It has come to my attention that mods can see the global data of how many times an user has been banned or timed out... And I can see how that can be easily abused or misleading, and I advise keeping those stats ONLY to staff, and ONLY take measures on those metrics on confirmed reports of spammers or clearly malicious individuals.

Examples of cases:

• Mob Mentality - Some groups of streamers may gang up on a target for petty reasons and ban them even if they have not been on their stream at all. Example: Viewer A has been banned by Streamer A for reasons (that may be or not valid), Streamer B, C, D, E, F, all of which are mods for Streamer A, bans Viewer A in their own channels, but some of those (let's say Streamer E and F) banned without knowing the context. That alone got 6 bans on Viewer A when it should have been 4 at most, and maybe what Viewer A did ONLY applies to the rules in Streamer A's channel, but not on B, C, D, E nor F.

• "Just a joke" - Sometimes a streamer or mod bans (and unbans immediately after) or times out on a joke. I have seen this happen a lot, and is never of ill intent, it's always an inside joke in the community... But imagine what other mods will see when that user goes to another channel. They'll get the wrong impression... Example: Streamer A goes to Streamer B's channel and makes a silly joke, to which Streamer A asks his mods to ban him, and his mods are in on it that Streamer A and B are good with each other, so they ban and unban him immediately. Alternatively, asks to time them out for ten minutes. It will add strikes against Streamer A for... An inside joke. Of the group he is part of.

• Accident - Oh my... How many times I've seen this happen... A streamer is, of course, busy focusing with his game; a mod is on alert because of bots and the like, and accidentally misclicks the mod tools and times out someone for ten minutes (and apologizes and rectifies that mistake immediately). But the damage is done though... One particular case involved a mod banning the viewer instead of a bot because he was trying to ban a bot but mistook the name when entering the command.

• Joke Redeems and Commands - I've seen people set joke redeems or commands that involves the viewer willingly asking to be timed out. Although it's more often than not used for when the viewer slipped a backseat or spoiler and wants to erase their last message (particularly with the command "!seppuku" that some bots have, that allows the user to request the bot to time them out for 1 second and clear chat messages)

• Misbehaving Moderator - Sometimes a moderator may go rogue because of someone they have a vendetta with, and they may more aggressively moderate a viewer and even be called out by the streamer a number of times to "take it easy" and "it is alright, they slipped", or "only a purge is enough". Sometimes the streamer may have enough and demote or ban the mod, but sometimes they keep the mod around because they are accustomed to them or they have no one that they can trust to take the mod position at that time.

In other words, as is, Twitch created a sort of echo chamber between moderators that allows for spread of misinformation about a viewer's behavior by neither allowing the viewer check AND contest those stats, nor having a context behind the actions that resulted in those bans and timeouts being recorded in a global manner. And even if context is provided, how can one tell the difference between "viewer was being a **" (serious) and "viewer was being a **" (joke)?

So yes, those metrics are harmful as is. Perhaps could have those metrics go down over time if the viewer behaves elsewhere, and not allow a ban to be lifted and reapplied to extend the metric through an exploit. Like let's say for example: Streamer A bans Viewer A, Viewer A has that strike against them for a month (the ban stays indefinite), after a month that specific ban's strike wears off, Streamer A is aware of the strike wearing off and then unbans and bans Viewer A to try to renew/extend the strike before the month wears off, but it doesn't extend because Viewer A hasn't interacted in Streamer A's channel since, so Streamer A has "no valid reason to renew the strike".
For Timeouts, that strike should wear off faster (based on the duration of the timeout).