PRODUCT UPDATES

What’s new in BluffKing.

Player-facing improvements and experience fixes, listed only after they are publicly shipped.

Latest updates
  1. Released
    H5BackendMarketing site

    48 updates in this release

    This release includes: Security bounty referral rewards; Release report transactions before website preparation; A worldwide community on the second tournament page; Hide deleted tournaments from public views; Registration rewards for friend-room invitations in Zha Jin Hua and Bull; Complete short deck loading diagnostics; and 42 more changes.

    1. Security bounty referral rewards

      Original solver bounties remain RMB 30,000 total; referrers earn an extra RMB 2027 for each challenge successfully solved by someone who registered through their invitation.

      • Earn RMB 6081 when referred users solve all three challenges, without reducing solver bounties.
      • Home, bounty page and share copy explain the rewards; personal invitations carry the referrer’s code into registration.
    2. Release report transactions before website preparation

      Read retained website keys instead of probing empty catalog entries, prepare statistics after commit, and record correlated transaction timings.

      • Page retained keys by catalog path to avoid empty per-entry probes
      • Preserve a consistent snapshot across feature and website statistics
      • Record window, transaction identity and read/commit timings without changing the 100 ms alert threshold
    3. A worldwide community on the second tournament page

      The tournament overview and second edition detail now feature a world player distribution map and an invitation to compete.

      • Compete with real Texas Hold’em players from around the world.
    4. Hide deleted tournaments from public views

      Tournaments deleted by staff are now hidden from the tournament landing page and public details.

      • Apply deletion filtering to listings, result announcements and honors while retaining results, rewards and audit records.
    5. Registration rewards for friend-room invitations in Zha Jin Hua and Bull

      Verified email sign-ups through eligible Zha Jin Hua and Bull friend rooms receive the existing mutual Max reward and host bean bonus. Public tables, solo practice and expired rooms do not qualify.

      • Retains one reward per new user and ten room referrals per host per rolling 30 days.
      • Reward processing remains separate from successful registration.
    6. Complete short deck loading diagnostics

      Capture original short deck import timings and probe observed failed dependencies separately.

      • Keep probe budgets and reload-loop guards; added diagnostics do not establish a historical fault fix.
    7. Simpler event and game descriptions

      Remove unnecessary affiliation and official-award disclaimers across the app and website while preserving play, entry and reward rules.

      • Participation bonuses describe eligibility and award conditions directly.
      • Simplify Chinese and English game help, lessons, website copy and historical release notes.
    8. Release workflow supports runtime-first builds

      Complete and verify backend builds before starting deployment timing, then recheck the live version before uploads.

      • Report preparation, deployment and total command duration separately, retaining post-deploy E2E.
      • Preserve runtime-first compiler settings and production release safeguards.
    9. Make public tables easier to join in the lobby

      Show human player counts, available seats and clearer join actions in the lobby to help solo players find a public table.

      • Prioritize occupied tables with open seats in the loaded list, with refresh and retry controls.
      • Encourage friends to leave two seats available for newcomers.
    10. Private tables default to Shark bots

      Private table setup now defaults to Shark bots while allowing manual opponent selection.

      • Unspecified bot seats are filled with Sharks.
    11. New poker sticker: Just teasing my friend

      A cheeky panda with X-ray goggles brings a little fun to the table.

      • Send it in chat, from your reaction menu or to another player.
    12. Keep page content and help in their proper places

      Remove the duplicate feedback FAQ, simplify About links, and place casual-game rules and records in the table menu.

      • Feedback focuses on sending feedback, support contacts and replies; FAQs remain available through Help.
      • Legacy app-information and FAQ pages link to their matching website topics; About no longer repeats a destination.
      • Call or Fold and Lucky Draw use the game menu for rules and records while retaining current controls and results.
    13. Shorter MVP celebration

      The A5 offsuit MVP impact now lasts 1.5 seconds, matching other reactions and reducing visual distraction.

      • Keep the card reveal, gold halo and MVP label while shortening both animation and cleanup together.
    14. MVP hand celebration

      Keep table reaction captions within the screen and celebrate the first tournament MVP's A5 offsuit hand.

      • Reaction captions stay within the viewport and wrap when needed.
      • The Hadi reaction features an A5 card reveal, gold halo and first tournament MVP label.
    15. Improve back navigation

      Returning to Mine restores its scroll position, and bottom-tab switches no longer pile up in back history.

      • Restore Mine's scroll position on return.
      • Avoid repeated back steps after switching between Lobby and Clubs.
    16. Close idle short-deck rooms promptly

      When an idle short-deck room is due for cleanup, immediately resolve the current hand using timeout rules and close without waiting for bot turns or dealing again.

      • Stop automatic new deals after ten minutes without human input, even when a player's socket remains connected.
      • Preserve the shared cleanup schedule; fresh human activity cancels pending closure.
    17. Move inactive guest cleanup to 04:00

      Daily cleanup of inactive guest and test accounts now runs at 04:00 Beijing/Singapore time.

      • Retain the seven-day inactivity rule and protection checks for active accounts and unfinished games.
    18. Improve email address validation

      Reject email addresses with a one-character top-level domain before sending and improve delivery failure diagnostics.

      • Report this address error before accepting a verification-code request.
      • Diagnostics retain only fixed error labels, without addresses, codes or raw provider responses.
    19. Reduce repeated reads during card settlement

      Reuse the settlement roster's bot information to avoid repeated database reads when recording card hands such as Dou Dizhu.

      • Preserve atomic balance, ledger and history writes and retry protection.
    20. Public lobby bean balance ranking

      Switch between bean profit and current balance rankings in the public lobby.

      • Current balances rank from highest to lowest, with up to 50 players per page.
    21. New article on trust and anti-cheat defenses

      A bilingual article uses a player conversation as its cover to explain dealing proofs, public source code, the bounty and their limits.

      • Original chat screenshot appears on the article, listing and social previews.
      • Clarifies the three RMB 10,000 challenges and links to verification and reporting.
    22. Additional conversation in the anti-cheat article

      Adds an original chat screenshot about software and bug fixes, with bilingual context and changelog links.

      • The responsive inline screenshot links to the original image.
    23. Zero-knowledge dealing proofs for Zha Jin Hua

      Zha Jin Hua reuses Hold’em’s encrypted shuffle and decryption proofs to verify visible cards locally while keeping unshown cards private.

      • Lock the deck before collecting player randomness, with reconnect recovery.
      • Verify or download history proofs, distinguishing fully witnessed hands from valid proofs without complete local evidence.
    24. Shared dealing proofs across card games

      Extend shared encrypted-deck, shuffle and selective-opening proofs to Sangong, Short Deck, Dou Dizhu, Guandan, Blackjack, Baccarat and Color Clash while retaining existing Hold’em, Zha Jin Hua and Niuniu verification.

      • Cover short decks, jokers, multiple decks and Color Clash reshuffles while keeping unshown cards and secret seeds private.
      • Baccarat commits its shoe before betting and mixes player randomness; browser verification distinguishes valid proofs from matching local commitment records.
    25. Fix leaving consecutive drills

      Leave a drill session with one back action, without retracing every question.

      • Next question reuses the current history entry; completion and next-question loading failures share the same exit behavior.
    26. Fix tournament registration ticker controls

      Fix the recent entries ticker staying paused after Play because of button focus or touch hover.

      • Pause and Play directly control scrolling while preserving reduced motion preferences.
    27. Register for multiple tournaments

      An upcoming tournament registration no longer blocks registration for another edition.

      • Players remain limited to one tournament in play at a time.
      • Per-edition fees, duplicate registration handling and withdrawal refunds remain unchanged.
    28. Clearer table markers and banners

      Match room label and number sizing, move blind markers beside avatars, trim pot backgrounds and center result banners.

      • SB and BB use each seat's dealer-marker position; heads-up keeps only D.
      • Pot amounts stay readable with tighter backgrounds; mobile result banners align with the table center.
    29. Bottom tab settings feedback

      Bottom tab settings now offer Done and Cancel with clear result messages.

      • Done applies the selection and shows an update message; cancelling or dismissing preserves the existing tabs.
    30. Snappier scratch cards and slots

      Scratch cards reveal with shorter strokes and immediate match highlights; slots stop faster with crisp reel feedback.

      • Wider scratch strokes, quicker reveals and inline ordinary payouts, with a result sheet for big rewards.
      • Slot reels stop in about 1.36 seconds with distinct stop and win feedback.
    31. Preserve request failure correlation

      Keep correlation fields for network failures and startup probes so the same request can be investigated.

      • Retain the sent trace identifier when a request fails before response headers arrive.
      • Preserve startup trace identifiers, preload flags and session states through collection and admin details.
    32. Attribute stalled page recovery

      Retain the last recovery navigation stage to distinguish waiting from a navigation already invoked.

      • Record bounded stages for foreground wait, asset repair, navigation wait, superseded or changed routes, and navigation calls.
      • Keep existing recovery decisions, attempt limits and timing.
    33. Player randomness across card games

      Reuse Hold’em and Zha Jin Hua commitment/reveal for player entropy and device verification in Niuniu, Sangong, Short Deck, Dou Dizhu, Blackjack, Guandan and Color Clash.

      • Lock decks and seats before contributions; exchange fresh entropy for each Guandan deal and Color Clash reshuffle.
      • Missing own contributions, withheld reveals and missing device evidence cannot report full verification; historical proofs remain compatible.
    34. Show account email in notification inbox alerts

      Admin notification inbox alerts now identify readers by their account email.

      • Readers and internal accounts show their email instead of a shortened user ID, with an explicit fallback when no email is linked.
    35. Zero-knowledge card verification for Niuniu

      Niuniu reuses Hold’em’s encrypted shuffle and opening proofs, with completed-hand verification and downloads in the table menu.

      • The browser checks five cards per seat against its saved commitment; undealt cards stay hidden.
      • Uses dealer randomness, without player seed mixing or server-blind dealing.
    36. Mobile notifications and tournament reach

      Add opt-in native notifications and improve tournament entry and reach reporting. Push requires provider setup and the updated app.

      • Enable or disable system notifications from the inbox
      • Keep an open tournament visible on home with manual promotion switching
      • Separate queued, provider-accepted, opened and subsequent registration metrics
    37. Quieter poker nickname styling

      Your nickname stays above your avatar with smaller, lighter text to keep attention on cards and chips.

      • Nickname text scales with table width within readable size limits.
    38. Correct Guandan waiting observations

      Distinguish waiting for the host from active deal preparation so normal waits are not recorded as timeouts.

      • Do not start a deal timeout while waiting for the host or assigned event players
      • Retain observations during active deal preparation and preserve historical records
    39. Clearer Guandan card selection

      Selected cards use a blue border and checkmark to distinguish them from gold bomb hints.

      • Selected groups also turn blue, and selection styling takes priority over bomb hints.
    40. Fix guaranteed catches in the claw machine

      All toys now share one catch probability after a valid aim; retries and reloads retain the same paid result.

      • Empty aims still miss; eligible aims can also fail
      • Updated rules and miss messages
    41. Retain request evidence for startup timeouts

      Startup probe timeouts retain the exact request phase and correlation identifiers for diagnosis.

      • Early session probes carry a random correlation identifier.
      • The four-second deadline and background recovery are preserved; this does not establish a fix for historical timeouts.
    42. Bean membership rewards, saving goals and permanent collectibles

      Redeem 1,000,000 beans for 30 days of Max once per account per Singapore calendar month. Collect permanent profile badges, save a goal across devices and view redemption receipts. Shortfalls link directly to player groups for product testers. Initial grants become 2,000, daily claims 200 and accepted feedback rewards 500 and referral bonuses 5,000; new tournament defaults are lower. Existing balances and published tournament prizes are preserved.

      • Atomic bean debit, membership grant and monthly limit with retry-safe receipts
      • Permanent collection badges appear beside profile and table identities
      • Tester entry offers WeChat and Telegram groups instead of admin contacts
    43. Admin workspace usability improvements

      Improve report refresh state, shareable filters, page names and read-failure messages across the admin workspace.

      • Refresh the active reports and tools with explicit loading, failure and empty states
      • Preserve device, geography and evidence navigation in links and clear old data when periods change
      • Name known pages and tournament states, filter experience groups and reveal the active navigation item on narrow screens
    44. Filter host noise in admin errors

      Inspect other errors when historical host bridge rejections fill the list.

      • Optionally hide identified stackless host bridge rejections while preserving raw records.
      • The list explicitly shows at most 100 matching recent errors, not a complete issue inventory.
    45. Publish corresponding source for lottery committed views

      Update public source for lottery read and persistence changes, including the committed-view module.

      • Keep published source aligned with the deployed implementation.
    46. Tournament arrival counts and second-edition prize places

      Count registrations and actual arrivals separately. The second edition awards nine prize places and requires actual seating and qualifying play.

      • Places 4–9 receive twice the entry fee in gold beans, with no general participation bonus; ties split occupied prize slots.
      • Reconnection, transfers and elimination retain cumulative arrival counts; historical granted rewards remain unchanged.
    47. Tournament deletion and time filters

      Cancel and delete upcoming tournaments directly, with standard quick time filters.

      • Delete draft and published tournaments through the existing cancellation and audit flow.
      • Use standard admin time windows, defaulting to all time to include future events.
    48. Separate lottery reads from persistence

      Lottery current state reads committed memory directly. Purchases and reveals publish after their own commit, while history loads independently.

      • Current-state reads no longer wait for the background writer or history queries
      • Explicit scratch preparation preserves idempotent retries, wallet consistency and bounded memory
  2. Released
    H5Backend

    4 updates in this release

    This release includes: Refresh the tournament WeChat QR code; Clearer tournament results; Keep tournament group chat available; Stronger server build optimization.

    1. Refresh the tournament WeChat QR code

      Tournament group chat now uses the new WeChat QR image, marked valid until November 8.

      • The tournament landing and detail pages share the updated QR code.
    2. Clearer tournament results

      Redesigned podium and standings with elimination times and 100-player pages.

      • Elimination times use the actual start and distinguish champions and survivors.
      • Separate podium cards and paginated standings.
    3. Keep tournament group chat available

      Keep WeChat and Telegram QR-code access on the tournament landing page regardless of edition or phase.

      • Group chat remains available while loading, with no events, or after a loading failure.
      • Both group options remain available on tournament detail pages.
    4. Stronger server build optimization

      Production server builds prioritize more extensive compiler optimization while preserving accounting and panic recovery protections.

      • Runtime optimization takes priority over build time and memory; performance gains remain subject to measurement.
  3. Released
    H5Backend

    3 updates in this release

    This release includes: Reduce telemetry ingestion waits; More mixed-language default nicknames; Nailong custom reaction artwork.

    1. Reduce telemetry ingestion waits

      Telemetry ingestion no longer waits for unrelated game history recovery and returns promptly when unavailable during a release handoff.

      • Preserve runtime ownership, identity readiness, release handoff guards and existing bounded queues.
    2. More mixed-language default nicknames

      Default nicknames for new accounts now include city and alternate-account prefixes with English handles alongside existing Chinese combinations.

      • Adds combinations such as 新加坡tanxuan and 小号briney; existing account nicknames stay unchanged.
    3. Nailong custom reaction artwork

      Replace the sunglasses character with Nailong in the custom sticker and sending animation, preserving the crown, cards and animation effects.

      • The sticker and animation share the Nailong artwork with existing text, sizing and effects unchanged.
  4. Released
    H5BackendMarketing site

    16 updates in this release

    This release includes: Start guidance in private Zha Jin Hua rooms; Complete the tournament statistics migration pair; Retain local slow settlement diagnostics; Report code quality fixes; Simpler game pages and membership information; Correlate play attempts and prepare gated entry measurement; and 10 more changes.

    1. Start guidance in private Zha Jin Hua rooms

      Before the first deal in a private room, hosts can directly fill empty seats with AI and start, or invite friends.

      • Reuses the existing start command and adds AI only after an explicit click.
      • Adds bounded guide exposure, click and first-hand observations for follow-up evaluation.
    2. Complete the tournament statistics migration pair

      Add the matching rollback definition for tournament completion statistics without executing a rollback.

      • Restore prior statistics functions and queue derived-fact reconciliation through the existing background writer.
    3. Retain local slow settlement diagnostics

      Settlement phase samples no longer depend on alert configuration, helping distinguish queue and persistence delays.

      • Retain process and settlement identities, queue wait and eight phase timings.
      • Rate-limit samples and use a nonblocking log queue to protect settlements.
    4. Report code quality fixes

      Resolve strict lint errors in daily reports and admin metrics while preserving reporting semantics.

      • Remove redundant string formatting and unused fields, and simplify counting.
    5. Simpler game pages and membership information

      App help links open the static website for full instructions, while pricing focuses on plans and benefits.

      • Calculator guides live on the website; H5 retains a help link.
      • Detailed membership answers are grouped in the website FAQ and expand on demand.
    6. Correlate play attempts and prepare gated entry measurement

      Play requests correlate accepted and failed responses by attempt; short-lived website-to-app click correlation stays disabled pending notice and separate activation.

      • Distinguish accepted requests from seats and completed hands
      • Time-limited random click tokens without transferring persistent anonymous IDs
    7. Include every account in admin reporting

      Count actual activity from staff, test accounts and QA browsers in admin reports and performance statistics.

      • Remove identity exclusions from growth, retention, revenue, games, activity and daily reports.
      • Include retained test samples in latency trends and show test account emails on live tables.
    8. Add up to two custom bottom tabs

      Keep Games, Join Game and Me as the three default tabs, and choose up to two additional shortcuts in app settings. Choices are saved to your account.

      • Choose zero to two shortcuts while keeping existing preferences
    9. Count completed tournament hands in first-play statistics

      Include settled tournament participation in account completion and retention facts, with bounded asynchronous historical catch-up.

      • Deduplicate by account and hand; count only dealt-in participants of settled hands.
      • Keep pre-registration completions separate from registered cohorts without adding regular records or financial totals.
    10. Correct registration upload-wait alerts

      Allow evidenced upload waits for successful registrations within 500ms total, while retaining the 100ms server-work budget.

      • Body waits need not exceed half of total time; missing evidence, failures and slow server work still alert.
      • Alerts and server timing share the rule, preserving diagnostics and multi-client wait escalation.
    11. Notification inbox opening alerts

      Notify administrators when a user opens the notification inbox, including the latest notification.

      • A single inbox-opening trigger supports empty and already-read inboxes.
      • Preserves request deduplication, paced delivery and retries.
    12. Simplify notification opening alerts

      Remove the extra explanatory sentence from Telegram inbox-opening alerts.

      • Retain the user identifier, opening action, latest notification and timestamp.
    13. Complete record-history request diagnostics

      Measure connection acquisition, database queries and response processing for record-history requests.

      • Reuse bounded sampling and request correlation to distinguish database and response-processing stages.
    14. Clarify membership prices and interface languages

      Dedicated website answers explain Free, Pro and Max web prices, store pricing differences, and Chinese/English interface support.

      • Price answers reuse existing plan data and distinguish web and store amounts.
      • Language guidance is shared by visible answers, structured data and public AI summaries.
    15. Smaller card-history preload batches

      Zha Jin Hua and Short Deck preload full rounds in smaller batches and report slow hydration fetch and indexing times.

      • Preserve complete history, the single transaction snapshot and existing capacity limits.
    16. Paginated lobby bean leaderboard

      Browse the bean leaderboard with up to 50 players per page.

      • Page numbers and previous/next controls preserve continuous rankings.
  5. Released
    H5Backend

    5 updates in this release

    This release includes: Tournament invitations include text and a link; Tournament sharing invites friends to compete; Restore the tournament overview entry; Correct recovery notices without a summary; Improve alert delivery diagnostics.

    1. Tournament invitations include text and a link

      Copy the tournament name, prize invitation and registration link together.

      • Preview, select and copy an invitation featuring the event’s champion prize.
    2. Tournament sharing invites friends to compete

      Invitations highlight the champion prize and final-table challenge, with the start time and registration link.

      • Chinese invitations display large gold-bean prizes in compact units without losing precision.
    3. Restore the tournament overview entry

      The home tournament banner now always opens the tournament landing page with current events, results and past tournaments.

      • The banner opens the tournament landing page even when an event is accepting entries.
    4. Correct recovery notices without a summary

      Recovery notices without a monitor summary also describe prior alert delivery as unconfirmed.

      • Use consistent delivery status wording with and without a summary.
    5. Improve alert delivery diagnostics

      Alert monitoring distinguishes transport timeouts from API rejection and accurately labels unconfirmed delivery.

      • Retain credential-free HTTP status and connection, TLS, and response timings.
      • Avoid treating a request timeout as authorization failure or proof of non-delivery.
  6. Released
    H5BackendMarketing site

    37 updates in this release

    This release includes: Improve script diagnostics on foreground return; Tournament time filters and removal; Fix joining busy friend rooms; Fix first deals on busy tables; Refresh the public source snapshot; Clearer Zha Jin Hua rules navigation; and 31 more changes.

    1. Improve script diagnostics on foreground return

      Preserve startup evidence and briefly observe script activity after foreground return to investigate browser-redacted errors.

      • Bounded sampling preserves gameplay and reporting of unknown errors.
    2. Tournament time filters and removal

      Filter tournaments by start time and confirm removal of completed or cancelled editions from the staff list.

      • Time filters remain active across pages
      • Preserve entries, results, rewards and audit history
    3. Fix joining busy friend rooms

      Friend-room admission waits for persistence capacity before checking and binding a seat, avoiding room-unavailable responses from a temporarily full queue.

      • Handle queue backpressure for mid-hand friend-room admission
    4. Fix first deals on busy tables

      First deals wait for persistence queue capacity before starting the room clock, so a temporarily full queue cannot leave the table waiting indefinitely.

      • Handle persistence backpressure before the first deal
    5. Refresh the public source snapshot

      The public source snapshot matches the dependencies and corresponding implementation used by this release.

      • Update the source link and verify the published snapshot.
    6. Clearer Zha Jin Hua rules navigation

      The bilingual guide now highlights hand rankings, 2-3-5 and showdown, with section navigation and a clearer link from the game page.

      • The hand-ranking table has row and column headers, and rule sections have direct anchors.
    7. Clearer official identity and game boundaries

      Share bilingual identity, language, source-code scope and no-real-money answers across pages and AI summaries.

      • About directly answers whether BluffKing offers real-money poker.
      • Visible pages, structured data and AI summaries share one answer source.
    8. Complete Squid settlement timing evidence

      Retain queue, writer admission and transaction phase timing for Squid settlements.

      • Reuse bounded diagnostic sampling while preserving settlement, retry and commit semantics.
    9. Retain more evidence for shared latency investigations

      Increase bounded proxy diagnostic retention and update offline file discovery capacity.

      • Preserve alert thresholds, compression chunk size and age limits; this does not establish a network latency fix.
    10. Retain correlated samples for shared latency alerts

      Shared latency alerts retain bounded slow observations from distinct clients in the triggering scope to support transport and server timing diagnosis.

      • Keep existing alert thresholds; retain only correlation IDs and fixed timing fields, without account information.
    11. Reduce settlement database trips and improve diagnostic coverage

      Use optimistic wallet versions, remove redundant participant prelocks and batch bean settlements while preserving atomic receipts and replay safety.

      • Gameplay keeps in-memory balances; background balances and ledgers commit together.
      • Remove periodic observer disconnects and decouple process discovery from host sampling.
    12. Request attribution details and checkers computation

      Show unregistered request reasons and slow counts, reuse repeated checkers position scores, and retain bounded slow-operation evidence.

      • Distinguish unmatched routes from slow requests; mark unavailable historical breakdowns.
      • Preserve checkers search and move selection while reusing unchanged opponent position scores within each decision.
    13. Restore proxy timing evidence for history requests

      Include the current history endpoint in bounded proxy timing observations to separate application work from external waiting.

      • Preserve sampling limits and pass-through behavior; this diagnostic change does not establish a latency fix.
    14. Unified database writes and registration rewards

      Business writes share one execution queue while reads remain independent; nested transactions and savepoints are removed.

      • Single statements execute directly; related writes retain atomic commit and rollback.
      • Optional room invitation rewards run after registration commits, so reward failures do not undo registration.
    15. Fix diagnostic recorder task exhaustion

      Allow bounded overlapping observer and discovery tasks to retain application and proxy process evidence.

      • Keep the existing CPU, memory and sampling limits.
    16. Reduce independent diagnostic recorder overhead

      Discover processes through bounded local read-only requests instead of repeated CLI startup.

      • Preserve resource limits, process identity checks and failure isolation.
    17. Complete guest sign-in promotion timing

      Measure guest promotion and login response preparation stages to help diagnose slow sign-ins.

      • Preserve login checks, transactions and alert thresholds; diagnostics retain fixed stage labels only.
    18. Time filters for player experience

      Player experience monitoring now offers the shared admin time dropdown, defaulting to the last 24 hours.

      • Choose 1, 3, 6, 12 or 24 hours, 3 days or 7 days for operations, timings and errors.
    19. Reduce account preference queue entry size

      Store appearance updates indirectly to remove the enum size lint while preserving preference saving behavior.

      • Smaller preference updates no longer reserve inline space for a full appearance value.
    20. Clearer ZJH compare pricing and auto-showdown rules

      Comparing costs twice the initiator’s current call, with clearer auto-showdown rounds and improved multiway bot decisions.

      • Auto-showdown defaults to 3 rounds, selectable from 1–10; looking at your own cards stays available at any time.
      • Ordinary compares, short-stack all-in compares and automatic showdowns have distinct explanations in the app and website.
      • Bots preserve strong hands in low-pressure multiway play, with bounded decision-reason diagnostics.
    21. Tournaments finish within two hours

      New tournaments raise blinds every 5 minutes with no scheduled break, and after 120 minutes finish the current hand and rank players by remaining chips.

      • A full 90-player field usually produces a champion around level 23; otherwise remaining stacks decide the places, with equal stacks tied.
      • Events without breaks no longer show a next-break timer, and every table keeps dealing independently.
      • Copied editions also have no scheduled break.
    22. Tournament tables gain the cash table's interactions

      Tournament tables now have hand history, live standings, a visible time bank, seat reporting, chat remarks with a return-to-table prompt, an elimination card, and desktop side rails.

      • New Hands menu: review your hole cards, board, chip change and street-by-street actions per hand.
      • Standings shows the live chip ranking, provisional places of eliminated players, level/break countdowns and prizes.
      • The tournament time bank applies automatically after the 15-second action clock and is now visible on the table.
      • Long-press an opponent to report; chat gains remark names and a return-to-table prompt; a card shows your provisional place after elimination.
    23. Back button works again while spectating a tournament

      The spectating notice no longer covers the table's top bar, so spectators can go back to the tournament page.

      • On tournament, Sangong, Niuniu and Blackjack tables, notices sit under the top bar so the back and menu buttons always respond.
    24. Tournament early seating and honors

      Take a seat 30 minutes before play; each champion receives an MVP medal and podium finishers receive colored names.

      • Waiting tables show a countdown and play begins at the scheduled start.
      • Gold champions, purple runners-up, orange third-place finishers; honors persist from sealed results.
    25. Tournament results highlight the podium

      Final standings show the champion, runner-up and third place as medal-coloured cards, with the rest in a compact list.

      • Rank and name now sit together instead of at opposite edges.
    26. Tournament entry fees and blind alerts

      Add atomic bean entry fees, refunds and full-table blind alerts.

      • Insufficient-balance guidance links to bug-report rewards
      • Timed events support blinds doubling every 15 minutes
    27. Texas tournaments: 3-second clock for players who stay offline

      A player who times out three decisions in a row while offline gets only 3 seconds per decision afterwards, so absent players no longer stall the table.

      • After three consecutive timeouts while offline, the decision clock drops from 15 to 3 seconds with no time bank.
      • Reconnecting or acting manually restores the full 15-second clock immediately.
    28. Tournaments start with 1,000,000 chips

      New tournaments give each player 1,000,000 chips, starting at 1,000/2,000 with a 1,000 big-blind ante.

      • The big-blind ante is half of each level’s big blind; the blind structure lists every level’s amounts.
      • Published events keep their original stacks and blinds. Rebuys and re-entry remain unavailable.
    29. Independent tournament table pacing

      Remove hand-for-hand play from 18 remaining so each table continues at its own pace.

      • Seat transfers and table consolidation still wait for the affected hands to finish.
      • Scheduled breaks, staff pauses and historical result recovery remain supported; event rules are updated in both languages.
    30. Tournament staff can force a player out

      Staff can force a rule-breaking or long-absent player out from the tournament entries list: they leave before the next hand, finish in their current place and keep it. Disqualify stays separate. The list now also shows each seated player's table, seat and connection state.

      • Force-out takes effect at the hand boundary and survives a restart
      • Entries list shows online state and offline minutes
      • Audit log gains admin_remove / removed events
    31. Tournament arrival seating and live table counts

      Tournaments share the Hold’em surface and player avatars, open random seating 30 minutes before the start, and deduct absent entrants’ blinds off-table.

      • First admission closes 15 minutes after the start; seated players can reconnect.
      • Event details and the table show actual table and seated-player counts, including each table’s occupancy.
      • Absent stacks lose a small blind, big blind and ante (if enabled) per minute of blind-clock time, excluding pauses and breaks.
    32. Texas Hold'em action clock unified at 15 seconds

      Every Texas Hold'em decision — bean tables, private chip tables, Short Deck and tournaments — now allows 15 seconds instead of 20; the +60 s time bank is unchanged.

      • Bean and private Texas tables count down 15 s instead of 20 s
      • Short Deck decisions are 15 s
      • Tournaments default to 15 s per action; events not yet started are updated
      • Timeout rule unchanged: check when possible, otherwise fold
    33. Two-way player report conversations

      Staff can contact reporters from the report queue, and players can reply through notifications and My Feedback.

      • Report conversations include message history and read status.
      • Player follow-ups reopen resolved reports for investigation.
    34. Hold’em report reviews and verification guidance

      Two anonymized case studies explain cheating and collusion report evidence, findings and player verification steps.

      • Historical deal, pot and settlement checks distinguish legacy and current proofs.
      • Shared-hand fold and all-in comparisons explain the scope of zero-knowledge verification.
    35. New “Not everyone gets to be Hadi” reaction

      A new “Not everyone gets to be Hadi” reaction joins the table stickers: a crowned, shades-on face shows off the champion A-5 offsuit, gives an unbothered nod and a glint off the shades. Send it in chat or throw it at an opponent's avatar for a crown-drop impact.

      • New “Not everyone gets to be Hadi” reaction (champion A-5 offsuit) for chat and seat throws
    36. Dou Dizhu card counter preference

      The card counter defaults to off, with a direct table toggle that preserves your saved preference on this device.

      • See and change the counter setting directly at the table.
      • Off by default; your choice persists when you return on this device.
    37. Clarify login refusal and shared latency evidence

      Record the Apple account-link protection refusal and the service scope of shared latency alerts.

      • Preserve account protection, grant revocation, and alert thresholds.
  7. Released
    H5BackendMarketing site

    9 updates in this release

    This release includes: Update the zero-knowledge proof blog for launch; Tournament flip countdown; Smoother Sic Bo settlement; Daily reports separate chips and gold beans; Align registered account counts in admin reports; Clearer membership comparisons and public source information; and 3 more changes.

    1. Update the zero-knowledge proof blog for launch

      Update both languages to describe live default Hold’em verifiable dealing, with verification access and requirements.

      • Replace pre-launch wording and refresh summaries and the update date.
      • Explain verification for dealt-in players, local evidence, legacy hands and the data supplied by the server.
    2. Tournament flip countdown

      Tournament countdowns now use a prominent full-width gold flip clock.

      • Stable hours, minutes and seconds with digit flips and reduced-motion support.
    3. Smoother Sic Bo settlement

      Collect losing stakes and distribute winnings across the table, then move into the next round with each seat’s previous result visible.

      • Dealer collects losses and pays winning seats
      • Shorter settlement without an extra next-round countdown
      • Seat labels show the round, verdict and net change
    4. Daily reports separate chips and gold beans

      Completed-game players are counted separately by chips and gold-bean mode; popularity explicitly shows page visitors across all modes.

      • Separate players and results by settlement mode for each game
      • Count players in each mode they played and label unknown modes separately
      • Include settled gold-bean Sic Bo rounds in completed-game totals
    5. Align registered account counts in admin reports

      Registration totals, recent signups and weekly counts include Apple, Telegram and accounts without email, excluding current staff and test accounts.

      • Admin and daily reports share registration counts using the time each account became registered.
    6. Clearer membership comparisons and public source information

      English and Chinese pricing pages explain Free, Pro, Max and the paused Coach Chat; About directly states product identity, Chinese and English interface support, and the scope of public source code.

      • Free and paid feature answers are visible with stable citation anchors
      • Pricing, FAQ structured data and existing membership summaries share answers
      • Clarifies that public engines and solvers do not make the entire hosted service open source or independently audited
    7. More complete membership comparison answers

      The pricing page's Free, Pro and Max comparison now states advanced-review scope and clarifies that bot practice and calculators need no registration.

      • Advanced review covers 14 days on Free, 365 days on Pro and all history on Max
      • Shorter English pricing page title
    8. Gold-bean mode for Zha Jin Hua, Bull and Short Deck

      All three games retain ordinary play and add gold-bean tables through the Gold Beans hub.

      • Buy-ins, top-ups and hand settlements use the shared gold-bean wallet.
      • Each room preserves its denomination across reconnects.
    9. Improve transient Apple sign-in recovery

      Retry a transient Apple signing-key fetch once and distinguish service failures from invalid credentials.

      • Bounded retry preserves signature and replay checks.
      • Record correlated signing-key fetch outcomes.
  8. Released
    H5BackendMarketing site

    3 updates in this release

    This release includes: Synchronize website interaction catalog; Tournament start and end countdowns; Gold bean codes default to the last day.

    1. Synchronize website interaction catalog

      Update the website interaction catalog so backend reporting matches the pages in this release.

      • Synchronize interaction identifiers on release notes and archive pages.
    2. Tournament start and end countdowns

      Tournament details and the tournament hub count down during the final 24 hours before play and to the end of timed events.

      • Countdowns tick every second using server time.
      • Freezeouts explain that play ends when a champion is decided.
    3. Gold bean codes default to the last day

      The admin page defaults to gold bean codes created in the last 24 hours, using the same time dropdown as other admin pages.

      • Reuses the admin time options, filters before pagination, and keeps the same time window when loading more.
  9. Released
    H5BackendMarketing site

    3 updates in this release

    This release includes: WeChat contact and group diagnostics; Scheduled statistics retention; Track home bounty announcement clicks separately.

    1. WeChat contact and group diagnostics

      Separate personal and group entry points and observe QR loading, copy and save attempts.

      • Show event and distinct identity counts with explicit coverage limits.
      • Record hidden entries and failures without claiming confirmed WeChat joins.
    2. Scheduled statistics retention

      Expired statistics are cleaned in bounded batches even when no new telemetry arrives.

      • Preserves existing retention windows and records cleanup progress and failures.
    3. Track home bounty announcement clicks separately

      Give the home bounty announcement a stable click label so existing usage analytics can report clicks and estimated unique visitors separately from generic promotions.

      • Bounty and tournament entries have separate click labels; historical generic promotion records are not backfilled
  10. Released
    H5Backend

    3 updates in this release

    This release includes: Synchronize corresponding source for this release; Live tables show time since the last human input; Error reports default to the last 12 hours.

    1. Synchronize corresponding source for this release

      Update public source references to match this backend release.

      • The public source tag has been verified byte for byte.
    2. Live tables show time since the last human input

      The live table list now shows seconds since the latest human input at each table, updating every second.

      • Bot turns and heartbeats do not reset the timer; missing observations display —.
    3. Error reports default to the last 12 hours

      Admin error reports now default to the last 12 hours, with other time ranges still selectable.

      • Opening error reports queries the last 12 hours by default.
Browse the release archiveExplore 750 earlier public updates