,000,000 bytes = 80,000,000 bits. Transfer time = 80,000,000 bits ÷ 100,000,000 bits/s = 0.8 s. RTT = 80 ms = 0.08 s. BDP = 100,000,000 bits/s × 0.08 s = 8,000,000 bits = 1,000,000 bytes. Window rate = 1,000,000 bytes ÷ 0.08 s = 12,500,000 bytes/s = 100,000,000 bits/s. Effective throughput = min(100,000,000, 100,000,000) = 100,000,000 bits/s."
}

Failure boundary

BREAK_CLAIMED · SELF_FLAGGED

A zero-bandwidth input makes transfer time undefined and must be rejected explicitly. The unit-ambiguity vector also produces materially different results if GB/MB are interpreted as binary rather than SI prefixes, so the interface must enforce or display the selected convention.

Other records

Sources and evidence

These pointers bind the page to evidence. They do not grant authority or collapse verification into deployment.

• VERIFIED_LEDGER.json
• evidence-tranche4/tranche4-results.json
• BOUNDARY_HAZARD_CATALOG.json