What 2027 invalidates

Most of what the field learned between 2022 and 2026 still applies to 2027. Three changes genuinely break earlier work: the mandatory bioassay, the key-and-lock and hose hardware, and the commanded astronaut interaction. Everything else carries forward with a raised bar, and the schedule consequence is larger than any single task change.

The invalidation matrix

The three that genuinely break earlier work

1. The bioassay definition. Two life detections are required, and at least one must be a bioassay that detects a molecule formed only by biology or in its presence S-URC27-RULES S-URC27-QA §1.b.v AQ9. Amino-acid chemistry alone does not satisfy it, and neither does a mineral spectrum. A payload that scored 94 in 2026 therefore fails in 2027 without a hardware change S-WVU-2026-SCIPLAN, while a three-channel wet-chemistry payload from 2022 was already compliant.

2. The Equipment hardware. The key in a tubular pin-tumbler lock has no URC precedent, and its 1 × 1 cm tags are the smallest targets in the rulebook. The hose moves from a fitting with the gasket removed to a 1.5-inch cam lock with a half-thickness gasket, which is a stiffer push than 2026’s. The USB Type-C task is gone S-URC27-RULES S-URC26-RULES §1.d.ii §1.e.ii.

3. The commanded astronaut interaction. Follow!, Stay!, Fetch!, Come! and Give! are new as a command loop S-URC27-RULES §1.e.iv §1.e.x. The geometry around them has nine years of precedent, but no earlier rulebook contains a command gesture or a follow clause.

What carries forward

What the removals give back

• The stratigraphic profile leaves documentation and debrief, which frees one to three minutes of site time and one debrief bullet S-URC26-RULES S-URC27-RULES §1.b.iii §1.b.iii.
• The Delivery marked-path task removes one path-following family and its rehearsals. It does not reduce the terrain requirement.
• The USB Type-C task deletes one bankable Equipment task and all the fixtures and poses built for it.
• The 10-minute transition is not a removal but a constraint: any change in mass or configuration between Autonomy and Equipment forces a re-weigh inside the window S-URC27-QA EQ4.

What looks invalidated and is not

• A 2022 payload is not obsolete. The wet three-channel payload that scored 96 in 2022 was closer to the 2027 clause than the Raman instrument that replaced it in 2024.
• Mission scores are still evidence of capability. They transfer as evidence of documentation, sampling and debrief skill, even where the payload behind them does not transfer.
• The 2026 autonomy stack survives its mission. The 2026 mission structure is invalidated, and the stack that scored 86 in it is not.
• The 2026 discipline carries over. The LED and the hands-off rule remain, and the C2 banner becomes optional rather than required S-URC27-QA AUQ18.
• The Equipment base rate is unchanged. The five-season mean of 29–35 remains the best estimate for 2027, with the key task as extra downside and the removed USB task as a small offset.
• Terrain capability still decides Delivery. The marked path is gone, but the objects still sit across sand, gravel, boulders and steep loose slopes.

What this means for the rover

1. The design follows the score shape, not the mean. The guaranteed-floor work, which is documentation, sample, cache and debrief in Astrobiology and the suit plus short behaviours in Autonomy, is protected and never traded for 1 m route targets, gesture Follow! or the key.
2. The two 2027-specific payloads are frozen before the robotics. The bioassay pair and the astronaut interaction both need review evidence by 2027-03-05, while the drive and the arm can be tuned until June.
3. “Two life detections” is a payload-architecture decision. Two orthogonal biomolecule assays share one detector, with the AQ9 justification written into the Science Plan.
4. The geometry the rules give away is worth taking. The astronaut’s GNSS point, the 20 m standoff with the astronaut in view, and exact route coordinates convert hard perception into easy localisation.
5. Follow! is built as a ladder, not a modality. Gesture, then audio, then a sign or a device, so the floor is 5 points and the ceiling is 15.
6. Equipment is a precision problem with a time budget. The cheap tasks come first, because any order is allowed, then the drawer and cache, then typing, the key and the hose.
7. Autonomy stays boring and the stop discipline excellent. The best-documented 2026 gain came from replacing a sophisticated planner with a simpler one.
8. A link check never gates autonomous driving. The second route target is deliberately out of radio sight, and one link check is a 25-point failure.
9. The transition and the weigh-in are rehearsed as mission tasks. The 10-minute window, the 1.2 m box and the 50 kg limit can cost more than a weak arm, and they are cheap to practise.
10. Documentation is generated as the system is tested. The review is 100 points and the only route to the field, and the teams that average 94 have the artefacts in hand when the deadline arrives.

Further reading

• Mission-by-mission backfit carries the master task table and the score shapes behind each verdict here.
• Astrobiology mission works the payload change in detail.
• Equipment servicing works the key, the hose and the removed USB task.
• Autonomy: astronaut assistance works the command interaction.
• System Acceptance Review explains why the March deadline governs the build order.
• 110 engineering problems turns each item above into a task with an acceptance test.