Against Arianism

“I need some love like I’ve never needed love before” – Geri, Mel C, Mel B, Victoria, Emma (noted Arianists) 

I spent most of today on a sequence of busses shuttling between cities in Ontario, so I’ve been thinking a lot about fourth century heresies. 

That’s an obvious lie. But I think we all know by now that I love to torture a metaphor. (Never forget Diamanda Galas. This is not a metaphor. Just solid advice.)

But why Arianism specifically? Well it’s an early Christian heresy that posited that Jesus was created by God and thus not the same as God. This view was emphatically rejected by the church leadership and in 325 the First Council of Nicaea wrote this rejection explicitly into the Nicene Creed, which these days reads 

I believe in one Lord Jesus Christ / … / Begotten, not made, consubstantial with the Father

(Yes to all the other Catholics who have lapsed or fallen: a few years back they decided “of one being with the Father” was too understandable. If you don’t know the difference between a lapsed and a fallen Catholic either it doesn’t matter to you, or you’re the former.)

This isn’t even my favourite early Christian heresy: if I ever find a solid reason to use Docetism (short version: Jesus was a hologram) as a metaphor, you better believe I will. (They didn’t need to construct a whole creed to rid themselves of that one.)

Whyyyyyyyyyyy? (extreme Annie Lennox voice)

The first sentence of this post was a lie. I actually spent those five hours thinking about Satan.

I get horrifically motion sick if I try to read anything on a bus, so I used today as an opportunity to catch up on the only podcast I subscribe to. And that podcast is a combination of an Audiobook and an academic discussion of the meaning and context (past and current) of Milton’s Paradise Lost.  It’s done by Anthony Oliveira, and as well as being an excellent performance of the epic poem, the discussion is so much fun you’ll be struggling to stop thinking about Satan. (Anthony has a PhD in this and unlike people with stats PhDs, he has good communication skills. So you get expert knowledge in an accessible package!) It is well worth the $3 per month (at this point $3 gets 21 episodes, so it’s an absolute bargain).

The Perils of Pauline (Theology) 

If I have a point in all of this, I probably should’ve expressed it by now. But I do. I’m just bad a writing. And unsurprisingly it will be a point that sort of I’ve made before. But one I made without such a grouse metaphor.

Like Gaul, Bayesian data analysis is usually divided into three parts: the likelihood, the prior, and … well, depending on the point that I’m going to make I’d either say the data or the computation. But as I don’t care about the third part at the moment, feel free to pick your favourite. (And let’s take it as a sign of personal growth [or Miltonic inspiration] that I’ve swerved into a pre-Christian metaphor even though there’s a perfectly obvious other option.) (Added later: John Ormerod suggested the third part should be the posterior, proving he has a better grip on my metaphors than I do.)

I’ve written before (with Andrew and Michael) about how the prior can (often) only be understood in the context of the likelihood [qualifying adjective added in review], but this is a more realistic metaphor. Because there is not compulsion to speak only of Jesus in the context of God. Instead, they are consubstantial; different but not interchangeable beings made of the same stuff.

But we’ve maybe hit the point of the post where my metaphor falls apart. Because the trinitarian view of god is often stated hierarchically as the father, son, and holy spirit, even if they are co-eternal and of the same substance. And our previous paper was also hierarchical: the prior was only to be understood in the light of the (superior) likelihood.

But reality is more nuanced. If I had to write that title again, I’d say this: The prior is consubstantial with the likelihood. (This is why post-publication revisions shouldn’t be encouraged)

That is not to say that they’re the same. The likelihood typically contains our hypothesized generative mechanism as well as information about how that mechanism was measured. On the other hand, the prior will encode our hypotheses about the constituent parts of both the generative and measurement processes. My point is that while it’s true that the prior should be considered in light of the likelihood, it is equally true that the likelihood  should be considered in light of the prior. 

Bayes from a homoousianism viewpoint. (Rather than a homoiousian one)

So Data Science isn’t so much ability to set a reasonable prior for a given problem (as suggested by a twitter soul with the excellent name “daniel”). Instead, it is the ability to use the same scientific knowledge (substance) to simultaneously build both the prior and the likelihood. 

Work that only considers one of these problems (and I’ve definitely written a bunch of those sorts of papers, including my favourite of my papers) is definitely useful but is incomplete. We must resist Arianism at all times!

The life of the world to come

I want to round out this post with some cultural things. It’s summer, so I’ve been actually enjoying myself, which is to say I’ve been consuming media like something that consumes a lot of media. So here’s some stuff (also – its nice to show that I know how to make a fold, just to point out that all these overbearingly long posts have been on purpose).

TV: (Shows that I watched because Patti LuPone is in them at some point)

  • Steven Universe, which is a cartoon and is very good and very sad and very queer.
  • Penny Dreadful, which is not a cartoon and is very good and not very sad

TV: (Shows that I watched because they’re good enough that Patti LuPone may eventually be in them)

  • Pose. 10s across the board. Fast forward through the white people who are far from essential.
  • Killing Eve. Phoebe Waller-Bridge should write all the things

Films (Recent):

  • Hereditary. Wow. (If this post convinces you of nothing, it should be clear that I love a cult)
  • Sorry To Bother You. Wow! (Also a cult)
  • Mamma Mia: Here We Go Again. WOO! (Also a cult)

Films (Old):

  • Alice Sweet Alice (If you think I don’t like Catholic guilt…)
  • Calvary (If you think I don’t like Irish Catholic guilt…)
  • Torch Song Trilogy (If you think Anne Bancroft as a grieving Jewish widow having a bruising argument with her drag queen son at her husband’s grave is not in my wheelhouse, you’ve not understood me at all)

The post Against Arianism appeared first on Statistical Modeling, Causal Inference, and Social Science.