Fallacy Baseball: A Brief Study in Apologetics, AI, and the Motte-and-Bailey Mind
A Theatre of Control essay.
Some exchanges on social media don’t follow the expected pattern. Every now and then, someone admits something – directly or implicitly – that shifts the entire frame. How you adapt to that shift becomes crucial. What follows is a tale of two exchanges, and the small “game” I use to ensure I am speaking to the person I believe is on the other end, rather than something rehearsed, automated, or inherited.
The Statement That Made Me Pause
‘You’re misunderstanding my point – I just don’t waste time debating people who’ve already decided there’s no God. The Bible says exactly why: ‘The fool hath said in his heart, There is no God’ (Psalm 14:1).
I focus on people who at least acknowledge God but don’t understand who Jesus is… [etc.]’
Before I even entered the conversation, a few things were already clear:
This reply wasn’t directed at me. I hadn’t said a word yet.
A claim that he ‘doesn’t debate’ non-believers, yet he’s speaking to a well-known atheist in the thread – an inconsistency worth noting.
Psalm 14:1 appears as proof-texting and a soft ad hominem: ‘You don’t believe; therefore you are a fool.’
He positions himself as the authority – others ‘don’t understand who Jesus is,’ but he does.
He sets up a protective loop: if the foundation (God exists) is questioned, everything else is ‘pointless.’ This is a form of enclosed reasoning.
Then comes the appeal to authority: Sir Lionel Luckhoo, a successful attorney whose expertise in law is presumed to extend to biblical scholarship.
Finally, the burden is shifted: the atheist must change their worldview before a conversation can happen.
Yet the believer is exempt.
It reads like someone barricaded in the motte-and-bailey argument, protected by doctrinal fortifications but needing to venture out just far enough to proselytise.
These observations, noted before the conversation had even properly begun, revealed a crucial point: I was dealing with a position that was structurally fortified before any evidence had been exchanged. The first move had been to pre-emptively dismiss any foundational challenge, shift the evidential burden, and lean on extraneous authority to defend the perimeter of the argument. My initial reply, therefore, was not aimed at winning a point but simply resetting the field of play.
My Opening Reply
‘Burden of proof is on the claimant regarding the Bible’s reliability.
Appeal to authority: Sir Luckhoo being a successful attorney has no relevance. It’s also cherry-picking a single voice against the consensus of scholars.’
I was interested precisely because of those walls – and whether they could be lowered.
My reply did two small things:
1. Reset the burden of proof. I didn’t want to inherit the baggage of his previous argument with another atheist.
2. Address the appeal to authority.
Luckhoo’s legal brilliance doesn’t grant expertise in biblical languages, textual history, doctrine, or manuscript transmission. Nor does his personal conversion overturn scholarly consensus on New Testament history or textual reliability.
Then I waited.
Time zones have a habit of making you forget these conversations even happened until the notification reappears.
His Reply
‘Appeal to authority? No, that’s your misread. Forget Luckhoo entirely. I’m talking data:
25,000+ manuscripts
99% textual accuracy
Dead Sea Scroll confirmation
Multiple hostile sources confirming Jesus’ life and death
Earliest creed within 3–5 years
If you dismiss the Bible with that level of evidence, then you’d have to throw out Caesar, Tacitus, Homer…
So the burden of proof is met. If someone refuses the evidence because it threatens their worldview, that’s not critical thinking.’
This is where the tone changes – and where the second exchange begins.
Fishing and Fallacy Baseball
As humans we are pattern-recognition machines. Sometimes we notice a pattern without knowing where we’ve seen it before; other times it’s like being hit by a train at full speed. For me, this was the latter.
I’ve spent enough time working with large language models and assessing others’ use of them – AI, if you prefer – to recognise the tell-tale cadence of a certain kind of response: structured, over-confident, rapid-fire, and mechanically comprehensive. And as soon as his reply arrived, I saw the same pattern emerge.
It wasn’t the content; it was the form:
a numbered list of “facts”,
framed as settled consensus,
presented with no textual nuance,
delivered in a single breathless sweep,
designed to overwhelm rather than clarify.
It was the conversational equivalent of loading all the bases and swinging at every pitch. Not one argument, but several stitched together, hoping the sheer volume might compensate for the lack of precision. A gish gallop in a jersey.
I’m not claiming he is an AI, or anything of the sort. The key, rather, is the structural fingerprint of old apologetics redressed in modern framing. Whether a person is consciously reciting a script memorised from apologetics websites, or if they have leveraged a large language model to synthesise a reply, the resulting conversational pattern is functionally identical: it is an accumulative performance of arguments that prioritises volume over genuine engagement. If the reply is pre-fabricated, the debate is hollow. But the behaviour resembled what I’ve repeatedly seen when an LLM tries too hard to ‘cover everything’ without understanding the underlying question. A pattern of argumentation that isn’t responsive – only accumulative.
That’s when the penny dropped. If I wanted an honest exchange, I needed to check whether I was speaking to a person defending a belief or a person defending a script. The difference is subtle but important. One may shift when presented with new information; the other loops.
So I reached for the little test I’ve developed for moments like this – a way to see whether there’s reflective thought beneath the surface, or just a rehearsed pattern. Occasionally, you have to be a little playful with your test.
And with that, Fallacy Baseball began.
It occurred to me that I had to ask the question – not accuse, that would be unproductive. But I needed to be honest, and honesty sometimes comes out with sharper edges than we intend. Even as I wrote that last line, a part of me wondered if I’d pushed a little too far for a first exchange. That hesitation matters, not because I regret saying it, but because it shows the tension between wanting clarity and wanting kindness. It’s a reminder that even in analysis, I’m as prone to missteps and self-doubt as anyone else – and that’s worth bringing the reader into.
So I typed my suspicion.
‘Your reply reads more like a dump of apologetics talking-points than an actual engagement with what I said. The structure, pacing, and the “if you reject this, you must reject all ancient history” plus the closing ‘If someone refuses...’, are exactly the stock lines large language models and pre-packaged scripts use. It doesn’t address the argument. That’s just you being lazy.’
They responded with ‘I’m sorry you feel that way’ and the usual decorative blessings I would fully expect.
But wait – I wasn’t done.
Or rather, they thought I’d blown it.
Or I thought I had.
The interesting thing about online exchanges is that you’re not just debating claims; you’re reading behavioural patterns. And this person had a glaring tell: they always wanted the last word.
Fine.
I can work with that.
My response – risky, but calculated:
‘That’s you throwing out a deflection and not an answer.
When the argument collapses you pivot to “sorry you feel that way” and a blessing.
That isn’t a response – it’s an exit line. Bye.’
Now, you might wonder why I ended with ‘Bye’.
Simple.
People who need the last word will reopen doors you appear to close.
And the rest of the message was the hook.
It worked almost instantly.
They came back with:
‘You called my reply dumb’
and
‘I’m not here to win a debate.’
This was the turn – the tell-tale shift from defensive certainty to wounded self-presentation. And yes, I still had to address the elephant in the room.
My reply:
‘I wasn’t calling your reply dumb. I said it strongly resembles LLM/AI script.
I’m not wasting time arguing with generated apologetics.
So I’ll leave it there.’
And then – right on cue:
‘Ah gotcha. But you don’t have anything to present for which Jesus is not God?’
The ‘Ah, gotcha’ made me laugh. Because that was precisely my reaction when they took the bait. Their desire for the final word reopened the conversation on my terms. And that’s when I introduced the game.
Fallacy Baseball.
It’s simple.
Three fallacies – three clear, identifiable reasoning errors – and you’re out.
Not out of the conversation, necessarily, but out of credibility for that topic.
It forces clarity. It cuts through fog. And, crucially, it sets a standard the other person can’t easily wriggle away from.
My opening pitch:
‘Try this: it’s your claim, not mine. The burden of proof lies with you.
If you can have a conversation without fallacious reasoning, or relying on AI-generated responses (which, as noted, you haven’t denied), then maybe a proper discussion could happen.
Until then…’
Strike one.
And suddenly, for the first time in the entire exchange, the playing field was mine.
Once the conversation reopened on my terms, I introduced the framework I use when things start drifting from argument into performance: Fallacy Baseball.
The first pitch had already landed.
Strike One: The Shifted Burden of Proof
Strike one was fair: burden of proof is not an option; it’s the price of making a claim.
His immediate pivot – ‘you don’t have anything to present for which Jesus is not God’ – was textbook. It reversed the burden of proof, a basic fallacy in argumentation.
He made the claim; the evidential responsibility is his.
I called it gently but firmly:
‘That’s your claim, not mine.’
The first strike wasn’t about winning a point; it was about grounding the conversation in something resembling shared reality. Without that, nothing else matters.
And here’s something essential for the reader:
I wasn’t calling him insincere.
I was calling the pattern what it was – the pattern you see when someone uses apologetic scripts (whether consciously memorised or lifted wholesale from AI-generated content). The rhythm, the phrasing, the structure… they were all pre-fabricated. Humans can use scripts too, but scripts have tells.
And this one had them all.
Strike Two: The Goalpost Shift
His next reply tried a new angle:
‘No matter what evidence is mentioned, you immediately say it “doesn’t meet the burden,”, even though you haven’t defined what level of evidence would ever satisfy you.’
This was two things at once:
A strawman – he attributed a position to me that I hadn’t taken, a logical fallacy where an opponent’s argument is misrepresented or distorted to make it easier to attack.
A shifting of goalposts accusation – one I never committed and a rhetorical tactic where changing the criteria for ‘winning’ an argument after evidence has already been presented to meet the original criteria.
I wasn’t shifting anything; I was holding the line he abandoned. I never said evidence must ‘prove everything in one step.’ I simply said that claims require appropriate support – especially claims as extraordinary as ‘the Abrahamic God is real and the Bible is His literal communication.’
He reframed my position to make it easier to dismiss.
I did clarify this position later in the exchange:
‘If God were real, they’d know what evidence would make me believe. Instead, we have an ambiguously authored book that claims to come from God, and believers all saying they have the true understanding.’
Still – Strike two.
Strike Three: The Pre-Emptive Excusal
Finally, when the argument began to sag under its own weight, he retreated into familiar territory – the existential safety net:
‘If someone refuses to follow the evidence because it threatens their worldview, that’s not critical thinking, that’s just protecting a belief system.’
A neat rhetorical move.
But it was a projection dressed up as insight.
And more importantly, it was the third fallacy:
Mind reading (claiming to know my motives).
Psychological ad hominem (framing disbelief as emotional deficiency).
Immunising strategy (any rejection of his argument is recast as proof of bias).
This was a familiar endgame – the late-inning appeal to motive.
Strike three.
‘You’re out.’
Not insultingly. Not triumphantly. Just structurally. The conversation had revealed its shape.
And here’s the fascinating part:
Once I said it, he immediately shifted tone – softer, spiritual, disengaging. The performance ended, and the person peeked through. That’s the moment I ended the exchange:
‘Enjoy your day.’
Because after fallacy baseball, there’s no need to keep swinging.
Author’s Note
They didn’t just commit three singular fallacies. Fallacies rarely travel alone; they tend to arrive like buses – you wait patiently, and then three turn up at once. The aim wasn’t to catalogue every misstep. This wasn’t about shaming. It was about signalling something important: AI apologetics is simply old apologetics wearing new syntactic clothes. It looks modern because the sentences are tidy, structured, and linguistically polished – but the underlying arguments have been circulating for millennia.
And here’s the irony:
When someone leans on AI-shaped replies, they inherit AI’s flaws along with its fluency. LLMs are trained on us – fallible, emotional, irrational humans – so their apologetics faithfully reproduce the same circular logic and the same fallacious structures. They do not discern. They cannot interrogate. They can only repeat.
My interlocutor claimed discernment as a spiritual virtue, yet his replies bore the exact fingerprints of generated apologetics. Even after I raised it, I still suspect he simply reworded what had already been built by prior scripts, whether from human apologists or AI systems.
The Motte and Bailey
His entire strategy rested on a classic Motte-and-Bailey manoeuvre.
It’s a defensive dance that looks like argument but functions as insulation.
It offers the emotional comfort of bold claims within the safety of modest ones.
A brief explanation for readers:
The ‘motte’ is the safe, defensible claim – vague, modest, unfalsifiable.
‘I’m just sharing why I believe.’
‘Christ changed my life.’
The ‘bailey’ is the bold, expansive claim – assertive, sweeping, evidence-heavy.
‘The resurrection is historically proven.’
‘If you reject this, you reject all ancient history.’
The tactic works like this:
Step out into the risky bailey to make grand assertions; retreat immediately into the motte when challenged.
Back and forth.
Advance and retreat.
Claim and conceal.
In AI apologetics this retreat is almost instantaneous; in human apologetics, it simply feels slower.
Final Thoughts
By the end of the exchange, I did reach the human behind the apologetic scaffolding – a human who still wanted the last word. And that’s fine. Wanting the last word is not a crime; it’s a tell. It means the performance needed a closing bow.
So I let him have it:
‘Thank you! You as well!
Psalm 118:24
“This is the day which the LORD hath made; we will rejoice and be glad in it.”’.
Sometimes the most productive way to end a conversation is simply to recognise that the curtain has fallen – and allow the performance its final line.