A 106-minute comparison during an ongoing Twitch LIVE

During a 106-minute, 21.621-second observation, we found 5,899 of 6,003 new author-and-text groups from a sampled native Twitch chat in the production TikChat view. 104 groups remained unresolved. The resulting 98.268% describes group presence in this sample, not the percentage of all Twitch messages captured.

We joined an already-running broadcast and ended the observation at the owner's request. At that point, the native Twitch page still indicated LIVE and TikChat showed Twitch Connected. We did not observe the complete broadcast or measure its audience size.

This was our own product test, commissioned by Matt Olech, the owner of TikChat. The owner started the production session; an AI assistant operated the browser views and performed the comparison. The channel, broadcast title and viewer identities are omitted from the report and its evidence files.

Test setup

The reference was the native browser chat of one public Twitch channel. The application under observation was the owner's existing production TikChat session, with Twitch as the source being compared. The MultiChat setup guide explains how sources are connected to the operator.

  • Study dates: November 15–16, 2026.
  • Platform: Twitch only.
  • Environment: the real production TikChat operator.
  • Reference window: 106 minutes and 21.621 seconds.
  • Final comparison window: a further 90 seconds, reading TikChat only.
  • Unit: a distinct normalized author-and-plain-text group, after excluding the initial native buffer.
  • Selection: one available public broadcast, without random selection or a control group.
  • Evidence: anonymized aggregates, relative sampling times, protocol, measurement code and product fingerprints.

The recorded application release was release-sales-multichat-b5fc1dfbf0a6; the website release was youtube-study-links-79d826914425. Release paths, selected file hashes, service process identifiers and start times agreed at the checks before, during and after observation. The downloadable version record retains releases and file hashes without publishing process identifiers or exact session clocks.

The broadcast and TikChat source were already running when formal sampling began. Observing a source after it has been connected does not test the source-attachment process.

What was the purpose of the test?

We asked whether content observed in the native Twitch browser chat could also be found in the production TikChat view, and whether its complete rendered representation agreed under a defined method.

The native chat’s rendered browser structure (DOM) did not expose stable provider message IDs to our reader. We therefore compared groups defined by the displayed author and normalized plain text. Repeated identical text from the same displayed author collapses into one group. A match establishes corresponding content, but cannot distinguish every occurrence of that content.

We also kept four stages separate: sending, native-chat visibility, independent backend receipt and TikChat DOM presence. Only the two browser stages were measured in this passive observation.

Method

Establish the native chat and production session

Before formal collection, we saved the protocol, DOM reader, counter and prepared controlled-message plan with their SHA-256 hashes. Calibration was excluded from the reported sample, and fresh starting buffers were read immediately before the formal window.

The operator's source scope was checked. Twitch showed Connected, search was empty and the Hide A–D replies filter was off. On this third-party channel we observed natural chat only: no synthetic messages were sent and no provider disconnects were forced.

Exclude the initial reference buffer

We excluded 130 groups already present in the starting native buffer. Those groups remained excluded if the same author-and-text combination appeared again later. The initial TikChat buffer could establish prior presence for a group first observed later in the native reference.

This prevents the starting backlog from being counted as new reference material. It also means the result is not a count of every message posted during the window.

Read both browser views repeatedly

The target was approximately one paired read every five seconds. The completed record contains 968 native Twitch reads and 978 TikChat reads after the baseline; ten TikChat reads fell within the final comparison window.

Each read inspected up to 500 recent native rows or 800 recent matching Twitch rows in the operator. The native buffer actually contained 145–150 rows per read. Actual intervals, row exclusions, bounded reads and buffer-overlap warnings were recorded rather than assuming uninterrupted sampling.

Match author-and-text groups and compare content

Displayed authors were normalized to Unicode NFC, whitespace was collapsed and letter case was lowered. Plain text used NFC and collapsed whitespace with trimmed edges; its letter case and punctuation remained significant. Recognized CTCP ACTION wrapping was removed for presentation comparison.

The grouping text excluded inline images. Their alt labels were included in a separate complete-content comparison. Emote-only rows with no usable plain text, and rows without a usable author or body, were excluded from the group denominator. Their repeated DOM observations are recorded separately.

An ephemeral salt and temporary digests supported comparison without writing comments or viewer names to the evidence. A group was marked ambiguous when the same signature occurred multiple times in one buffer or its complete content representation changed. Ambiguous matched groups were excluded from unambiguous content-equality claims.

Save aggregates and complete the final comparison

Progress snapshots were saved approximately every five minutes. An unmatched group was initially pending. After 60 seconds without a match it became unresolved, while remaining eligible for a later match.

At the owner's stop request, we closed the native-reference window and applied the planned 90-second final comparison window. Only operator reads inside its cutoff were counted. Working comparison maps and the salt were disposed after aggregate export.

The observer made no deployments, service restarts, production fault injections, additional TikChat collectors or additional private-API polling loops. Normal application collection continued. Server resource use was not measured.

What happened during the observation?

The first five-minute snapshot contained 814 reference groups, with 799 already matched, five pending and ten unresolved. At approximately 60 minutes, the sample had reached 4,260 groups, with 4,207 matched and 53 unresolved. Later matches could change the status of earlier groups.

At the end of the comparison, the sample contained 6,003 groups: 5,899 matched, 104 unresolved and none pending. Eleven supplementary checks showed Twitch Connected and empty search. These were point checks, not a continuous connection trace.

The reader recorded zero chat-read errors and zero rejected reads. It recorded one native buffer-overlap warning. These observations concern the measurement process; they do not prove complete capture or uninterrupted provider connectivity.

The native page still indicated LIVE at the stop request. The 106-minute interval describes our observation, not the broadcast's total duration or natural ending.

Results and evidence

Did sampled Twitch content appear in TikChat?

5,899 of 6,003 new reference groups were also found in the production TikChat DOM. A group means the same normalized displayed author and plain text, not a uniquely identified Twitch message. The observed group-match ratio was 98.268%.

Sampled author-and-text groups compared with the production TikChat DOM

Group comparisonGroups
New groups in the native Twitch sample6,003
Found in the production TikChat DOM5,899
Unresolved at finalization104
Still pending at finalization0
Initial native groups excluded130


What was measured at each stage?

Separate measurement status for sending, native visibility, backend receipt and rendering

StageMeasurementResult
SendingNot measuredNo controlled sender; observer sent zero test messages
Native-chat visibilitySampled DOM groups6,003 new groups in 968 reads
Backend receiptNot independently measuredNo database or collector-ingress read
TikChat renderingSampled DOM presence5,899 reference groups found in 978 reads


DOM presence does not establish that every row was inside the visible viewport or was read by the creator. Unmeasured sending and receipt values remain unknown, rather than being reported as zero.

How did the recorded result develop?

Cumulative group results at selected progress snapshots

Progress snapshotReference groupsMatched thenPendingUnresolved
About 5 minutes814799510
About 25 minutes2,1012,070526
About 50 minutes3,8203,777142
About 75 minutes4,7814,704077
About 100 minutes5,8465,750294
About 105 minutes5,9755,8713101
Final comparison6,0035,8990104


These are cumulative snapshots at approximately the stated times. All 21 five-minute checkpoints, their actual relative times and the final evaluated cohorts are in the attached data.

What does an unresolved group mean?

An unresolved group is one for which this method did not establish a TikChat-view match before finalization. The 104 groups are not 104 messages proven lost by the backend. Their causes were not established. Reception, presentation, moderation, bounded buffers and observation limits are areas for further investigation, not diagnosed causes.

Did the message content agree?

Among the 5,899 matched groups, 5,413 had matching complete normalized rendered content. Another 486 were ambiguous and were excluded from unambiguous content-pair claims.

Complete content comparison among 5,899 matched groups

Content comparison among matched groupsGroupsInterpretation
Matching complete normalized representation5,413Includes inline-image alt labels under the stated normalization
Ambiguous matched groups486Repetition or representation changes prevented an unambiguous pair claim
Total matched groups5,899Denominator for this comparison


Matching already requires the same normalized author and plain text. The separate content check therefore does not independently validate the wording of every comment. The 104 unresolved groups have no established pair, and the 486 ambiguous groups retain uncertainty.

The counter recorded 29 reference-side and 253 operator-side content-digest revisions, including baseline observations. These are revision events, not necessarily that many distinct groups or messages. Their causes were not established. No additional differences were classified in the remaining unambiguous subset; this is not a zero-defect result for all chat traffic.

Evidence files and verification

The attached files omit the channel name and URL, broadcast title, viewer names, comments, message and account identifiers, working fingerprints and salt. Dates remain visible; measurement times are relative to the formal observation start.

The original ZIP contains the reader, counter, frozen hashes, coverage data, all checkpoints and verification tools. Twelve synthetic counter tests passed; they validate observer logic, not the live Twitch transport or TikChat recovery.

Readers can check arithmetic, timing and file integrity, and use the code for a fresh observation. The original group-by-group comparison cannot be independently replayed after disposal of the working data. Checksums detect changes relative to the supplied manifest; they are not independent certification. More context about the product's evidence is available in Product & sources.

What does the evidence support?

The supported observation is specific: during one ongoing Twitch broadcast, 5,899 of 6,003 new author-and-text groups in our native browser sample also occurred in the production TikChat DOM. The report separately identifies matching complete representations, ambiguous groups and unresolved cases.

This documents corresponding chat content in the real product over an extended observation window. It does not establish how many individual messages were sent, what proportion of all messages was captured, or whether every repeated occurrence appeared in TikChat.

The 106-minute duration and 6,003-group sample describe this session. Audience size, engagement and server capacity were not measured.

Limitations

Group identity is weaker than message identity

Repeated author-and-text combinations collapse, and a later repetition can establish a group match without proving the earlier occurrence was received. Baseline groups remain excluded on repetition, and emote-only messages are outside the denominator.

Twitch does provide message IDs in its IRC protocol; our limitation was the browser DOM available to the reader. See Twitch's IRC documentation. Our 99-minute YouTube chat study used message IDs, so its percentage and this group ratio must not be ranked as equivalent capture-rate measurements.

The native reference was sampled

Actual intervals between browser reads

Read intervalNative TwitchTikChat operator
Median5.002 s5.002 s
95th percentile13.090 s13.164 s
Longest118.726 s118.729 s


The first formal native read started 16.141 seconds after the window began; the last ended 24.452 seconds before it closed. One consecutive nonempty native-buffer pair had no shared group keys. Overlap elsewhere does not prove full coverage, and the number of missed messages is unknown.

The final 90-second comparison window contained ten operator reads. The first began 27.609 seconds after reference closure and the last ended after 73.311 seconds. The final 16.689 seconds were not sampled. No post-cutoff reads were included. This was not continuous recording.

Both browser views had bounded buffers

The operator reader deliberately used at most 800 recent matching rows; 966 reads were marked as bounded. That flag is not an application-loss count. Excluded or repeated rows in the log are repeated DOM observations, not distinct failed comments or confirmed duplicate-message defects.

Observation timing was not transport latency

Seven groups were first matched more than 60 seconds after their initial native observation. Sampling pauses and repeated content can contribute to that interval. It is not a measurement of provider delay or a transport-latency percentile.

This was one first-party production session

There was no independent reviewer, random sample of channels or separate backend-receipt measurement. The version checks were point checks, not continuous runtime attestation. CPU, RAM, I/O and network use were not measured, so avoiding deployments and restarts is not proof of zero resource impact.

During initial calibration, Twitch displayed a video-playback-unavailable notice; the native chat subsequently loaded and was sampled. Video and audio were not the reference. An optional-variable cleanup warning after final export was corrected without affecting counts; it is disclosed in the error record.

The planned controlled series of 120 synthetic messages was prepared but not sent. Late source attachment and provider reconnect were not tested. Joining our observation midway through a broadcast is different from measuring the application's source attachment, and refreshing an operator tab would not test the server's provider connection.

What this means for creators

This test provides an inspectable record of the production Twitch chat view during one ongoing broadcast. It shows that thousands of author-and-text groups observed in the native sample also appeared in TikChat, while retaining unresolved cases and the limits of group matching.

Creators can use that evidence alongside the TikChat feature overview and setup documentation when assessing the tool. The practical result concerns the chat presented in this session. It does not measure improved engagement, creator response time or complete message capture.

Frequently asked questions

Was this a test of the real TikChat application?

Yes. The owner started a production TikChat session with a Twitch source. The observer read that operator alongside the native Twitch chat. A local replacement application was not used for the observation.

What does 98.268% mean?

It means 5,899 of 6,003 new normalized author-and-text groups observed in the native browser sample were also found in TikChat. Repeated identical text from the same displayed author forms one group. The percentage does not describe all Twitch messages or individual-message capture.

Did the comments themselves agree?

Complete normalized rendered content agreed in 5,413 matched groups. Another 486 matched groups were ambiguous. Pairing already fixes plain text and author, so this is not independent verification of every comment or every repeated occurrence.

Were 104 messages definitely lost by TikChat?

The study did not establish that. It retained 104 groups without a confirmed operator-view match. The causes remain unresolved, and backend receipt was not independently measured.

Did the observation cover the whole Twitch LIVE?

No. Sampling began during an already-running broadcast and ended at the owner's request while the native page still indicated LIVE and TikChat showed Connected. The broadcast's full duration and audience size were not measured.

Did this test validate reconnect or adding a Twitch source?

No. The source was already connected when observation began. No provider disconnect was forced, and the planned synthetic-message series was not sent. Those scenarios require a separate controlled test.

Can readers verify the result?

Readers can inspect the aggregates, sampling log, protocol, version fingerprints, code and checksums. They can rerun consistency checks, but cannot independently reconstruct each original group match after the working fingerprints were disposed.

Conclusion

Across a 106-minute, 21.621-second first-party observation, 5,899 of 6,003 sampled Twitch author-and-text groups were also found in the production TikChat DOM. There were 104 unresolved groups, with content agreement assessed separately and ambiguous matches retained.

We joined the broadcast in progress and ended observation while it still reported LIVE. The method, data and limitations document this single-session result and provide a basis for a subsequent controlled Twitch test.

For connection instructions, see the TikChat MultiChat setup guide.