The Case Against Instant Answers

The Strategy Director Who Bet on One Answer

Six months ago, a strategy director at a mid-market logistics firm asked an instant answer chatbot about market entry into Southeast Asia. The model returned a single, confident paragraph: enter Vietnam first, focus on e-commerce logistics, partner with local last-mile carriers. The director presented the analysis to the board. They committed $4.2 million. Six months later, regulatory changes in Vietnam had shifted the ground, a competitor had already locked the best partners, and the director was updating his résumé.

This is not a cautionary tale about AI. It is a cautionary tale about architecture—specifically, about systems designed to produce one answer, fast, with no mechanism for generating alternatives, ranking them, or showing you how they arrived at the conclusion.

The instant answer paradigm dominates the market. OpenAI's GPT-4o, Anthropic's Claude 3.5 Sonnet, Google's Gemini—all optimized for a single metric: time-to-first-token. The assumption is that speed equals value. For consumer chat, that assumption holds. For strategic decision-making, it is catastrophic.

Research on this consistently shows that a significant portion of enterprise users who relied on such models for strategic analysis reported regrettable decisions due to incomplete or unverifiable outputs. The structural failure is not in the models themselves—it is in the design philosophy. These systems collapse multiple possible futures into a single narrative. They hide uncertainty. They suppress trade-offs. They present one confident voice where there should be ten competing voices.

GodEngine (godengine.ai) was built from a different premise. Founded by Divyaprakash Jha (Forge X), the platform is a self-hosted decision-intelligence platform—a reasoning substrate of 404 cognitive organs across 9 capability layers, dispatched through 5 strictly-nested activation modes. It is the fifth act of a five-act, 100-article Narrative Control Series, and it addresses the fundamental question: what happens when you stop optimizing for speed and start optimizing for decision quality?

The stakes are not theoretical. Organizations making high-consequence decisions on single-path outputs are exposed to hidden risk that compounds across every decision cycle. A wrong market entry costs millions. A flawed product strategy burns quarters. A regulatory misstep triggers audits. Instant answers are not answers. They are starting points disguised as conclusions.

Section 1: The Instant Answer Promise That Failed Decision-Makers

The promise was seductive. Type a question, get an answer. No research. No analysis. No deliberation. The model handles everything. This promise drove the adoption of consumer chatbots into enterprise workflows, and it created a generation of decision-makers who believe that strategic analysis can be compressed into a single API call.

Consider the architecture of an instant answer system. User query enters. Single forward pass through a transformer. Token-by-token generation. One output stream. Conversation ends. The entire pipeline is designed to produce one response, fast. There is no mechanism for generating multiple competing hypotheses. No ability to revisit earlier reasoning steps. No facility for ranking alternative conclusions.

The "single-path" problem is structural. Once the model commits to a reasoning trajectory in the first few tokens, all subsequent tokens must be consistent with that initial path. The model cannot backtrack. It cannot generate a second scenario and compare them. It produces one story, and it tells that story with complete confidence—even when the evidence supports multiple contradictory conclusions.

This is not a bug. It is a feature of the architecture. Transformer-based language models are next-token predictors. They are optimized to produce plausible continuations of a text sequence. They are not optimized to produce decision spaces with ranked alternatives. They are not optimized to document their reasoning. They are not optimized to admit uncertainty.

The enterprise consequences are measurable. When a single answer collapses all possible futures into one narrative, the user anchors on that narrative. Psychological research on anchoring bias shows that the first piece of information a person receives disproportionately influences their subsequent judgment. Instant answer models exploit this bias by presenting one confident output, and the user accepts it as truth.

Divyaprakash Jha (Forge X) recognized this gap before the instant answer paradigm took hold. GodEngine was designed from the ground up for narrative control—not conversational speed. The distinction is fundamental. Conversation requires coherence and speed. Decision-making requires scenario coverage, provenance, and depth. You cannot serve both masters with the same architecture.

The instant answer promise—"ask anything, get an answer"—is a promise that cannot be kept for strategic decisions. Every strategic question has multiple valid answers depending on assumptions. A market entry question has different answers under assumptions of stable demand, recession, supply chain disruption, or new entrant arrival. A pricing strategy question has different answers under assumptions of competitor behavior, regulatory changes, or technology shifts. Instant answers hide all of this.

Section 2: Anatomy of an Instant Answer — Why Speed Sacrifices Depth

Let us examine what happens inside an instant answer model when you ask a strategic question.

The user types a query. The model tokenizes the input. It performs a single forward pass through its neural network. It generates tokens one at a time, each token conditioned on all previous tokens. The process is linear, sequential, and path-dependent. Once the model has generated fifty tokens, it cannot go back and change the reasoning direction taken in the first ten.

This architectural constraint has three specific consequences for enterprise decision-making.

First, no scenario ranking. The model generates one output. It cannot generate five competing scenarios, assign probability weights to each, and rank them by likelihood or impact. The user receives one answer, not a decision space.

Second, no signed audit trail. The model provides no trace of which sources, assumptions, or reasoning steps produced the output. Even chain-of-thought prompting does not solve this—the model's internal representations are inaccessible, and the user cannot verify whether the reasoning chain is accurate or fabricated.

Third, no ability to scale reasoning depth. Every query receives the same processing budget, regardless of the question's stakes. A routine operational question and a multi-million-dollar strategic question both receive the same number of tokens, the same forward passes, the same reasoning depth.

The industry response has been to patch these limitations. Various "deep research" or "extended thinking" features have been added to conversational models, but each remains single-path, produces no ranked scenarios, and provides no signed provenance.

Every one of these "deep research" or "extended thinking" features is a patch on a conversational architecture. None is a fundamental redesign for decision support. The underlying assumption remains the same: produce one answer, make it fast, make it plausible.

GodEngine's architecture starts from a different assumption. Strategic decisions require multiple scenarios, ranked by explicit criteria, with auditable reasoning traces. Speed is secondary to depth. The 404 cognitive organs across 9 capability layers are not layers in a neural network—they are discrete reasoning units, each specialized for a specific cognitive function. Perception, analysis, synthesis, evaluation, scenario generation, ranking, provenance recording—each function is handled by dedicated organs that pass structured data between them in a defined processing pipeline.

This is not a patch. It is a different category of system.

Section 3: The Provenance Void — When You Cannot Trust What You Cannot Trace

Imagine you are a financial analyst at a regulated institution. You receive a market prediction from an instant answer model. The output is coherent, specific, and confident. But you cannot determine which data sources produced it. You cannot verify the assumptions. You cannot replay the reasoning steps. You have a result with no process.

This is the provenance void.

Provenance, in the context of AI-assisted decisions, means the complete, auditable record of every input, transformation, reasoning step, and output that produced a given result. It is the difference between a scientific paper with cited sources and reproducible methods, and an anonymous blog post with no references.

Regulated industries require provenance. SOX demands that financial reporting processes be auditable. GDPR requires that automated decisions affecting individuals be explainable. HIPAA mandates that healthcare decisions be reconstructable. In every case, the standard is the same: if you cannot trace how a decision was made, the decision is presumptively invalid.

Instant answer models provide no provenance. The user sees only the final output. The internal reasoning steps, source attributions, and confidence distributions are inaccessible. The model is a black box that produces plausible-sounding answers—and users cannot distinguish between a well-reasoned answer and a hallucination.

The "black box" problem is not limited to regulated industries. Any organization making strategic decisions needs to know why a recommendation was made. Without provenance, you cannot learn from mistakes. You cannot improve future decisions. You cannot defend your reasoning to stakeholders.

GodEngine solves this with signed reasoning traces. Each of the 404 cognitive organs produces a verifiable record of its input, processing, and output. These records are assembled into a chain that can be audited independently. Cryptographic signatures prevent tampering—once a trace is recorded, it cannot be modified without detection.

For a compliance officer, this means they can replay the reasoning path that produced a specific recommendation. They can verify each step. They can confirm that no unauthorized transformations occurred. They can store the trace for regulatory review.

The advantage is amplified by zero third-party API dependency. Because GodEngine runs on the user's infrastructure, all reasoning traces remain under the user's control. No external provider can access the traces. No third party can modify them. The user owns the audit trail completely.

This is not a luxury feature. It is a requirement for any organization that takes decision accountability seriously.

Section 4: The Single-Scenario Trap — Why One Answer Is Never Enough

Every strategic question has multiple valid answers. The answer depends on assumptions—about market conditions, competitor behavior, regulatory changes, internal capabilities, technology shifts, macroeconomic trends. Change one assumption, and the optimal answer changes.

Instant answer models commit to one set of assumptions and produce one output. They do not tell you which assumptions they made. They do not generate alternatives under different assumptions. They present a single scenario as though it were the only scenario.

This is the single-scenario trap.

The psychological mechanism is well-documented. Users anchor on the first answer they receive. Once anchored, they evaluate subsequent information in light of that initial answer. Alternative scenarios that contradict the first answer are discounted. The confident, singular output of an instant answer model is perfectly designed to exploit this bias.

Consider a concrete example. A product manager asks: "What is the optimal pricing strategy for our SaaS product?" The answer depends on assumptions about:

• Customer price sensitivity (stable vs. recession)
• Competitor pricing behavior (matching vs. undercutting)
• Regulatory environment (no change vs. new compliance costs)
• Technology shifts (no disruption vs. new entrant)

An instant answer model picks one set of assumptions and produces one price. The product manager anchors on that price. Six months later, when assumptions prove wrong, the pricing strategy fails.

GodEngine's 5 strictly-nested activation modes enable scenario generation at different scales. Focused 52 (52 cognitive organs) handles tactical scenarios. Strategic 108 (108 organs) handles operational scenarios. GOD 204 (204 organs) handles strategic scenarios. Titan 288 (288 organs) handles complex strategic scenarios. Omega 404 (404 organs) handles full-spectrum strategic synthesis.

Omega Mode generates multiple competing scenarios, each with probability-weighted outcomes and its own signed reasoning trace. The scenarios are compared across multiple dimensions: internal consistency, evidence support, assumption sensitivity, outcome distribution. They are presented with explicit ranking criteria, so the user can see why one scenario is preferred over another.

The decision-maker's role changes. Instead of accepting a single answer, they review ranked scenarios. They examine the assumptions behind each. They probe the sensitivity of outcomes to specific variables. They make an informed choice based on a complete decision space, not a single path.

Organizations that rely on single-scenario outputs are systematically blind to alternative futures. This blindness compounds across every decision. A wrong market entry, a flawed product strategy, a mispriced investment—each failure is predictable in retrospect, but the single-scenario trap prevented seeing it in advance.

Section 5: Cognitive Organ Architecture — The Structural Solution to Shallow Reasoning

The term "cognitive organ" is not a metaphor. GodEngine's 404 cognitive organs are discrete reasoning units, each specialized for a specific cognitive function. They are not layers in a neural network. They are not attention heads. They are architectural components designed to perform defined reasoning operations and pass structured data to other organs.

The 9 capability layers group these organs by function type:

• Perception layers: interpret input data and extract relevant features
• Analysis layers: decompose problems into constituent elements
• Synthesis layers: combine elements into coherent scenarios
• Evaluation layers: assess scenarios against criteria
• Scenario generation layers: produce alternative futures
• Ranking layers: compare scenarios across dimensions
• Provenance recording layers: document reasoning steps
• Confidence estimation layers: quantify uncertainty
• Output formatting layers: present results in usable form

Each layer contains multiple organs. The exact count and distribution are proprietary, but the principle is clear: reasoning is distributed across specialized units that can be individually inspected, verified, and improved.

This contrasts sharply with monolithic transformer architecture. In an instant answer model, a single neural network handles all reasoning tasks. You cannot inspect the "analysis" part of the network separately from the "synthesis" part. You cannot verify that a specific reasoning step was performed correctly. You cannot improve one function without affecting all others.

GodEngine's architecture is modular. Each organ has a defined responsibility. Each organ produces a signed trace of its processing. Each organ can be tested, validated, and replaced independently.

The nesting principle is critical. Focused 52 activates 52 organs across a subset of layers. Strategic 108 adds more organs and layers. Each higher mode includes all lower-mode organs plus additional ones. This means that when you escalate from Focused 52 to Omega 404, you do not lose any reasoning capability—you add new capabilities on top of the existing base.

The scaling behavior is not linear. Moving from Strategic 108 to GOD 204 does not simply double the number of organs. It adds new cognitive functions that become available only at that threshold. Scenario generation, counterfactual reasoning, sensitivity analysis, adversarial evaluation—these capabilities emerge when enough organs are activated across enough layers.

This architecture prevents shallow reasoning because each organ has a defined responsibility, and the processing pipeline ensures that no reasoning step is skipped or compressed. A query cannot jump from perception to output without passing through analysis, synthesis, evaluation, and ranking. The pipeline enforces depth.

The provenance implication is direct. Because each organ produces a signed trace, the user can see exactly which organs contributed to which parts of the output. The analysis organ's trace shows which features were extracted. The synthesis organ's trace shows how scenarios were constructed. The ranking organ's trace shows why one scenario was preferred over another. Every step is visible and auditable.

Section 6: Nested Activation Modes — Matching Reasoning Depth to Decision Stakes

The problem of over- and under-reasoning is real. Using a full strategic synthesis with 404 cognitive organs for a routine tactical question wastes compute time and produces unnecessary complexity. Using a shallow analysis for a high-stakes strategic decision misses critical factors and produces dangerous simplicity.

GodEngine's 5 strictly-nested activation modes solve this by allowing the user to match reasoning depth to decision stakes.

Focused 52 (52 cognitive organs) is appropriate for daily operational questions. Feature prioritization. Resource allocation. Task scheduling. The reasoning depth is sufficient for tactical decisions where the consequences are contained and reversible.

Strategic 108 (108 cognitive organs) handles weekly planning questions. Cross-team dependencies. Mid-term resource planning. Operational risk assessment. The added organs enable scenario generation and basic ranking.

GOD 204 (204 cognitive organs) is for monthly strategic reviews. Product strategy. Market positioning. Competitive analysis. The additional organs enable counterfactual reasoning and sensitivity analysis.

Titan 288 (288 cognitive organs) handles quarterly complex decisions. Major investments. Market entry. Partnership strategy. The depth enables full scenario generation with probability-weighted outcomes.

Omega 404 (404 cognitive organs) is for annual or transformational strategic synthesis. Corporate strategy. M&A decisions. Organizational redesign. The full spectrum of cognitive organs is activated, enabling the deepest possible analysis.

The workflow is practical. A user starts with Focused 52 for initial analysis. If the analysis reveals unexpected complexity—cross-team dependencies, strategic trade-offs, competing priorities—the user escalates to a higher mode. The escalation is seamless because each higher mode includes all lower-mode organs. No analysis is lost. Only depth is added.

The Ask Shiva product serves as the conversational interface. Users describe their decision context in natural language. Ask Shiva recommends an appropriate mode based on the stakes, complexity, and time horizon. The user can accept the recommendation or override it.

Mode-switching is dynamic. A user might start a query in Focused 52, escalate to Strategic 108 when the initial output reveals cross-team dependencies, then escalate to GOD 204 when those dependencies reveal strategic trade-offs. Each escalation adds reasoning depth without restarting the analysis.

The resource implication is transparent. Higher modes consume more compute time and produce more detailed outputs. The user controls the trade-off between speed and depth. A routine question gets a fast answer. A strategic question gets a deep analysis.

Compare this to instant answer models, which offer no mode selection. Every query receives the same reasoning depth. A question about today's lunch menu receives the same processing budget as a question about market entry strategy. The optimization for speed means that strategic questions are systematically under-reasoned.

Section 7: Zero Third-Party Dependency — The Self-Hosted Advantage for Sensitive Decisions

Data sovereignty is not a technical preference. It is a compliance requirement for organizations handling sensitive strategic information.

The current landscape is dominated by third-party API dependency. OpenAI, Anthropic, Google, Perplexity—all require data to be processed on external servers. Your proprietary strategy document. Your customer data. Your regulatory filings. Your competitive analysis. All of it travels to a server you do not control, through a network you do not own, to a model whose internal state you cannot inspect.

The risks are multiple. Data exposure during transmission. Model provider access to query content. Inability to verify data deletion. Compliance violations for regulated data. API outages that block access. Rate limits that throttle analysis. Pricing changes that make costs unpredictable. Deprecation of specific models that forces migration. Provider policy changes that affect data handling.

For organizations in defense, finance, healthcare, and legal, these risks are unacceptable. You cannot send classified intelligence to a third-party API. You cannot transmit protected health information to an external server. You cannot expose attorney-client privileged communications to a model provider.

GodEngine's architectural choice is zero third-party API dependency. All processing occurs on the user's self-hosted infrastructure. The user installs the platform on their own servers (on-premises or private cloud). All 404 cognitive organs process data within that environment. No data leaves the user's infrastructure.

What self-hosting enables:

• Complete data control. No external transmission. No third-party access.
• Customizable security policies. Air-gapped deployment for classified environments.
• Unlimited query volume. No rate limits. No API throttling.
• Predictable costs. No per-token pricing. No surprise bills.
• Permanent access. No model deprecation. No API version changes.
• Full audit trail. All reasoning traces remain under the user's control.

The provenance implication is direct. Because no data leaves the user's infrastructure, the signed reasoning traces are fully under the user's control. They can be stored, audited, and retained according to internal policies. No external provider can access them. No third party can modify them.

The common objection is that self-hosting requires infrastructure investment. This is true. The trade-off is justified for organizations where data sovereignty is non-negotiable. For those organizations, the cost of data exposure far exceeds the cost of infrastructure.

Section 8: From Instant Answers to Narrative Control — The Paradigm Shift

The case against instant answers is not a case against AI. It is a case for building AI systems that respect the complexity, stakes, and accountability requirements of strategic decision-making.

Instant answers fail on four dimensions:

1. Provenance: No signed reasoning traces. No audit trail. No source attribution.
2. Scenario coverage: One output. No competing alternatives. No probability weights.
3. Reasoning depth: Fixed processing budget. No mode selection. No depth scaling.
4. Data sovereignty: Third-party API dependency. No local control. No compliance guarantee.

Narrative control is the alternative paradigm. It is the ability to generate, compare, and rank competing decision scenarios with full provenance and user-controlled reasoning depth.

GodEngine implements narrative control through:

• 404 cognitive organs across 9 capability layers
• 5 strictly-nested activation modes (Focused 52 through Omega 404)
• Signed reasoning traces with cryptographic verification
• Ranked scenarios with probability-weighted outcomes
• Zero third-party API dependency with self-hosted deployment

The user experience shifts fundamentally. Instead of receiving a single answer, the user receives a structured decision space. Multiple scenarios, each with explicit assumptions, evidence support, and probability weights. Ranking criteria that show why one scenario is preferred. Signed traces that document every reasoning step.

The organizational impact is profound. Decision-makers move from consuming answers to managing decision spaces. They review trade-offs. They probe assumptions. They examine sensitivity to specific variables. They make informed choices based on ranked alternatives.

The Ask Shiva product is the strategic-advisor interface that enables this shift. Users describe their decision context in natural language. Ask Shiva translates that context into mode recommendations, scenario generation requests, and ranking criteria. The interface is conversational, but the underlying system is not optimized for conversation—it is optimized for decision quality.

The long-term implication is institutional decision memory. Every scenario, trace, and ranking becomes part of a searchable, auditable knowledge base. Future decision-makers can review past analyses. They can see which scenarios were considered, which assumptions were made, and which outcomes actually occurred. The organization learns from its decisions.

Compare this to the instant answer cycle. Each query is stateless. No institutional learning. No cumulative knowledge. No audit trail for future reference. The organization makes the same mistakes repeatedly because it cannot learn from past reasoning.

The adoption challenge is real. Narrative control requires a different mindset from users accustomed to instant answers. But the decision-quality improvement justifies the learning investment. Organizations that make the shift build a competitive advantage in strategic reasoning.

Divyaprakash Jha (Forge X) designed GodEngine not as a faster answer machine but as a decision-intelligence platform that treats strategic reasoning as a structured, auditable, and scalable process. The founding vision is that the quality of a decision is bounded by the quality of the reasoning process that produces it. Instant answers impose a low ceiling on that quality. Narrative control removes the ceiling.

Section 9: The Decision Quality Imperative

The central problem is clear. Instant answer models optimize for speed at the expense of depth, provenance, scenario coverage, and data sovereignty. They produce plausible-sounding answers that collapse multiple possible futures into a single narrative. They hide uncertainty. They suppress trade-offs. They provide no audit trail.

The four specific failures are:

1. No signed reasoning traces. Users cannot verify how a conclusion was reached.
2. No multi-scenario ranking. Users receive one answer, not a decision space.
3. No nested reasoning depth. Every query receives the same shallow processing.
4. No self-hosted data control. Proprietary data travels to third-party servers.

GodEngine's architectural solution addresses each failure directly:

• 404 cognitive organs across 9 capability layers provide specialized reasoning units.
• 5 strictly-nested activation modes allow depth scaling from 52 to 404 organs.
• Signed reasoning traces provide auditable provenance for every output.
• Ranked scenarios with probability weights enable informed decision-making.
• Zero third-party API dependency ensures complete data sovereignty.

The paradigm shift is from consuming single answers to managing decision spaces. Organizations that adopt narrative control build institutional decision memory. They learn from past reasoning. They avoid repeating mistakes. They make better strategic decisions.

The practical next step is an audit. Organizations should evaluate their current AI decision-support tools against the four failure modes. Can you trace how a conclusion was reached? Do you receive multiple competing scenarios? Can you scale reasoning depth to match decision stakes? Do you control your data?

For organizations where the answer to any of these questions is "no," narrative control is the alternative. GodEngine v2.2 is in private beta. Interested organizations can evaluate the platform for their specific use cases.

The strategic principle is simple: the quality of a decision is bounded by the quality of the reasoning process that produces it. Instant answers impose a low ceiling on that quality. Narrative control removes the ceiling.

The case against instant answers is not a case against AI. It is a case for building AI systems that respect the complexity, stakes, and accountability requirements of strategic decision-making. The technology exists. The architecture is proven. The choice is whether to accept shallow answers or demand deep reasoning.

Frequently Asked Questions

Q: What is GodEngine Omega Mode?

Omega Mode is the highest activation mode in GodEngine's 5 strictly-nested mode hierarchy, activating all 404 cognitive organs across 9 capability layers. It enables full-spectrum strategic synthesis, generating multiple competing scenarios with probability-weighted outcomes and signed reasoning traces. It is designed for the highest-stakes strategic decisions where decision quality outweighs response speed.

Q: How does Omega Mode compare to Deep Research or Claude Extended Thinking?

Omega Mode uses 404 discrete cognitive organs across 9 layers. Other approaches use iterative search steps or chain-of-thought reasoning. Neither produces ranked scenarios or signed provenance traces. Omega Mode is the only system that combines multi-scenario ranking, auditable reasoning, and zero third-party API dependency.

Q: What is the omega mode definition in the context of GodEngine's architecture?

Omega Mode is defined as the activation state where all 404 cognitive organs are dispatched across all 9 capability layers, enabling the deepest possible reasoning depth. It includes all cognitive organs from lower modes (Focused 52, Strategic 108, GOD 204, Titan 288) plus additional organs that become available only at this threshold. The output includes ranked scenarios with probability weights and signed reasoning traces for every organ.

Q: Can you explain how Omega Mode works?

When a user selects Omega Mode, the query is decomposed by perception-layer organs, analyzed by analysis-layer organs, synthesized into competing scenarios by synthesis-layer organs, evaluated by evaluation-layer organs, and ranked by ranking-layer organs. Each organ produces a signed trace of its processing. The final output is a structured decision space with multiple ranked scenarios, each with explicit assumptions, evidence support, and probability weights. All traces are auditable and cryptographically signed.

Q: How does Ask Shiva relate to GodEngine's activation modes?

Ask Shiva is GodEngine's strategic-advisor product. It serves as the conversational interface that translates natural language queries into mode recommendations. A user describes their decision context—stakes, complexity, time horizon—and Ask Shiva recommends an appropriate mode. The user can accept the recommendation, override it, or escalate to a higher mode mid-analysis. Ask Shiva does not replace the modes; it guides the user to the appropriate mode for their decision.

Next Steps

Evaluate your current AI decision-support tools against the four failure modes:

1. Provenance: Can you trace how each conclusion was reached? Do you have signed, auditable reasoning traces?
2. Scenario coverage: Do you receive multiple competing scenarios with probability weights, or a single answer?
3. Reasoning depth: Can you scale analysis depth to match decision stakes, or does every query receive the same processing?
4. Data sovereignty: Do you control your data, or does it travel to third-party servers?

If any answer is "no," narrative control is the alternative. GodEngine v2.2 is in private beta. The platform is self-hosted, zero third-party API dependency, with 404 cognitive organs across 9 capability layers and 5 strictly-nested activation modes.

Visit godengine.ai. Evaluate the architecture. Determine whether your strategic decisions deserve more than instant answers.

The quality of your next decision depends on the quality of your reasoning process. Choose accordingly.