INQUIRING LINE

How do implicit signals like clicks capture preference more reliably than explicit ratings?

This explores why behavioral traces like clicks and watches often capture user preference better than star ratings — and the corpus suggests the real answer is that implicit signals carry an extra dimension ratings throw away, while also dragging in biases you have to model around.


This explores why behavioral traces like clicks and watches often capture user preference better than star ratings. The sharpest reframing in the corpus is that implicit and explicit signals aren't just noisier and cleaner versions of the same thing — they're shaped differently. Implicit feedback splits into *two* paired magnitudes: preference (did they engage or not) and confidence (how much, how often) Can implicit feedback reveal both preference and confidence?. A single star rating collapses both into one number and loses the certainty information. Watching a show ten times and watching it once both might map to 'liked it,' but the implicit signal keeps the volume knob that the rating discards.

The deeper reason explicit ratings underperform is that a rating isn't always a preference at all. Decomposing annotation responses shows they contain three different things — genuine preferences, non-attitudes (people answering when they don't actually have an opinion), and constructed preferences (opinions invented on the spot by the act of being asked) Do all annotation responses measure the same underlying thing?. Asking forces a number into existence even when no real preference exists, contaminating the signal. Clicks don't have this problem: nobody clicks to be polite or to fill in a survey box. The behavior happened because the interest was real, which is also why agents can often learn more by watching than by asking Can agents learn preferences by watching rather than asking?.

But 'more reliable' comes with a catch the corpus is blunt about: implicit signals are reliable about *what users did*, not cleanly about *what they wanted*, because the system itself shaped what they could do. You only click what you were shown. YouTube's ranking work makes this concrete — without explicitly modeling selection and position bias, recommenders converge on degenerate loops that amplify their own past decisions, mistaking 'we showed it at the top' for 'they preferred it' Why do ranking systems need to model selection bias explicitly?. So the reliability is real but conditional: you have to subtract out the artifacts of exposure.

There's also a modeling subtlety in *how* you treat click data once you trust it. Clicks behave like a competition, not independent yes/no judgments — a multinomial likelihood that forces items to compete for a fixed probability budget outperforms Gaussian or logistic treatments, because it bakes in the top-N ranking objective that implicit feedback is really about Why does multinomial likelihood work better for click prediction?. And the same accuracy-chasing that implicit signals enable can quietly crowd out a user's minority interests unless you rerank for calibration Why do accuracy-optimized recommenders crowd out minority interests?.

The thing you might not have expected to learn: the field is increasingly trying to recover the *richness of explicit feedback without the cost of asking for it*. Negative comments like 'doesn't look good for a date' can be transformed by an LLM into a positive, retrievable preference ('prefer more romantic') Can language models bridge the gap between critique and preference?, and scalar reward signals are being shown to throw away directional information that natural language feedback preserves Can scalar rewards capture all the information in agent feedback?. So the trajectory isn't 'implicit beats explicit' — it's that the most useful signal carries *both* preference and the surrounding context of how strong and which-direction it points, which clicks supply abundantly and ratings supply thinly.


Sources 8 notes

Can implicit feedback reveal both preference and confidence?

Hu, Koren, and Volinsky show that implicit signals (watches, purchases, clicks) encode preference and confidence as two distinct dimensions. Explicit ratings collapse these into one number, losing information about certainty in the preference estimate.

Do all annotation responses measure the same underlying thing?

Behavioral science reveals that annotations contain genuine preferences, non-attitudes, and constructed preferences—distinguishable by consistency across measurement conditions. Treating them uniformly contaminates reward model training and downstream alignment.

Can agents learn preferences by watching rather than asking?

M3-Agent demonstrates that separating episodic events from semantic knowledge in an entity-centric graph, combined with parallel memorization and control processes, allows agents to infer and act on user preferences without asking. This architecture mirrors human cognitive systems that bind disparate information about individuals across sensory modalities.

Why do ranking systems need to model selection bias explicitly?

YouTube's multi-objective ranker uses MMoE for conflicting objectives and a shallow position tower to remove selection bias from training data. Without both mechanisms, models converge on degenerate equilibria that amplify their own past decisions.

Why does multinomial likelihood work better for click prediction?

Multinomial likelihood better models click data because it forces items to compete for a fixed probability budget, implicitly optimizing for top-N ranking. Gaussian and logistic likelihoods allow high probability across many items simultaneously, misaligning training with ranking objectives.

Why do accuracy-optimized recommenders crowd out minority interests?

Accuracy-optimized models systematically miscalibrate by over-weighting dominant user interests. A post-processing reranking algorithm that enforces calibration constraints can restore proportional representation without retraining the underlying model.

Can language models bridge the gap between critique and preference?

Few-shot LLM prompting can convert natural negative feedback like "doesn't look good for a date" into positive preferences like "prefer more romantic," enabling retrieval systems to find better-matching recommendations without fine-tuning.

Can scalar rewards capture all the information in agent feedback?

Natural feedback carries two orthogonal types of information: evaluative (how well an action performed) and directive (how it should change). Scalar rewards capture evaluation but discard directional specifics that token-level distillation can recover, making the two complementary rather than redundant.

Research prompt for your LLMexpand ↓

Copy into ChatGPT or Claude to take this line of inquiry further — it asks the model to find newer work and re-test which earlier constraints still hold.

You are a research analyst re-testing claims about implicit vs. explicit preference signals in recommender systems and LLM personalization. The question remains open: do behavioral traces (clicks, watches) capture user preference more reliably than explicit ratings, and if so, why?

What a curated library found — and when (findings span 2017–2026; treat as dated claims, not current truth):
• Implicit feedback splits into two paired magnitudes — preference (engagement: yes/no) and confidence (volume, frequency) — while star ratings collapse both into one number, discarding certainty (2017–2021).
• Explicit ratings contain three contaminating signals: genuine preferences, non-attitudes (answers to forced questions), and constructed preferences (invented on the spot), whereas clicks are behaviorally genuine (~2021).
• Without explicitly modeling selection and position bias, recommenders mistake 'we showed it at the top' for 'they preferred it,' creating degenerate loops (2023–2024).
• Multinomial likelihoods outperform Gaussian/logistic models for click data because they bake in the top-N ranking objective implicit feedback is about (~2024).
• LLM-based critique-to-preference transformation and reward signal decomposition (evaluative vs. directive) are recovering richness of explicit feedback without asking users (~2025–2026).

Anchor papers (verify; mind their dates):
• arXiv:1708.05031 (2017) — Neural Collaborative Filtering
• arXiv:2307.15142 (2023) — Reconciling accuracy-diversity trade-off
• arXiv:2604.03238 (2026) — Measuring Human Preferences in RLHF as Social Science
• arXiv:2507.04607 (2025) — PRIME: LLM Personalization with Cognitive Memory

Your task:
(1) RE-TEST EACH CONSTRAINT. For every finding, judge whether newer models (multimodal LLMs, agentic systems), orchestration (multi-turn dialogue, memory synthesis), or evaluation methods (online A/B, human studies) have since relaxed or overturned it. Separate the durable question (implicit signals capture richer preference structure) from perishable limitations (e.g., 'you can't extract preference from clicks without debiasing'—has this been automated?). Flag what resolved it.
(2) Surface the strongest CONTRADICTING or SUPERSEDING work from the last ~6 months. Has any recent paper shown explicit signals (natural language, Likert scales with confidence, ranker feedback) outperform clicks at scale, or shown clicks fail under new distribution shifts (synthetic data, adversarial users, cross-domain transfer)?
(3) Propose 2 research questions that ASSUME the regime has moved: (a) given LLMs can now synthesize preference from natural language artifacts (critiques, reviews, conversation context), does the implicit–explicit boundary still matter, or does it dissolve into "all signals are noisy reconstructions of latent preference structure"? (b) in multi-agent or long-horizon settings, do click traces and explicit feedback diverge in ways that single-turn recommenders don't reveal?

Cite arXiv IDs; flag anything you cannot ground in a real paper.

Next inquiring lines