What the scores tell us about arguments, debate roles and recurring speakers.
Try changing the scoring weights below, then explore 10 studies of why scores differ, how arguments and replies work, and whether speakers tend to perform similarly across different debates. Each study explains its finding, how to read its figure, and what remains uncertain. The linked papers and methods page provide the details.
Research snapshot: October 9, 2026 · 308 assessments. The October edition adds 55 assessments to the September research and recalculates the score comparisons. The slogan counts still come from the September 5 AI review of 187 transcripts; no newer transcripts were reviewed for slogans. The debates were selected for the site, not sampled at random, so they are not representative of all public debate. Scores assess the arguments presented; they do not establish whether a worldview is true. Slugfester’s assessment reviewers and adjudicators are independent AI agents called by the administrative AI. Independence means separate review contexts with controlled inputs; the agents may use the same model and share biases. Review arrangements vary across assessment periods.
What changed since September: the average skeptical-side lead is smaller. People who have taken both debate roles show a small average advantage when opposing the stated claim, although its strength depends on how their records are counted. The ranking study now includes 57 people with at least three assessed debates each. The research update did not change any earlier debate scores.
Decide how much each scoring area should count, then see what happens to the pro-God and con-God sides. The sliders start at Slugfester’s current weights. This experiment changes the calculation—not the underlying assessments or published scorecards.
234 reviewed debates · October 9, 2026 research snapshot. “Pro-God” means the side defending God, a creator or a religious claim; “con-God” means the side challenging it. This includes debates about religion, not only God’s existence. The labels follow the position argued, not the speaker’s personal beliefs or the scorecard’s PRO/CON order.
The default all-topic results are shown. Enable JavaScript to adjust the sliders or filters, or to restore shared settings.
Black marks show the averages with the current site weights.
Con-God ahead by 5.74 points on average.
The gap is unchanged.
33 Pro-God ahead
193 Con-God ahead
8 Tied
0 of 234 debates change which side leads or whether the scores are tied.
This tests how the recorded scores respond to different weights. It does not ask the AI to judge the arguments again, establish which worldview is true, or prove the judgments are unbiased.
See individual debate resultsHow this calculation works
Each argument or reply keeps its six recorded dimension scores. Your weights produce a new whole-number move score. We then apply the original argument-importance weights, round each section score, apply the original section weights and final burden-completion adjustment, and round the debate-side score. Scores stay between 0 and 100. Each debate contributes equally to the averages above.
The default calculation reproduces every published side score in this set. A changed result means a switch between pro-God ahead, con-God ahead and tied. Small slider changes may leave a score unchanged because the original calculation rounds at several steps. Even a 100% setting retains the original argument weights, section weights and final adjustments; it is an extreme scenario, not a recommended rubric.
The comparison uses the same reviewed religious-versus-skeptical set as the first Insights study, excluding team debates and cases without a clear contrasting position. New assessments are not added automatically until their inclusion and side labels have been reviewed for a research update.
Topic and procedure filters apply together to the averages, original-weight baselines, counts and individual results. Each debate belongs to one research topic group. These are the eight broad groups used by the papers, not the finer categories used to browse the site. Selections with fewer than ten debates carry a small-selection reminder; ten is a display threshold, not a test of statistical reliability.
Shared links store the exact whole-percentage weights, both filters and the research-edition identifier. They contain no personal details. Unsupported settings or a different edition produce a visible warning and restore the defaults, rather than silently presenting a different comparison.
Skeptical sides still score higher on average, but the recently added debates have a smaller gap.
5.74 pointsAverage skeptical-side lead on the 100-point overall score · 234 debates
Across 234 debates comparing a religious claim with a skeptical challenge, skeptical sides average 5.74 points higher. They score higher in 193 debates, religious sides in 33, and 8 are tied. The September average gap was 6.34 points. The 47 debates added since then have a smaller average gap of 3.34 points, lowering the combined average.
The largest difference among the six scoring areas is in Support: how well evidence and reasons justify a claim. Support, logical reasoning and replies to objections together account for 74.6% of the overall point gap. This identifies where the points differ, not why the speakers performed differently. Religious sides do sometimes score higher, as in Meyer–Atkins on design. The study also distinguishes an argument for a creator from an argument for a particular religious God. A separate check excluding debates about religion’s cultural benefits and selected historical or doctrinal questions still finds a 5.77-point gap.
What this cannot establish. These scores judge the arguments presented in selected debates; they do not establish whether a religion or worldview is true. The proposed link between faith and weaker evidence is an explanation to investigate, not a demonstrated psychological cause. Repeating calculations with the same records cannot remove the AI’s judging preferences or the influence of speakers who appear many times.
Reading the figure. Each dot is the average skeptical-side score minus the religious-side score in one scoring area. A dot to the right of zero means the skeptical side scores higher; zero means equal averages. Each line covers the middle 95% of results from 20,000 samples drawn from the saved debates, allowing the same debate to be selected more than once. Longer lines mean the estimate changes more with the mix of debates. The six area differences have not yet been multiplied by their weights in the overall score, so do not add them together. These lines do not account for every possible judging error.
Skeptical sides average higher scores in all eight research topic groups, but the size of the gap varies.
7.28 vs 3.04Average skeptical-side leads, in points: religion, culture and meaning vs resurrection and history
The largest average gap is 7.28 points in the 25 debates about religion, culture and meaning. The smallest is 3.04 points in the 24 debates about resurrection and historical evidence. However, when the analysis repeatedly draws new samples from the recorded debates, religion, culture and meaning has the largest gap only about 32% of the time. It leads in the current records, but another topic often leads when the mix of debates changes.
The paper examines the extra reasoning needed to get from a religion being useful to its being true, from a cause of the universe to a religious God, or from sincere testimony to a miracle actually occurring. The skeptical-side advantage in evidence and support appears both in arguments presenting a case and in replies to objections. The study retains September’s eight topic groups so the two editions can be compared; the website’s browsing categories are more detailed.
What this cannot establish. These results do not show that the topic itself causes a larger or smaller gap. Different topics include different speakers, opponents and questions. The estimates are not precise enough to establish a reliable ordering of all eight topics from largest to smallest gap.
Both sides must contain the stated move kind; equal debate weights
Reading the figure. This figure compares evidence-and-support scores for individual arguments and replies, not overall debate scores. Each dot shows how far the skeptical side leads on average for the stated kind of move. Both sides must contain that kind to enter the comparison. Each line covers the middle 95% of results when the recorded debates are sampled 20,000 times, allowing repeats. A line entirely right of zero keeps the skeptical advantage throughout that range. The rows overlap, so their counts must not be added.
Slogans used to block criticism were detected more often on religious sides in the September transcript review.
0.56 vs 0.16Criticism-blocking slogan uses per 10,000 words · religious vs skeptical sides · 187 transcripts
A ‘protected slogan’ means a phrase used instead of a supporting reason and also used to shut down criticism. The September 5 review detected 77 such uses on religious sides and 19 on skeptical sides. After allowing for how much each side spoke and giving each debate equal weight, the rates are 0.56 and 0.16 per 10,000 words. Most debates—144 of 187—had none detected on either side. For the broader category of slogans lacking support, without necessarily blocking criticism, the evidence does not clearly establish a difference between the sides.
A separate check of 234 current scorecards finds that 33.87% of religious-side moves and 13.40% of skeptical-side moves score below all three thresholds: 70 for Support, 80 for Clarity and 80 for Care. Each debate contributes equally to those percentages. A move is an individual argument or reply; Care measures whether confidence fits the evidence and the opponent is represented fairly. Low scores in these areas can flag weak argumentation, but they do not identify a slogan. No additional transcripts were reviewed for slogans in the October update.
What this cannot establish. Each transcript received one AI review, not two independent AI reviews. Transcription errors, missed examples and consistent AI misreadings could affect the counts. The slogan finding applies to the 187 September transcripts; the newer scorecard check cannot establish how often slogans occur in the added debates.
187 previously reviewed transcripts, not the expanded archive
Reading the figure. Rust bars represent religious sides; teal bars represent skeptical sides. The heights show slogan uses per 10,000 words, first calculated within each debate and then averaged with each debate counting equally. Both panels use the same scale and start at zero. ‘Protected’ slogans are the criticism-blocking subset of unsupported slogans: do not add the two panels’ counts. All direct slogan counts come from the September review, not the expanded October archive.
The side opposing the debate’s stated claim scores higher on average, but this does not prove that opposing is inherently easier.
4.44 → 1.33CON’s average lead in points: all comparable debates → comparing the same people in both roles
PRO supports the debate’s stated claim; CON opposes it. Neither label necessarily means religious or skeptical. Across 291 comparable debates, CON averages 4.44 points higher. Among the 37 people who have taken both roles, their CON scores average 1.33 points higher when people with more appearances in both roles count more heavily. Repeatedly sampling these speakers gives a middle-95% range of 0.32–2.30 points: the advantage stays above zero throughout that range.
If each of those 37 people counts equally instead, the average CON lead is 1.09 points, and its range narrowly includes no advantage. Another check gives equal weight to debates where the religious side is PRO and debates where it is CON; that gap is 1.73 points. These checks show that the apparent advantage depends partly on who fills each role and how the records are counted. They do not measure how much of the gap is caused by the role itself.
What this cannot establish. Speakers were not randomly assigned to roles, and their topics and opponents change between debates. The smaller gap among people who took both roles is evidence of an association, not proof that taking CON raises a person’s score. These results do not justify automatically adding or subtracting points for either role.
Debate draws for first two rows; speaker draws for last two
Reading the figure. Each dot is an average CON-minus-PRO score. Right of zero favors CON; left favors PRO. The rows use different comparisons: all debates, equal weighting of the two religious-side role assignments, and the same speakers compared across roles with equal or unequal speaker weights. Lines show the middle 95% of 20,000 repeated samples. The first two rows sample debates; the last two sample speakers. The weighted same-speaker line stays above zero, but none of these checks randomly assigned people to roles.
A side can score lower without a named fallacy—and score higher despite having one.
54.4%Lower-scoring sides without a named-fallacy tag · 161 / 296
In 161 of the 296 assessments with unequal scores, the lower-scoring side has no named logical-fallacy label. That is 54.4%, down from 61.7% in September. The rate differs sharply between assessments made with earlier procedures (80.8%) and later procedures (17.8%). This percentage therefore depends on which assessment procedures are represented, not just on the arguments in the debates.
Of 158 lower-scoring sides with no fallacy label and fully comparable scoring records, 118 score lower in five or all six scoring areas. Their weaknesses extend beyond a missing fallacy label. Conversely, 75 higher-scoring sides do have a fallacy label. To understand a score, examine the argument, the criticism and their importance to the debate; counting labels alone is not enough.
What this cannot establish. A missing label does not prove an argument contains no fallacy. The labels are part of the AI assessments, not a separate check for every possible error. The counts refer to assessments: assessments 13 and 125 cover the same source video in different formats.
Reading the figure. Each fraction counts lower-scoring sides with no named-fallacy label, divided by all assessments with unequal scores in that group. The bars show the corresponding percentages. The dark overall bar includes the other groups; it is not an additional set of debates. A missing label does not mean an argument is error-free. These are assessment counts, including two assessments of one shared video.
Later assessment procedures produce lower average scores; we have not established whether they judge the same performance more strictly.
81.32 vs 78.98Earlier versus later average side scores · 179 / 112 debates
The 179 debates assessed with earlier procedures have an average side score of 81.32; the 112 assessed with later procedures average 78.98. The difference is 2.35 points, smaller than the 2.82-point difference in September. Of 71 people appearing in both groups, 56 average lower scores in their later-procedure debates, 13 average higher scores, and 2 are unchanged. How often fallacies and biases are labeled, and which individual score values are used, also differs.
Recalculating from the saved assessment records reproduces all 7,032 comparable argument-and-reply scores and all 582 overall side scores. The arithmetic is correct. That does not establish that an 80 under one procedure represents the same quality as an 80 under another. A stronger test would have both procedures assess the same debates and then publish where their judgments differ.
What this cannot establish. People appearing in both groups gave different performances on different topics against different opponents. Their lower later averages could reflect those differences, stricter judging, or both. A newer AI model is not automatically more accurate. The current evidence does not justify raising later scores simply to match the earlier average.
Reading the figure. Each dot represents one person who appears in debates assessed by both procedures. Read their earlier-procedure average along the bottom axis and their later-procedure average along the left axis. A dot below the diagonal means that person’s later-procedure average is lower. The axes show only the range occupied by these scores, not the full 0–100 scale. These are different debates by the same person—not two judgments of the same performance.
Do Slugfester Rankings Measure Stable Performance?
Yes—speakers who rank higher using one half of their assessed debates generally also rank higher using the other half.
0.87Typical agreement between two rankings built from separate halves of each person’s debates · 1.00 means identical order
For the 31 people with at least six assessed debates, the study randomly divides each person’s debates into two halves, averages their scores in each half, and builds two rankings. It repeats this 3,000 times. The typical agreement between the two rankings is 0.87 on a scale where 1.00 means identical order and 0 means no agreement in rank order. The middle 95% of the results run from 0.81 to 0.92. This is strong evidence that the differences between higher- and lower-scoring speakers recur across their recorded debates.
The result also holds when each person’s debates are split by assessment number instead of at random: ranking agreement is 0.84. In a separate check, subtracting the average score for each assessment procedure from scores produced by that procedure leaves the full 57-person ranking almost unchanged (0.99 agreement). Readers can therefore take repeated differences between higher- and lower-scoring speakers seriously. The strongest direct split-half evidence comes from the 31 people with six or more assessed debates. It does not establish that someone in fifth place reliably outperforms someone in sixth: neighboring averages are typically just 0.20 points apart.
What this cannot establish. An agreement of 0.87 does not mean the rankings are 87% accurate or predict an 87% chance of winning. These tests concern the saved scores, not future debates. The full research ranking contains 57 people with at least three assessed debates each (416 appearances in total). They face different topics and opponents, and a judging system can repeat its preferences consistently without judging correctly.
Highest 15 means shown; all 57 eligible people included in calculation
Reading the figure. The figure shows the 15 people with the highest averages, but calculates ranks among all 57 eligible people. Dark dots mark their places using all their recorded scores. Solid teal lines cover the middle 95% of places obtained when each person’s recorded scores are sampled repeatedly. Dashed rust lines use a model that allows score variation estimated from the whole group, including for speakers whose few recorded scores are identical. Numbers in parentheses count assessed debates. Overlapping lines mean exact places are uncertain; they do not overturn the separate finding that higher- and lower-scoring speakers usually remain ordered similarly when their records are split in half.
A useful reply changes the support for the claim actually being made—not merely a nearby claim.
58.2%Of recorded reply-to-target links lead to another reply · 112 debates
The later-process assessments contain 2,444 replies and 2,978 links to the moves they address. Most links—1,733—lead to another reply, so following the exchange matters. All stored targets exist, and their excerpts start no later than the replies. These checks show that the recorded exchange can be followed; they do not establish that every reply succeeds.
Eight worked examples show the difference between challenging a premise, exposing a missing inference, clarifying what was claimed, conceding a point, and answering a position the opponent did not state. Meyer distinguishes physical order from functional DNA sequences; Craig identifies a disputed inference from foreknowledge to necessity; Goff acknowledges a clarified objection. Each local success leaves a clearly identified question open. The paper also examines inaccurate attribution and incomplete responses.
What this cannot establish. The eight cases were deliberately selected to explain different reasoning moves, not to measure how often each side replies well. Existing scores are context, not independent confirmation of the analysis. The close readings use saved caption excerpts, not a new full-video review.
112 later-process comparable debates; links, not unique replies.
Reading the figure. Bar length counts recorded links. A reply can have more than one target. Teal means its target is another reply; gold means a constructive argument. Both bars start at zero. These categories describe the stored links, not whether a reply succeeds.
From a Creator to a Personal God: Which Steps Are Defended?
An argument for a cause does not automatically establish a mind, perfect goodness or a particular religion.
8 passagesSix debates examining what supports each additional claim
The case studies separate a necessary foundation or cause from intelligence, purposes, unlimited attributes and religious identification. Several speakers explicitly defend the extra steps rather than simply assuming them. Craig offers an argument for personal agency; Jones argues for unlimited perfection and explores a route to the Trinity; Horn adds historical evidence for Christianity. Each proposed step has its own supporting reasons and unresolved questions.
The paper also includes deliberate limits. Meyer distinguishes an inference from biological data from an argument based on scripture; even a successful biological-design argument would not alone establish a cosmic creator. Goff gives up unlimited divine power in response to suffering, which creates a new question about what limits that power. The practical rule applies to both sides: credit the conclusion actually supported, and do not treat a local criticism as a disproof of every possible creator.
What this cannot establish. These eight passages are a deliberately varied case study, not a census of creator arguments or a count of how often religious speakers overreach. The diagram distinguishes claims; it is not a compulsory sequence or a proof. Speaker identity and personal religion do not determine how an argument is classified.
An explanatory framework illustrated by eight passages in six debates.
Reading the figure. The numbered boxes identify distinct claims and the question needed to support each. They are not measured frequencies, proof steps, or a ranking of religions. A successful argument for one claim does not automatically establish the others.
How Much Do Opponents Affect Recorded Performance?
The small opponent-score association is not stable when the eligible comparison set changes.
−0.49 → +0.06Fitted own-score change for a five-point higher opponent average · main and same-procedure checks
The study compares a person's own scores against opponents with different averages elsewhere, excluding every meeting of that pair from the opponent average. The main model uses 295 appearances and accounts for the speaker, assessment procedure and PRO/CON role. A five-point higher opponent average is associated with a 0.49-point lower own score. Requiring same-procedure opponent experience leaves 180 appearances and gives a 0.06-point increase. On those same 180 records, the original opponent measure already gives a near-zero 0.02-point decrease. The changed sample, not just the changed measure, matters.
Among 21 pairs who meet more than once, only two switch which person scores higher; both comparisons cross assessment procedures. None of the 12 same-procedure repeat groups has a strict reversal. The typical range of margins is three points across all repeated pairs and two within the same procedure. This supports stability within those observed matchups, but not an automatic opponent-adjusted ranking. The paper shows eight model checks and distinguishes score changes from changes in the margin between speakers.
What this cannot establish. Opponents were not assigned at random. Their outside averages also reflect topics, schedules and judging conditions. The checks describe associations, not causal effects or advance predictions. The spread between models is not a confidence interval; the same-procedure check changes both the proxy and the eligible sample.
The opponent association depends on the comparison set
Eight descriptive specifications; eligibility changes in several checks.
Reading the figure. Each dot is a fitted association, not a causal effect or uncertainty interval. Negative values mean a lower own score with a higher opponent average elsewhere. All meetings of the pair are excluded from that opponent average. The red final row also requires those outside scores to use the same procedure; its eligible sample is smaller. There are no error bars because sampling uncertainty is not estimated.
