openclaw - 💡(How to fix) Fix [Feature]: Dreaming: add wildcard candidates to deep phase ranking [1 participants]

Official PRs (…)
ON THIS PAGE

Recommended Tools

×6

Utilities matched from this issue’s tags and category — try them while you read without losing context.

GitHub issue graph ai analysis

Paste a GitHub issue URL. We fetch that issue, discover linked issues from bodies/comments/timeline, collect linked pull requests, and produce a structured English report.

The report is written in English Markdown for sharing and archival.

Helpful · Quick feedback

Loading…
GitHub stats
openclaw/openclaw#62029Fetched 2026-04-08 03:10:05
View on GitHub
Comments
0
Participants
1
Timeline
1
Reactions
0
Author
Participants
Timeline (top)
labeled ×1

Add random sub-threshold wildcard candidates to dreaming deep phase ranking so that unexpected memories can occasionally surface for promotion

Root Cause

Add random sub-threshold wildcard candidates to dreaming deep phase ranking so that unexpected memories can occasionally surface for promotion

RAW_BUFFERClick to expand / collapse

Summary

Add random sub-threshold wildcard candidates to dreaming deep phase ranking so that unexpected memories can occasionally surface for promotion

Problem to solve

The current dreaming system only promotes memories that score well on a convergence-oriented model — high recall frequency, strong relevance, multi-day consolidation. This means long-term memory becomes an echo chamber of things the user already repeatedly asks about. Memories that were recalled a few times in passing but touched on something novel or cross-cutting get permanently filtered out. Users lose serendipitous rediscovery. The thing that makes real memory (and real dreaming) useful isn't just reinforcing what you already know, it's surfacing forgotten connections you didn't know you needed.

Proposed solution

Implementation on branch feat/dreaming-wildcard-candidates — adds a wildcardRatio option to rankShortTermPromotionCandidates() (default 0.15). Sub-threshold candidates that passed hard gates but failed the composite score are randomly sampled into the final result set. ~20 lines in production code, fully backwards-compatible, set to 0 to disable.

Alternatives considered

No response

Impact

This is a low-severity, long-term quality concern rather than something blocking anyone:

Affected: Any user with dreaming enabled. The effect is invisible — it's memories they never see rather than something breaking. Severity: Minor. The system works correctly, it's just conservative. Long-term memory slowly becomes a filtered echo of what the user already asks about frequently, missing serendipitous connections.
Frequency: Every single deep phase run. The minScore gate silently drops 100% of sub-threshold candidates every time — there's no mechanism for surprises. Consequence: No errors or manual work. The practical cost is opportunity cost — the user's long-term memory converges toward a narrow band of "proven useful" topics and never surfaces the odd, one-off memory that might connect two unrelated projects or remind them of something they forgot they cared about.

Evidence/examples

No response

Additional information

No response

extent analysis

TL;DR

Implement the proposed solution on branch feat/dreaming-wildcard-candidates to introduce random sub-threshold wildcard candidates into the dreaming deep phase ranking.

Guidance

  • Review the implementation on branch feat/dreaming-wildcard-candidates to understand how the wildcardRatio option is added to rankShortTermPromotionCandidates().
  • Test the effect of the wildcardRatio option on the promotion of sub-threshold candidates, starting with the default value of 0.15.
  • Evaluate the impact of this change on the diversity of memories surfaced during deep phase runs, considering the trade-off between serendipitous rediscovery and the potential for irrelevant memories to be promoted.
  • Monitor user feedback and adjust the wildcardRatio as needed to balance the goals of promoting novel connections and avoiding noise in the memory recall process.

Example

No specific code example is provided, as the implementation details are assumed to be available on the feat/dreaming-wildcard-candidates branch.

Notes

The proposed solution aims to address a minor, long-term quality concern rather than a critical issue, and its impact may be subtle. The effectiveness of this change in promoting serendipitous connections without introducing too much noise will depend on the specific characteristics of the user base and their interaction with the dreaming system.

Recommendation

Apply the workaround by implementing the proposed solution on branch feat/dreaming-wildcard-candidates, as it offers a controlled way to introduce diversity into the memory promotion process without disrupting the existing functionality.

Vote matrix · Quick signals

Works
Did the solution work? Tap to confirm.
Easy Fix
Was it a quick fix?
Time Saver
Did it save you time?
Blocking
Was it severely blocking?
Common Issue
Are others likely hitting this too?
Flaky / Intermittent
Is it intermittent?
Verified / Reproducible
Can you reproduce it reliably?
Loading…

Still need to ship something?

×6

Another batch ranked right after the header list — different links, same matching logic.

Back to top recommendations

TRENDING

openclaw - 💡(How to fix) Fix [Feature]: Dreaming: add wildcard candidates to deep phase ranking [1 participants]