The verdict

The short version

1. Six rival accounts, one body of evidence.

   Naturalism, deism, other traditions, Judaism, Islam and Christianity are each scored against every line of evidence they make a prediction about.

   The ledger ↓
2. Every line is a likelihood ratio, with a range.

   The low end is the value a well-informed critic would defend; the high end, the strongest defender's. The evidence against sits in the same table.

   The ledger ↓
3. Combined with safeguards.

   Overlapping lines are discounted, and no single line can move the odds more than the chance that its analysis is wrong allows.

   Method →
4. The result depends on where you start, so we show every start.

   From even odds, our central estimates bring a Mind behind nature to 91.2% and Christianity to 68.4%, with wide uncertainty. A committed skeptic who gives any Mind behind nature 1 in 1,000 moves only from 0.01% to 0.12% on Christianity: weighed this way, the evidence barely moves a committed skeptic.

   The verdict ↓
5. The hinge is the resurrection; prophecy no longer helps; the strongest objection is suffering.

   Remove the resurrection and Christianity falls from 68.4% to 2.5% from a neutral start: it carries essentially all of the weight that is specific to Christianity. Prophecy, scored whole, gives no margin, and as the model reads its central values, it counts against. The evidence can be argued about line by line, and you can do exactly that below.

   The calculator ↓

The verdict, stated plainly

Read the last column as the honest form of the disagreement. If you start out more confident than that that there is nothing behind nature, this evidence, weighed this way, will not move you past even odds. If you start anywhere above it, it will. Where you stand is yours to decide; the calculator below lets you find it.

The running total

In odds form, weighing evidence is addition: start from your starting odds, add each line's weight, and you arrive at your result. This chart does exactly that with your current numbers (change them with Your numbers). Each bar is one line after the overlap discount and the error cap; blue moves toward Christianity, orange away. Switch the comparison to see which lines separate Christianity from Judaism, Islam or deism: most of them drop out, the two prophecy lines pull the other way, and the resurrection does the work.

The ledger

Every line of evidence on the site, with its likelihood ratio and range as the research found it. Lines above 1 count for the account named in "What it compares"; lines below 1 count against it. Click any line for the full analysis.

What each line can tell apart

Most evidence bears on only one question. Fine-tuning can tell a Mind from no Mind, but not Christianity from Judaism or Islam. The resurrection is where Christianity parts from its nearest rivals. Messianic prophecy was meant to be the other place, but scored whole it gives no margin, so in practice it no longer separates them. This is why the case has three tiers, and why Tier I alone can never favour any one religion.

Put in your own numbers

Start where you are. The calculator opens on the skeptic's starting point and our central estimates. Change the starting point, swap in the critics' or the defenders' numbers, move any line, switch lines off, or change how much the lines overlap and how often an analysis might be wrong. The results, the break-evens and a ledger of what each line contributes update as you go.

Robustness: what if we're wrong about the method?

A conclusion that holds only under one set of modelling choices is fragile. These checks change the choices, not the evidence:

Three things stand out. First, the case for a Mind needs both ends of the site: without Tier I it falls to 46.5%, below even odds, and without the resurrection to 72.7%. Even granted in full, the resurrection evidence does far less for the existence of God than for the event itself (the resurrection page shows the limit). Second, the resurrection carries essentially all of the weight that separates Christianity from its rivals. That is why its page has the deepest analysis and the strongest critics on the site. Third, the checks that bite are no longer about overlap. Treating each tier as one argument also discounts the evidence against and the prophecy lines that now sit below 1, so it raises the result; counting the evidence against twice, or dropping Tier I, are the stress tests that matter. Reading the two prophecy lines as flat rather than as evidence against would raise Christianity, from even odds, to about 93%.²

Rival traditions

The case is only fair if the evidence offered for other traditions gets the same rules. It does, on its own page, and those rows sit in the ledger above. In short: the evidence offered for Islam (the Quran's early text, its literary challenge, a claimed prophecy in 30:2-4) comes to about 2.4, and the Quran's own hard texts count against it by about the same factor the Bible's hard texts count against Christianity (0.41 against 0.42 by an identical rule). The historical record of Jesus' crucifixion counts against Islam's denial of it, modestly once the resurrection line is in the model. The evidence offered for the other traditions, driven mainly by children's past-life reports, comes to about 2.3 as a composite. Judaism's strongest case against Jesus, that the messianic age has not come, is real. It is scored inside the messianic prophecy line, where it now decides the result (Jesus misses all 19 military and political features the pre-Christian sources expected), so it is not entered twice. The Torah's "perpetual" statutes count a little against Christianity (0.83), the only row on that page that favours Judaism over Christianity (messianic prophecy, scored whole, also sits below 1).

The result: on the central numbers and the neutral settings, Christianity's advantage over Judaism rests entirely on one line, the resurrection; without it, the other lines leave Christianity behind Judaism by a factor of about 2. Over Islam, the resurrection is about three-fifths of the edge; the Gospels' reliability and the crucifixion record supply most of the rest.²

What would change our mind

A hypothesis that nothing could count against is not worth testing. These observations would move the verdict most:

• Tier I: a demonstrated unguided route from chemistry to a self-sustaining, evolving replicator; or a physical theory that fixes the cosmological constant's value without a landscape of universes or tuning of its own.
• Tier II: for prophecy to count for the case again, an independent human replication of the blind prophecy test, or a measured rate of embellished fulfilment claims below about 1 in 25; against it, evidence that the hits were written after their events, or a pre-Christian manuscript showing Isaiah 53 or Daniel 9 edited toward Jesus.
• Tier III: evidence that the 1 Corinthians 15 list of witnesses is a late addition; a credible natural explanation of the group appearances that fits the early sources better than the ones we score; or, most simply, the body.
• Against: a theodicy-free account of animal suffering over deep time that a good God would plainly prefer is the strongest single item on the other side; we score it as such.

Objections to the method

What we don't claim

• "The probability that God exists is X%." Every percentage on this page is conditional on a starting point you can change.
• "The case is proven." It is a cumulative, probabilistic case that a reasonable skeptic can still resist, and the break-even tells you exactly how.
• "No rival has any viability." On the critics' numbers naturalism leads from every starting point in the table, and from the skeptic's start it keeps 99.54% even on our central numbers.
• "The odds against chance are 1 in 10^X." No line in this ledger is allowed to carry such a factor.
• "Fulfilled prophecy confirms it." Scored whole, with the misses, both prophecy lines give no margin.
• "The resurrection proves God." It is evidence for the event. What it adds to the case for God is limited by how likely God was to do this, and in the model it raises the odds of a Mind about fourfold from even odds.²
• "The philosophical arguments settle which God." The best of them, at most, support a generic theism; they do not choose among Judaism, Christianity and Islam, or against deism. See the method page.

Sources

1. The model: model.py and its inputs model-inputs.json. Each line's range and reasoning is on its own page, with its sources.
2. Figures the model does not print (Christianity with the prophecy lines read as flat; the resurrection's effect on the odds of a Mind; each line's share of Christianity's edge over its rivals): truth/research/calc/14-lt-checks.py and its saved output, which call the model's own functions. The comparison with the model's earlier run is in truth/research/14-integration.md, §4.
3. The method and its literature (Good, Jeffreys, Kish, Swinburne, Draper, the McGrews, Sober, Collins, Hume, Earman): see the Method page's sources.