That's definitely a priority of mine. In a given day I'll usually get to read most all of the threads that hit the main page here. I don't comment in most of them because the community generally does a great job of answering questions for each other, but I do weigh in whenever I think there's something new or useful I can safely say.

Yes to both. It's almost always "hidden errata", or what we call a substitution. Just the parsed text is changed, the displayed text remains the same. [In Oketra's Name] is my favorite example of a card we subbed, to change "other creatures" to "non-zombie creatures".#wotc_staff

We agree; this is a great guide. For handy reference, it’s linked in the “Learn More” page in-game (the little ‘?’ In the top-right of the home screen).

Many of the things you want to know are from before my time (Arena's was quietly in development for quite a while!). It's hard to really estimate how much work this solution saves per set - after all, we've never done a set for Arena without using this system. As a general rule of thumb, a set has around 75% of the cards in it work right away with no need for developer work, around 10% work with less than a couple hours of work, 10% need around a day, and 5% are really hairy. #wotc_staff

As others have noticed, group-assignment of attackers (including All Attack when there is a planeswalker available to be attacked) and of blockers is not yet supported in scenarios involving creatures with combat stipulations. In these cases, the client submits the group to the server, but the server rejects the submission and prompts the client to try again. This then allows you to submit the same result one-by-one.

The intent is for group-assignments to work in all scenarios; we are working on a fix.

Neural nets are a bad fit for the problem because they're pretty "noisy" - they have lots of potential to make pretty much undiagnosable mistakes. We have a pretty classic-AI approach - a spelled-out grammar and dictionary that covers all of the rules texts we support, and a system that takes the resulting syntax trees, gleans their semantics, resolves anaphora (e.g. what does "it" mean?), and compiles the resulting ambiguity-free instruction flow into the final code. #wotc_staff

Heh. Chris is 6'4" (or more), which makes him about a foot taller than me. Me literally trying to fill his shoes would be a pretty humorous affair. On the metaphorical side, I'll definitely do my best to continue the great work he started, both with the game and with community interaction.

You are correct on both fronts. :D #wotc_staff

The parser. I like to describe my job as teaching a computer how to read M:tG cards. #wotc_staff

It's more about what "the chosen creature" means. It's actually lies, as rule 700.7 quoted by OP indicates. It really means "the chosen game object", but that reads worse in English. The problem is that the Arena rules-parser is using that English to write the ability's code, so... #wotc_staff

Ability texts are localized as entire strings, so modifying costs in those texts is non-trivial. We also don't want to raise the expectation that we'll do other such text modifications. Adding information is much simpler than changing it for us. #wotc_staff

The crown represents a generic "numeric aid" - something on the card has a variable that we've identified as worth communicating to the player, but it doesn't match some particular mechanic (like, say, Undergrowth). We use this "crown badge" to represent that number. #wotc_staff

Thanks for the report - I've confirmed that the Indestructible-granting layered effect is indeed constraining its recipient to be a creature incorrectly. I'll fix this up today. #wotc_staff