TLCTC · Evidence · CIS Controls v8.1 mapping

153 Safeguards. 74 objectives.

The full cell-by-cell mapping of every CIS Controls v8.1 Safeguard onto the TLCTC 10×6×2 matrix. The ≥60-objective reduction was a prediction. This is the count.

CIS Controls v8.1 (153 Safeguards, 18 Controls) · TLCTC 10×6×2 matrix · method & data below · CC BY 4.0 · updated 2 Oct 2026: data published
Update · 2 October 2026 · the per-Safeguard data is published

The data file this page promised was never published, and it could not be recovered. The per-Safeguard mapping is now public as a rebuild against CIS Controls v8.1.2 (March 2025; v8.1.1 and v8.1.2 changed glossary definitions only, the Safeguards are those of v8.1). It keeps the 34 Safeguards this page names exactly as shown below and places the rest on the same doctrine.

Mapping data (JSON) · README and method · load it as a starter in the Control Matrix (Templates → CIS Controls) or the DRE Control Matrix. The data holds Safeguard ids and TLCTC wording only: the CIS text is CC BY-NC-ND 4.0, so read it alongside your own copy of the Controls.

The rebuild does not reproduce the June counts below, for three structural reasons rather than disagreements over single rows:

Cluster matrixJune 2026 · prevention / D–RRebuild · prevention / D–R
#1 Abuse of Functions12 / 826 / 2
#2 Exploiting Server19 / 718 / 2
#3 Exploiting Client9 / 418 / 2
#4 Identity Theft17 / 426 / 1
#5 Man in the Middle5 / 28 / 0
#6 Flooding Attack1 / 23 / 1
#7 Malware12 / 1116 / 6
#8 Physical Attack8 / 13 / 1
#9 Social Engineering9 / 27 / 0
#10 Supply Chain7 / 310 / 1
Enablers / cluster-neutral4846 (+ 18 on the data layer)

The finding survives the rebuild: preventive load still piles onto #1, #2 and #4 while #6, #5 and #9 stay thin. The new data layer adds two gaps the June view could not show — no Safeguard of its own protects provenance (If), and almost nothing detects or responds at the Data Risk Event. The June analysis below is unchanged, as published.

RESULTThe count cascade · June 2026

Each CIS Safeguard was assigned the TLCTC cluster(s) whose generic vulnerability it acts upon, plus a Bow-Tie side (prevention / detection-response). Objectives are then counted at their true grain: (cluster, side, generic-strategy).

CIS count
153
Safeguards, each counted as one "control"
– Fail the floor
48
Pure enablers (inventory / process / governance) that act on no cluster — not control objectives
– De-duplicate
31
Repeated strategies & collapsed umbrella fragments hitting the same cell
Cell-pure objectives
74
Each granular enough to reach target-zero on one cell
Prediction confirmed

The blog forecast a reduction "on the order of 60 objectives." The mapping yields 153 → 74, a net reduction of 79 — the hypothesis holds, with margin. Two independent drivers produce it, pulling from opposite directions exactly as predicted.

DRIVER 148 Safeguards fail the granularity floor

Nearly a third of v8.1 acts on no cluster's generic vulnerability. These are inventories, processes, policies, and governance rows — they enable other controls but cannot themselves be measured to target-zero against any cause. Under the floor, they are not control objectives. Examples: 1.1 asset inventory, 3.1 data management process, all nine of Control 17 (Incident Response process rows), 16.1–16.3 dev-process rows, 18.1/18.3/18.4 pentest-program rows. Legitimate work — but not cause-acting controls.

DRIVER 234 Safeguards are umbrellas

These span more than one cluster and must de-aggregate. The classic case 7.4 Application patch management spans #2#3#10 — three causes, three velocity classes, one Safeguard. It cannot reach target-zero on any of them. After splitting, the fragments collapse into shared cells with other Safeguards (de-duplication), which is where the count comes back down.

SafeguardSpansWhy it's an umbrella
7.4#2#3#10App patching covers server-side, client-side, and third-party components
13.2 / 13.7#2#3#7Host IDS/IPS detects exploitation (server+client) and malware
8.11#1#4#7Log review spans abuse, identity, and malware indicators
2.5 / 2.6 / 2.7#1#7Allowlisting blocks both function-abuse and foreign code execution
5.4 / 6.8 / 3.3#1#4Access control acts on both abuse-of-function and identity
12.7#5#4VPN + AAA: anti-MITM transit and identity at once
13.9#8#4Port-level NAC: physical port control and identity
14.3 / 14.8#9 + #4/#5Awareness training folds social engineering with authn / transit

34 umbrellas total; representative rows shown. The full per-Safeguard data: mapping JSON (rebuild, see the update above).

COVERAGEAll 20 cause/side coordinates are touched

A reassuring finding for CIS: v8.1 does reach every cluster on both Bow-Tie sides — there are no empty cells at the cluster level. The problem is not absence of coverage; it is that coverage is distributed without a cause axis to balance it. Look at the load:

Prevention (left of SRE)Detection–Response (right of SRE)
#1 Abuse of Functions128
#2 Exploiting Server197
#3 Exploiting Client94
#4 Identity Theft174
#5 Man in the Middle52
#6 Flooding Attack12
#7 Malware1211
#8 Physical Attack81
#9 Social Engineering92
#10 Supply Chain73

Cell values = number of Safeguard-fragments landing in that (cluster, side) coordinate after de-aggregation (June 2026). In the rebuild two detection–response coordinates are empty (#5, #9) because general detection stays cluster-neutral; see the update above.

The imbalance the cause axis reveals

#2 Exploiting Server carries 19 preventive fragments (rebuild: 18). #6 Flooding Attack carries 1 (rebuild: 3, all incidental — no Safeguard is written for flooding). #5 Man in the Middle carries 5. This is not a risk-weighted distribution — it is an artifact of where the control set historically accreted. CIS RAM's expectancy engine cannot even see this skew, because its commonality axis is asset-class frequency, not cause. The matrix surfaces it instantly: massive redundancy on server exploitation, near-singleton coverage on flooding and MITM.

READWhat this proves — and what it doesn't

It proves: (1) the ≥60 prediction was conservative — the real figure is 79; (2) the reduction is not deletion of capability but reclassification — 48 enablers move out of the objective count into a separate "enabling" register, and umbrella fragments de-duplicate into shared cells; (3) every cluster is already covered, so adoption is a re-anchoring, not a rebuild.

It does not prove: the exact strategy-grain assignment is canonical. Cell counts depend on how finely the six generic strategies are cut within a cluster. A different but reasonable strategy taxonomy would move 74 by a handful either way. The direction and magnitude are robust; the precise integer is mapping-dependent and should be ratified against the canonical TLCTC strategy set before publication.

Bottom line for CIS

You do not have a coverage gap. You have a resolution problem: 153 line items that mix 48 non-controls, 34 umbrellas that cannot reach target-zero, and a server-exploitation cluster carrying 19× the load of flooding. Re-anchor onto the matrix and the set resolves to 74 cause-pure objectives plus a clean enabling register — every objective measurable, every gap nameable, every redundancy visible. DoCRA stays on top, untouched.

Context: Why CIS Cannot Answer Your Cyber Threat Risk · the blind spot, visualized: What CIS Cannot See.

Mapping: 153 v8.1 Safeguards → 10×6×2 matrix. 153 → 74 objectives (−79), June 2026. Strategy-grain assignment is illustrative and pending ratification against the canonical TLCTC strategy set. Per-Safeguard data (rebuild, October 2026): mappings/cis-controls-v8.1.