"What AI Suggests for TMP Polls" Topic

I asked AI to give me five suggestions for polls on TMP. (Despite discussing TMP previously, I had to remind the AI what TMP was.)

Since **The Miniatures Page (TMP)** is a community for wargaming and miniature hobbyists, the polls should lean into the "crunchy" details of the hobby – rules, scales, and the ever-growing "mountain of plastic."

Here are five poll suggestions tailored for **TMP**:

### 1. The Scale Debate
> **"What is your 'Goldilocks' scale for historical gaming?"**
* **Options:** * 6mm (Massive battles, tiny footprint)
* 15mm (The classic club standard)
* 28mm (The detail-seeker's choice)
* 54mm+ (Skirmish or bust)

### 2. The Rulebook Philosophy
> **"When starting a new period, what determines your choice of rules?"**
* **Options:** * Historical Accuracy/Realism (The 'Simulationist')
* Ease of Play/Speed (The 'Gameist')
* What the local club is already playing
* I usually just write my own house rules

### 3. The "Lead Mountain" Confessional
> **"What percentage of your current collection is actually painted?"**
* **Options:** * 0–25% (I'm a collector, not a painter)
* 26–50% (Making steady progress)
* 51–75% (The mountain is shrinking!)
* 76–100% (I am a hobbyist god)

### 4. Period Popularity
> **"If you could only wargame in one era for the rest of your life, which would it be?"**
* **Options:** * Ancients & Medieval
* Horse & Musket (Napoleonic/ACW)
* World Wars (Bolt Action/Chain of Command)
* Sci-Fi/Fantasy

### 5. The Basing Struggle
> **"Individual basing or multi-figure stands?"**
* **Why it works:** This is a perennial hot topic on TMP. It touches on everything from aesthetic preference to which rule systems a player can easily swap between.

Anyone surprised that AI knows so much about TMP?

AI seems to know a lot, but evidently nothing new.

AI seems to know a lot, but evidently nothing new.

AI, most likely in this case an LLM, doesn't know anything.

It mathematically synthesizes responses from (1) a large corpora of existing knowledge, and (2) a constraint (the "prompt").

The default on most systems is return pseudo-random results from the clusters with the highest correlations to the input constraint. So, in this case, it's not surprising it listed topics that have been posed multiple times in multiple ways here.

What that indicates is a strong correlation between our aggregate behaviour over time and the data in the corpora.

If you wanted "new" or "innovative" poll suggestions, you would need to engineer the prompt toward that.

I think there are two basic tacks:

(1) ask for the topics low frequency discussions that have a wide base of different users and are amenable to a multiple choice framing.

Identify discussion themes that:
* Have low historical frequency in wargaming forums (i.e., not commonly recurring topics)
* Can be meaningfully answered by both casual and expert players
*Can be expressed as a multiple-choice poll with 4–6 distinct, non-overlapping options
Avoid:
* Common topics (rules complexity, scale, realism vs playability, favorite periods, etc.)
* Topics requiring niche technical knowledge
For each suggestion:
* Explain why it is likely under-discussed
* Provide a ready-to-use poll question and answer options
* Indicate what kind of player diversity it might reveal

(2) ask for non-wargaming discussions that have a broad social base and an intersection with wargming topics.

Generate poll topics at the intersection of wargaming and the following domains:
* Cognitive psychology
* User experience / interface design
* Narrative storytelling
* Group decision-making and social dynamics
Requirements:
* The topic must be understandable without formal training in the external domain
* It must connect clearly back to player behavior or game design in wargaming
* It must be suitable for a multiple-choice poll (4–6 options)
Avoid:
* Traditional wargaming debates (realism vs abstraction, scale, rules complexity)
For each:
* Provide the poll question and answer options
* Briefly explain the cross-domain insight it explores

AI slop questions would be a great way to kill the polls for me.

If you go this route, might as well just change this to TMAIP.

The fact that you had to remind your AI what TMP was makes me wonder if perhaps your AI is getting old and senile. Please ask it what we can do to let it know that many of us are also aging and we sympathize.

I did like the four descriptions of different scales, and I wish it had provided descriptions for 10 mm, 20 mm, and 40 mm.

Grelber

Chat GPT did a good job of remembering everything I had told it. GROK, also, now seems to be up on that.

But yes, of course, an AI is NOT creative, so will tell you what "you" already know. Ah, but who is "you"? In the case of TMP, not even THE EDITOR knows all that is there, because he is human and there is too much over too long a period.

What an AI does is look for and spot patterns, and sometimes it sees a pattern that I had missed. The data was there, but my human brain skipped over it. Because, among other things, I have imagination, and desires, and distastes, all things that can warp my examination of phenomena. AI has none of those.

I think we've had all those polls already.

We could let AI ask – and answer – the polls and just cut the TMPers out of the loop.

Looking at some of the threads, maybe "we" already have…

AI slop questions would be a great way to kill the polls for me.

We have plenty of poll suggestions in the queue, so we don't need to ask AI for more.