Repeated Telegram messages: a practical anti-flood policy
Separate accidental duplicates, fast conversation and deliberate flooding. Choose targeted restrictions, test useful messages and understand Defendy’s current limits.
A support question posted twice is irritating. A dozen people discussing a release can be noisy. A seller repeatedly posting the same promotion is a different moderation problem. Treating all three as “too many messages” produces rules that punish useful participation without necessarily stopping abuse.
Start by identifying the disruption, then choose the smallest measure that addresses it. Telegram provides slow mode and member restrictions. Defendy can filter prohibited content and apply configured consequences, but its dedicated anti-flood module is currently marked as planned. There is no available Defendy setting that promises to detect every duplicate or enforce a custom messages-per-minute allowance.
If the disruption comes from reactions rather than new messages, follow the reaction-spam guide. Investigate that mechanism before limiting ordinary conversation.
Diagnose the pattern before choosing a limit
Read a small amount of surrounding conversation. Is one person repeating themselves, are several people answering the same prompt, or are new accounts arriving together? Did the author stop when asked? Does the message break a content rule even when posted once?
| Pattern | First response | Main risk |
|---|---|---|
| An accidental duplicate | Ask the author to keep one copy | Turning a mistake into a disciplinary record |
| Useful but rapid discussion | Ask for consolidated replies; consider a short slow-mode interval | Making follow-up questions difficult |
| One person repeatedly disrupts | Explain the rule, then use a targeted temporary restriction if needed | Restricting the wrong person or using an excessive duration |
| Recurring prohibited promotion | Configure the relevant content or link rule | Blocking legitimate recommendations with an overly broad match |
| Many accounts flood together | Review entry controls and moderator coverage | Expecting one per-person delay to stop a coordinated incident |
A message count cannot establish motive. Someone may use speech input, type in short bursts, or need an extra message to correct a translation. Build room for those cases into the rule, rather than treating a preferred writing style as proof of good behavior.
Three tools that solve different problems
Slow mode: make each participant pause
Telegram describes slow mode as a delay between a member’s messages, with a timer before the next send. It is useful when people need more time to read. See Telegram’s official explanation and our configuration guide.
The editorial implication is important: a room with many active participants can still move quickly. Slowing each person does not create a single shared queue or decide which comments are worthwhile. Tell members why you are enabling it and when you will review the setting.
Content filters: remove a prohibited message
Defendy’s text anti-spam, link rules and stop-word lists work on different content signals. A repeated advertisement may match one of those rules. A second copy of an otherwise ordinary question is not guaranteed to match merely because it is a duplicate.
Choose the rule based on what makes the message unacceptable. Do not add “help” or “any updates?” to a stop-word list simply because members repeat those phrases. That can suppress the purpose of a support community.
Targeted restrictions: stop a particular disruption
If one member continues after a clear request, use the consequence described in your community rules. A temporary personal restriction can leave everyone else’s conversation intact.
For example, imagine a member repeatedly reposting the same request after a moderator has acknowledged it. A moderator might reply to that member’s message with /tmute 10m repeated flooding. This is a hypothetical ten-minute intervention, not a universal punishment recommendation. Check the author, your authority and the bot’s permissions first, then verify the result. The moderation command guide explains supported commands.
Telegram also allows administrators to restrict categories of content through group permissions. Before removing media rights from everyone, consider whether you would also block troubleshooting screenshots, accessible voice contributions or course materials. One person’s sticker burst seldom justifies an indefinite ban on every attachment.
A hypothetical release-day incident
Suppose a small software community announces an update. Members post installation questions; several say “same here”; one person repeatedly promotes an unrelated service.
A sensible response separates those threads. A moderator gathers installation questions into a clear discussion, asks people to include the relevant details in one message and considers a short slow-mode interval if the conversation remains hard to follow. The promotional account is handled under the advertising rule.
A blanket duplicate policy would be a poor fit: several identical error reports may contain useful evidence, while slightly reworded advertisements remain unwanted. What matters is whether the intervention preserves the discussion members came for.
After the launch rush, review the temporary changes. Keeping a restrictive setting forever because it helped during one event creates unnecessary friction later.
Configure Defendy without inventing a flood detector
In the Mini App, select the intended group before changing anything. For text classification, open Anti-spam → Text. For a specific phrase, use the stop-word controls; for prohibited destinations, use the link filter.
Make the consequences explicit:
- Check whether the matching message will be deleted.
- Check the action applied to its author and any duration.
- If the action is a warning, review the warning thresholds and the member’s existing points. One more warning may trigger a restriction.
- Enable the relevant logging categories if you plan to use a moderation log for review.
- Change one rule at a time so you can identify the cause of a false positive.
A warning action does not by itself mean “keep the message.” Deletion and the author’s consequence are separate decisions. A warning counter also records configured violations; it is not a measurement of messages sent per minute.
Test what you want to preserve
Use a separate test group and a consenting ordinary member. Avoid using an administrator as the only test account: their experience can differ from a member’s. Use harmless sample text rather than real scam material, and do not generate a large burst of messages.
Write expected outcomes for a small test set:
- A normal question should remain readable.
- A single accidental duplicate should not cause an unexpected sanction.
- Several relevant clarifications should be possible under the chosen pacing rule.
- An allowed documentation link should remain usable.
- A harmless phrase deliberately added to a test rule should produce the configured outcome.
No automatic response to a duplicate can be the correct result. You have not enabled a dedicated duplicate detector in Defendy. The test is there to reveal what your settings actually do, including where a human still needs to intervene.
Repair mistakes at the right layer
If a legitimate participant is affected, identify whether they are waiting on slow mode, personally restricted, over a warning threshold or losing messages to a filter. Adjust that specific cause. Disabling a filter does not automatically reverse a restriction it already triggered.
For an incorrect mute, verify the member and use Telegram’s member controls or the supported /unmute command. Review any mistaken warning separately. Removing a personal restriction does not remove group-wide restrictions. Defendy does not offer an undo button for a message it deleted; once the rule is corrected, ask the author to repost the useful information if they still have it.
Publish an appeal contact that remains reachable when posting in the group is unavailable. A short explanation and the approximate incident time are usually a better starting point than requiring the member to argue their case publicly.
Review success in terms of useful participation: can someone ask, clarify and receive an answer? A quieter group is only an improvement if members can still use it.
Sources and review date
Reviewed on 5 October 2026 against Telegram’s official documentation and the current Defendy interface and implementation. The scenarios above are proposed checks, not reports of live tests. See the feature catalogue for module status and the text anti-spam reference for its separate controls.
