Obsidian/Writing/ERLM/ERLM-Proposal-Review-Summary.md

# ERLM Proposal Writing Review - Executive Summary

**Date**: December 2, 2025 **Reviewer**: Claude Code
**Framework**: Gopen's Sense of Structure

---

## Overview

This proposal demonstrates strong technical content, clear
methodology, and comprehensive coverage of all required
elements. The research approach is well-conceived, and the
progression from problem statement through solution is
logical. The writing is generally clear and professional.

**Key Strengths:**
- Excellent technical depth and specificity
- Strong motivation established through human factors
statistics
- Clear three-thrust research structure
- Comprehensive risk analysis with concrete contingencies
- Good use of specific examples (TMI accident, HARDENS
project)

**Priority Areas for Revision:**
- Sentence-level: Strengthen stress positions to emphasize
key claims
- Paragraph-level: Sharpen point-issue structure in some
sections
- Section-level: Tighten organization in State of the Art
section
- Big picture: Strengthen "so what" connections throughout

---

## Priority Issues (Top 10)

### 1. **SOTA Section Length and Organization**
[SECTION-LEVEL] **Location**: State of the Art section (358
lines) **Issue**: The SOTA section is the longest in the
proposal and covers multiple distinct topics (current
procedures, human factors, HARDENS). While comprehensive, it
risks overwhelming readers and obscuring your key
 contributions. **Impact**: HIGH - Reviewers may lose track
 of your argument in the density **Recommendation**:
 Consider restructuring with clearer signposting. Each
 subsection should explicitly connect back to what gap
 you're filling. The current "\textbf{LIMITATION:}" callouts
 are excellent—ensure every major subsection has one.

### 2. **Weak Stress Positions Throughout** [SENTENCE-LEVEL]
**Location**: All sections, especially Goals and State of
the Art **Issue**: Many sentences place old/known
information in stress position (sentence-final), missing
opportunities to emphasize new claims **Impact**:
MEDIUM-HIGH - Reduces rhetorical impact of key claims **See
Pattern**: "Stress Position Weakness" below for examples and
fixes

### 3. **Missing "So What" Connections** [BIG PICTURE]
**Location**: Transitions between major sections **Issue**:
The proposal moves from problem → approach → metrics without
always explicitly stating "this matters because..." at
transition points **Impact**: MEDIUM-HIGH - Reviewers may
not fully grasp significance **Recommendation**: Add
explicit "if successful, this enables..." statements at the
end of Goals section and beginning of Metrics section

### 4. **Passive Voice Obscuring Agency** [SENTENCE-LEVEL]
**Location**: Research Approach, especially subsection
introductions **Issue**: Passive constructions like "will be
employed" and "will be used" hide who does what and reduce
directness **Impact**: MEDIUM - Reduces clarity and makes
writing feel less confident **See Pattern**: "Passive Voice"
below

### 5. **Point-Issue Structure in Paragraphs**
[PARAGRAPH-LEVEL] **Location**: State of the Art, Risk
sections **Issue**: Some paragraphs present information
without first establishing why readers should care (the
"issue") **Impact**: MEDIUM - Readers may wonder "why are
you telling me this?" **See Pattern**: "Point-Issue
Structure" below

### 6. **Topic String Breaks** [PARAGRAPH-LEVEL]
**Location**: Research Approach, subsection transitions
**Issue**: Topic position doesn't always establish clear
continuity from previous sentence, forcing readers to
reconstruct connections **Impact**: MEDIUM - Increases
cognitive load **See Pattern**: "Topic Position &
Continuity" below

### 7. **Nominalization Hiding Action** [SENTENCE-LEVEL]
**Location**: Throughout, especially Research Approach
**Issue**: Action buried in nouns (e.g., "implementation"
instead of "implement", "verification" instead of "verify")
**Impact**: MEDIUM - Makes writing feel static rather than
dynamic **Recommendation**: Convert nominalizations to
active verbs where possible

### 8. **Long Complex Sentences** [SENTENCE-LEVEL]
**Location**: State of the Art (lines 45-51), Risks (lines
72-79) **Issue**: Some sentences exceed 40-50 words with
multiple subordinate clauses, challenging comprehension
**Impact**: MEDIUM - Reviewers may have to re-read
**Recommendation**: Break into 2-3 shorter sentences with
clear logical flow

### 9. **Subsection Balance in Risks Section**
[SECTION-LEVEL] **Location**: Risks and Contingencies
section **Issue**: Four subsections of vastly different
lengths (computational tractability gets more space than
discrete-continuous interface, despite latter being more
fundamental) **Impact**: LOW-MEDIUM - May suggest misaligned
priorities **Recommendation**: Consider whether space
allocation reflects actual risk magnitude

### 10. **Broader Impacts Underutilized** [BIG PICTURE]
**Location**: Broader Impacts section (75 lines vs 358 for
SOTA) **Issue**: This section is relatively brief given that
economic impact is a major motivation for SMRs **Impact**:
LOW-MEDIUM - Missing opportunity to strengthen value
proposition **Recommendation**: Consider expanding economic
analysis or adding brief discussion of workforce/educational
impacts

---

## Key Patterns Identified

### Pattern 1: Stress Position Weakness

**Principle** (Gopen): The stress position (end of sentence)
should contain the most important new information. Readers
expect climax at sentence-end and are disappointed when they
find old information or weak phrases there.

**Example 1** (Goals and Outcomes, lines 13-17): ```
Current: "Currently, nuclear plant operations rely on
extensively trained human operators who follow detailed
written procedures and strict regulatory requirements to
manage reactor control." ```
- **Issue**: Sentence ends with "manage reactor control"—a
restatement of the opening. The key claim is buried
mid-sentence: "extensively trained...detailed
procedures...strict requirements"
- **Fixed**: "Currently, nuclear plant operations require
extensively trained human operators following detailed
written procedures under strict regulatory requirements."

**Example 2** (State of the Art, lines 53-54): ``` Current:
"Procedures lack formal verification of correctness and
completeness." ```
- **Issue**: Ends weakly with "completeness" which is minor
compared to the bigger issue
- **Fixed**: "Procedures lack formal verification, leaving
correctness and completeness unproven."

**Example 3** (Research Approach, lines 41-42): ``` Current:
"The following sections discuss how these thrusts will be
accomplished." ```
- **Issue**: Pure metadiscourse in stress position, provides
no new information
- **Fixed**: Delete this sentence—the enumeration provides
sufficient transition, or combine with previous sentence:
"...through three main thrusts, each detailed below."

**Similar instances**:
- Goals lines 29-32: "...we will combine formal methods..."
- State of the Art lines 81-85: "...no application of hybrid
control theory exists..."
- Research Approach lines 115-116: "...enable progression to
the next step..."
- Metrics lines 29-31: "...makes this metric directly
relevant..."
- Risks lines 12-13: "...identification of remaining
barriers to deployment"

**How to fix**: Identify the most important new claim in
each sentence and move it to the end. Often this means
converting from "X does Y to achieve Z" to "X achieves Z by
doing Y."

---

### Pattern 2: Passive Voice Obscuring Agency

**Principle** (Gopen): Passive voice obscures who does what
and reduces directness. In proposal writing, active voice
demonstrates confidence and control. Use passive only when
the agent is truly unimportant or unknown.

**Example 1** (Research Approach, line 118): ``` Current:
"We will employ state-of-the-art reactive synthesis
tools..." ```
- **Issue**: "Employ" is weak; you're not hiring the tools,
you're using them
- **Better**: "We will use Strix, a state-of-the-art
reactive synthesis tool..."
- **Best**: "Strix will translate our temporal logic
specifications into deterministic automata..." (Shows what
the tool *does*, not just that you'll use it)

**Example 2** (Research Approach, line 207): ``` Current:
"Control barrier functions will be employed when..." ```
- **Issue**: Passive—who employs them? And "employed" sounds
formal/stuffy
- **Fixed**: "We will use control barrier functions to
verify..." or better "Control barrier functions verify..."

**Example 3** (Metrics, line 67): ``` Current: "This
milestone delivers an internal technical report..." ```
- **Issue**: Milestones don't deliver, people do
- **Fixed**: "We will deliver an internal technical report
documenting..."

**Similar instances**:
- Research Approach lines 161, 175, 206, 220: "will be
employed", "will be developed", "will be used"
- Metrics lines 69, 73, 79, 84: "...delivers a [document]"
- Risks lines 57, 109, 163: various passives

**How to fix**:
1. Identify the real agent (usually "we")
2. Make agent the subject: "We will X" or "X will Y"
3. Choose strong active verbs: use/apply/develop/verify (not
employ/utilize)

---

### Pattern 3: Point-Issue Structure Weakness

**Principle** (Gopen): Paragraphs should begin by
establishing (1) the point/claim being made and (2) why it
matters (the issue). Discussion then supports that point.
Readers need context before details.

**Example 1** (State of the Art, lines 88-107): ``` Current
paragraph begins: "The persistent role of human error in
nuclear safety incidents, despite decades of
improvements..." ```
- **Analysis**: This paragraph immediately dives into the
"persistent role" without first establishing why we're
discussing human factors at all. Reader thinks: "Wait,
weren't we just talking about procedures?"
- **Fixed**: Add issue statement first: "Human factors
provide the most compelling motivation for formal automated
control. Despite decades of improvements in training and
procedures, human error persists in 70-80% of nuclear
incidents—suggesting that operator-based control faces
fundamental, not remediable, limitations."

**Example 2** (Risks, first paragraph): ``` Current: "This
research relies on several critical assumptions that, if
invalidated, would require scope adjustment..." ```
- **Analysis**: Good—this establishes both point (critical
assumptions exist) and issue (invalidity requires
adjustment) immediately. The paragraph then delivers on this
promise. This is a good model!

**Example 3** (Research Approach, lines 166-169): ```
Current: "While discrete system components will be
synthesized with correctness guarantees, they represent only
half of the complete system." ```
- **Analysis**: Good issue statement (discrete alone
insufficient), but could be sharper about the point. What
will this section show?
- **Fixed**: "While discrete system components will be
synthesized with correctness guarantees, they represent only
half of the complete system. This section describes how we
will develop continuous control modes, verify their
correctness, and address the unique verification challenges
at the discrete-continuous interface."

**Similar instances**:
- State of the Art lines 13-34: long paragraph with delayed
point
- Goals lines 103-119: impact paragraph could be tighter
- Approach lines 178-208: three-mode classification needs
clearer framing

**How to fix**:
1. First sentence should state the paragraph's point
2. Second sentence (or same sentence) should state why this
matters
3. Remaining sentences provide supporting detail

---

### Pattern 4: Topic Position & Continuity

**Principle** (Gopen): The topic position (beginning of
sentence) should contain old/familiar information that links
to what came before. This creates flow and coherence. Abrupt
topic shifts disorient readers.

**Example 1** (Goals, lines 18-23): ``` Sentence 1: "...this
reliance on human operators prevents the introduction of
autonomous control capabilities..."

Sentence 2: "Emerging technologies like small modular
reactors face significantly higher per-megawatt staffing
costs..." ```
- **Issue**: Topic shifts abruptly from "reliance on
operators" to "emerging technologies". Connection exists
(both about staffing challenges) but isn't explicit
- **Fixed**: "...prevents autonomous control capabilities.
This limitation creates particular challenges for emerging
technologies like small modular reactors, which face
significantly higher per-megawatt staffing costs..."

**Example 2** (State of the Art, lines 234-243): ```
Sentence about what HARDENS addressed: "...discrete digital
control logic..."

Next sentence: "However, the project did not address
continuous dynamics..." ```
- **Analysis**: Good use of "however, the project" in topic
position—maintains focus on HARDENS while pivoting to
limitation. This is a good model!

**Example 3** (Research Approach, lines 56-58): ``` Sentence
1: "...we may be able to translate them into logical
formulae..."

Sentence 2: "Linear Temporal Logic (LTL) provides four
fundamental operators..." ```
- **Issue**: Abrupt topic shift from "translating
procedures" to "LTL provides". Missing: why LTL? Why now?
- **Fixed**: "...translate them into logical formulae. To
formalize these procedures, we will use Linear Temporal
Logic (LTL), which provides four fundamental operators..."

**Similar instances**:
- Goals lines 23-27: "emerging technologies" → "what is
needed"
- State of the Art lines 72-74: control modes → division
between automated/human
- Approach lines 183-185: stabilizing mode example →
transitory mode definition

**How to fix**:
1. Identify the topic of the previous sentence
2. Begin next sentence with something related to that topic
3. Use transitional phrases when shifting topics: "This
[previous thing] leads to [new thing]"

---

### Pattern 5: Long Complex Sentences

**Principle**: Sentences with multiple subordinate clauses
(especially over 35-40 words) tax reader working memory.
Breaking into multiple sentences often improves clarity
without losing sophistication.

**Example 1** (State of the Art, lines 48-51): ``` Current
(51 words): "Procedures undergo technical evaluation,
simulator validation testing, and biennial review as part of
operator requalification under 10 CFR 55.59, but despite
these rigorous development processes, procedures
fundamentally lack formal verification of key safety
properties." ```
- **Issue**: Long sentence with list, subordinate clause,
and contrast—hard to parse
- **Fixed (2 sentences)**: "Procedures undergo technical
evaluation, simulator validation testing, and biennial
review as part of operator requalification under 10 CFR
55.59. Despite these rigorous development processes,
procedures fundamentally lack formal verification of key
safety properties."

**Example 2** (Risks, lines 72-78): ``` Current (57 words):
"Temporal logic operates on boolean predicates, while
continuous control requires reasoning about differential
equations and reachable sets, and guard conditions that
require complex nonlinear predicates may resist boolean
abstraction, making synthesis intractable." ```
- **Issue**: Run-on with multiple clauses strung together
with commas
- **Fixed (3 sentences)**: "Temporal logic operates on
boolean predicates, while continuous control requires
reasoning about differential equations and reachable sets.
Guard conditions requiring complex nonlinear predicates may
resist boolean abstraction. This mismatch could make
synthesis intractable."

**Similar instances**:
- State of the Art lines 44-51: procedure development
description
- Research Approach lines 40-45: hybrid system description
- Risks lines 17-24: computational tractability discussion
- Broader Impacts lines 13-23: economic analysis

**How to fix**:
1. Identify natural breakpoints (usually where you have
"and" or "but")
2. Create new sentences at these breaks
3. Ensure each new sentence has clear topic position
4. May need to repeat/reference previous sentence's subject
for clarity

---

## Section-Level Issues

### Goals and Outcomes Section **Strengths**: Excellent
structure with clear goal → problem → approach → outcomes →
impact progression. The four-paragraph opening is very
strong.

**Issues**:
- Lines 29-53 (Approach paragraph): This is dense and tries
to cover too much. Consider breaking into two paragraphs:
one on the approach concept, one on the hypothesis and
rationale.
- Outcomes enumeration: Very clear, but could strengthen the
transition from strategy to outcome in each item. Currently
reads as "we'll do X. [new sentence] This enables Y."
Consider: "We'll do X, enabling Y."

### State of the Art Section **Strengths**: Comprehensive,
well-researched, excellent use of the HARDENS case study as
both positive example and gap identifier.

**Issues**:
- **Length**: At 358 lines, this risks losing readers. Most
concerning: readers may forget your framing by the time they
reach your contribution.
- **Organization**: Four major subsections (procedures,
human factors, HARDENS, research imperative) would benefit
from a roadmap sentence at the beginning: "To understand the
need for hybrid control synthesis, we first examine..."
- **Balance**: HARDENS subsection is 89 lines—nearly 25% of
SOTA. While impressive, consider whether this should be a
separate section or whether some detail could move to an
appendix.
- **Transition to Approach**: The "Research Imperative"
subsection is excellent but feels like it belongs at the
start of Research Approach rather than end of SOTA.

### Research Approach Section **Strengths**: Clear
three-thrust structure, good use of equations and examples,
strong technical detail.

**Issues**:
- **Subsection transitions**: The transitions between the
three main subsections (Procedures→Temporal,
Temporal→Discrete, Discrete→Continuous) could be smoother.
Each starts somewhat abruptly.
- **SmAHTR introduction**: The SmAHTR demonstration case is
introduced suddenly at line 253. Consider introducing it
earlier (perhaps in Goals section or at start of Approach)
so readers know it's coming.
- **Three-mode classification**: Lines 178-208 present the
stabilizing/transitory/expulsory framework, which is
innovative. This deserves more prominence—consider
highlighting it as a key contribution.

### Metrics of Success Section **Strengths**: TRL framework
is well-justified, progression through levels is clear.

**Issues**:
- **Defensive tone**: Lines 11-30 spend considerable space
justifying why TRL is appropriate. This is good but could be
more concise. Consider: one paragraph on why TRLs (lines
10-19) rather than two.
- **Grading criteria**: The TRL definitions (3, 4, 5) are
excellent. Very concrete and measurable.

### Risks and Contingencies Section **Strengths**:
Comprehensive, each risk has indicators and contingencies,
well-organized.

**Issues**:
- **Subsection balance**: Four subsections range from 41
lines (computational) to 65 lines (discrete-continuous).
Ensure space reflects actual risk level.
- **Mitigation vs. contingency**: Some subsections blur
"mitigation" (preventing problems) and "contingency"
(response if they occur). Consider clarifying this
structure.

### Broader Impacts Section **Strengths**: Clear economic
motivation, good connection to SMRs and datacenter
application.

**Issues**:
- **Brevity**: At 75 lines, this is the shortest technical
section. Given that economic viability is a key motivation,
consider expanding.
- **Missed opportunities**: Could briefly mention
workforce/educational impacts (training future engineers in
formal methods), equity (providing reliable clean energy to
underserved areas), broader applicability beyond nuclear.

### Budget Section **Brief review**: Budget is
comprehensive, well-justified, appropriate. Minor note:
Consider whether the high-performance workstation (Year 1)
might need upgrades in Year 2-3 as synthesis scales up.

### Schedule Section **Brief review**: Schedule is ambitious
but realistic. Six trimesters for dissertation research is
reasonable. Publication strategy is smart (nuclear community
first, then broader control theory community). Minor note:
Line 73 has a space issue ("t ranslation").

---

## Big Picture Observations

### Narrative and Argument Structure

**Strengths**:
- Clear problem-solution arc: operators make errors →
procedures lack formal guarantees → hybrid control synthesis
provides guarantees
- Good use of motivating examples (TMI, human error
statistics, HARDENS)
- Technical progression is logical: discrete synthesis →
continuous verification → integrated system

**Opportunities**:
1. **Strengthen "so what" transitions**: The proposal
sometimes presents information without explicitly stating
significance. Add more "This matters because..." statements.
2. **Emphasize novelty earlier**: The three-mode
classification and discrete-continuous interface
verification are novel contributions. Signal this earlier
and more explicitly.
3. **Create more callbacks**: When describing Research
Approach, refer back to specific limitations identified in
State of the Art. Currently these connections are implicit.

### Rhetorical Effectiveness

**Credibility established through**:
- Comprehensive literature review
- Specific technical detail
- Access to industry hardware (Emerson partnership)
- Prior conference recognition (best student paper)

**Value proposition**:
- Clear economic impact (O&M cost reduction)
- Safety improvement (mathematical guarantees vs. human
operators)
- Broader applicability (methodology generalizes)

**Could strengthen**:
- More explicit statements of what's novel vs. what's
established practice
- Stronger emphasis on the unique combination of discrete
synthesis + continuous verification (others do one or the
other, not both)

### Content Gaps and Consistency

**Terminology**:
- Generally consistent
- Good introduction of technical terms (hybrid automata,
temporal logic, reachability analysis)
- Minor: "correct by construction" vs. "provably
correct"—used interchangeably, which is fine, but could note
they're synonymous

**Scope consistency**:
- Excellent—stays focused on startup procedures for SmAHTR
- Appropriately acknowledges limitations (TRL 5, not
deployment-ready)
- Risk section addresses what happens if scope must narrow

**Potential gaps**:
1. **Cybersecurity**: Not mentioned. For autonomous nuclear
control, shouldn't there be at least a paragraph on security
verification?
2. **Regulatory path**: You mention "regulatory
requirements" but don't detail what NRC approval process
would look like. Even a paragraph would strengthen
credibility.
3. **Comparison with alternatives**: What about machine
learning approaches to autonomous control? Worth a paragraph
explaining why formal methods are superior for
safety-critical systems.

---

## Gopen Framework Quick Reference

**Stress Position**: End of sentence should contain most
important new information. Readers expect climax there.

**Topic Position**: Beginning of sentence should contain
familiar information that links to previous sentence.
Creates flow.

**Point-Issue Structure**: Paragraphs should open by stating
(1) the point/claim and (2) why it matters, before providing
supporting detail.

**Topic String**: The chain of topics across sentences in a
paragraph. Strong topic strings create coherence; broken
ones confuse readers.

**Old→New Information Flow**: Information should flow from
familiar (old) to unfamiliar (new) within sentences and
paragraphs.

---

## Next Steps

1. **Start with Priority Issues 1-3**: These have the
highest impact
2. **Apply Patterns**: Use the pattern examples to fix
similar instances throughout
3. **Consult Detailed Document**: For comprehensive
checkbox-by-checkbox revisions
4. **Section-by-section revision**: Work through one section
at a time, applying patterns
5. **Final pass for consistency**: Ensure changes maintain
consistent terminology and tone

This proposal has strong technical content and a solid
structure. The revisions suggested here will strengthen
clarity, emphasize key contributions, and make the argument
even more compelling for reviewers. Good luck with your
revisions!