SFW deadlock; The most serious long-term problem

[hushed burrow]

This can be a sensitive issue for some people in the rankings, but I must speak out for Myshell's future.

What is SFW deadlock?

** The ranking I am talking is based on displayed ranking. For more info about ranking, check my another writing about ranking : https://discord.com/channels/1122227993805336617/1214592655112147014 **

For past 1 month, many NSFW bot has risen and fallen.
The bot in 1st place changed like, (Don't remember) -> Ayumi -> The day after -> Mimi -> Stuck in wall (currently). If we include all 1st~12th place, the variance goes way up.

How about SFW bot?
There were only a minor change. Aria came down to 1st to 8th after being excluded from the mission, Utena / Rumi / Victoria sored after being included in mission, and only minor changes...

This is a critical situation. It really demoralizes whomever wants to make an SFW bot.

1. Is being ranked in top SFW bot rewarding?

-> Definitely yes. look at the picture above.

Rank 1 earning bot in NSFW is similar to 6th earning bot, I am flaire.
Rank 2 earning bot in NSFW is similar to 9th earning bot, Rumi.

It is reasonable to say that top SFW bots will earn a lot.

Of course, we should not mess up Correlation and Causation. I will mention it again.

2. Then why creators don't make SFW bots?

-> Mainly 2 reasons.

A. If it is not ranked on top, It sucks.

Look at the graph. Out the main page, No SFW bots are earning 50+, and even half are under 100.
How about NSFW bots? No bot is ranked under 100.

Even if we make an SFW bot, If we can't manage to go in Top 12, It will not give enough return.

• How about putting the bot on featured? Sadly, featured bot's maximum rank is usually around 60s. MAXIMUM, not average. It was seen in Valentine's day event, and now? most of the featured bots are 100+. This... is not a good return.

B. It is TERRIBLY hard to go in top 12.

As I told you from beginning, for past 1 month, many NSFW bot has risen and fallen, and changed lots of time. However, SFW bots never change. 3 causes I can think of;

• New users tend to use whatever is in front of them.
  People who turned on NSFW is not a user who just joined. At least they pressed this and that. Controversally, people using other categories will not be picky and use whatever the bot is no matter the quality.
  This goes double for Emma and Emma | CEFR A2; This bot is counted both in Original character and Learning category.
• Quest.
  Without having the bot in quest, it is almost impossible to be ranked in the top 9. (though 10~12 changes time to time...)
• Probable cheaters
  I will not speak much since I don't have data. But I can definitely say some bots are cheating for keeping their rank.

3. So, What's your solution?

This is really really hard matter, and I know it can't be solved with only one answer. This is only a type of suggestion I can think of.

A. Not ranked on top, It sucks

-> I think it's a consequence not the cause; All the good bot makers move to NSFW category, whether or not their bot is NSFW or not, and users tend to find good bots from NSFW category only.

Still, we need incentives for low rank bot makers in SFW category. Maybe multiplier can be an answer; but it must be really heavy weighted, and no weights on top bots.

B. It is TERRIBLY hard to go in top 12

Randomly display Original Characters

• Plus : It is the most simple way, and it will judge all the bots within specific rank fair enough.
• Minus : going in to specific rank will be another hurdle. / Putting your bot into higher rank will not be a motivation anymore.

Using weights on Rank : (Time dependent - Longer the bot's precense, lower the rank / Rank dependent - Higher the bot's rank is, reduce point earning)

• Plus : Time dependent can be a long term solution for bot makers. It will prevent any bots from keeping high rank for long time.
• Minus : Bot makers will get stressed to think of new bots everyday, and watch their bots wearing away. (I already experience that kind of stress even without weight...) / Rank dependent weight will take long time to show it's effect, since... you can see the point gap. / It is hard to set the weight. It must be set arbitarily.

⭐ Cut the cheaters immediately ⭐

I know, cutting cheaters will give the cheaters chance to retry cheating. However, for creator, it is really really frustrating experience to see them in main; not because they are cheating, but because of the cheaters my bots can't be displayed. It is taking away opportunity.
That's why, when I mean 'Cut', It does not mean just cutting their points. It means total deletion of the bot.

Also, we can't use delayed punishment after TGE too. We must think of a way to kill them before each season ends.

For the quest...

Don't use anyone's bot in quest. Yes, I know, it might be a reward toward not famous sections. However, putting it in quest is too much...

After putting Rumi into quest, immediately, Rumi earned almost 100k points. seriously, yes, I know I am Rumi's owner, but even for me it was absurd. Rumi the day before was 100+ earning bot.

Then what will be good substitution? Well, I only found out one solution... Buy the bot from the bot maker in return of certain amount of shellcoin or points. (Also, remove them from ranking so more creators can benefit) That way all the points previously came from quest will go back to ecosystem. Remember the Pepe bot... It is not impossible.

(end of essay)
Feedbacks always welcomed!

[still turtle]

[hushed burrow]

Strict separation between each category should be done too.

1. Most of the 'game' category is not used by game bots, rather used with 'Famous character' for game character bots.
2. Some bots in learning sector uses original + learning.
3. Many SFW bots uses NSFW tags... I can definitely say that Ayumi wouldn't make to 1st place if it was NSFW bot, (I heard Ayumi was misclassified due to AI error ) but it is not good situation for SFW category. People need to look up NSFW category to find SFW bot, and NSFW users must test whether the bot is SFW or not by themselves.

To achieve it, maybe giving CV permission to reassign category might be considered...
Also need clear guideline for each category tags. NSFW itself may contain several different criteria. (sexual, violence, criminal...)

[hushed burrow]

I just found out new thing; People who did not check NSFW would be probably seeing almost same page with same bots everyday. This would be another big reason for solving SFW deadlock.

[timber wing]

maybe we need a ‘random’ filter

[icy rose]

[formal saddle]

Good thread, pointing to a clear problem.
As for me, creating NSFW is a personal choice... but I know how hard it is to earn points when you're not in the TOP visibility; and yes, I think winning with bots is a matter of visibility, over quality or time invested (ROI).
The problem is that appearing in the top 12, feeds back into that condition: it's an easy and direct choice for casual users, which will grant those bots more usage, therefore, they will continue to be the top earners.

Until now, developers offered two ways to get visibility for SFW bots:

• Featured bots - after careful selection, participation in events, etc. provide visibility to a bot, allowing more people to interact with it. This is a time-limited action that usually drives bot usage, and after that period comes the fall from grace.
• Rankings - Popularity for usage and Trending and Top Gainers for investment. Still, only trending and top gainers show some churn, however these rankings are not used by users to choose a bot to chat with.

How to improve this? I think @hushed burrow shared some good ideas, and I would like to add one:

• Dynamic Tabs / Marquee lists for SFW and NSFW rankings. Imagine different rankings instead of one, rotating in the same place every 5-7 seconds. If rotation features (like the featured bots big images) is difficult, we can keep it with tabs (like the Dashboard trending - top gainers - etc.). Each tab / list can have different focus:
• Discover. The "random" function that @hushed burrow was commenting.
• Popularity. The one we already know, the top 12.
• Top scored / reviewed. I shared time ago https://discord.com/channels/1122227993805336617/1218162427678687283 aiming partly to this problem. Scores can be more reliable than usage.
• Top creators. This was another thing I was considering, a tab to check creators. This is tricky, tho, because depending on the ranking criteria, it will happen something similar to Popularity ranking.

[hushed burrow]

Hope they can develop these features when dev team has enough time! Currently they are focusing on the new mode too much...

[smoky atlas]

1. Instead of selecting 3 specific bots for each season mission, plan a bot pool in advance and randomly select 3 bots in the pool for each user for mission.

2. The team conducts a preliminary review of each bot. As long as it meets basic quality standards （not blank or garbage promt）, it can enter the bot pool. TBH I don’t think the quality of some low-ranking bots is necessarily worse than the bots that entered the task.

This way, all serious bot makers have an equal chance. Moreover, pool review also encourages makers not to remove their bots too casual (this will make him need to wait again to enter the pool)

[hushed burrow]

I have tons of good ideas for SFW bots... but I need to turn them into NSFW bots I changed isekai travel into NSFW too.
Hope there will be a solution soon.