But sometimes the last corridor train of the day leaves not long after a particular LD train gets in (thinking specifically of the CZ's mid-afternoon scheduled arrival). Sometimes the corridor train is too early to be safe but the LD train that leaves later is usually safe. If your goal is to eliminate situations in which people miss the last available train out of Chicago because their arriving LD is late -- not necessarily an bad idea -- why not look at the actual trains, not general categories? It's not a hard information problem and Amtrak has all the data.
Making sure that no one misses a connection when their train is 12-14 hours late seems a bit extreme. How often is any train 12-14 hours late outside of blizzards, when everyone understands that transportation will have big delays? Are there really that many derailments etc. that block up the tracks? My sense of the LDs is that they are on time at the final station 50-75% of the time and that most of the delays are 1-4 hours. Trying to stop people from depending on connections within those few hours seems a lot more reasonable than essentially planning the schedule to suit blizzards and major freight wrecks.
Finally, you've said two things that are a little contradictory: you've estimated that only 50 people a day try to make the transfers you want to prohibit, and that that makes it OK to ban them. But you've also said that this is a huge problem costing Amtrak all kinds of money and goodwill. Which is it? Amtrak probably has a pretty good deal with whatever hotel it uses -- not free but less that you or I would pay walking up to the desk. How many people are missing their connections and what is it actually costing? Are there any routings where someone has a 50% chance of missing the connection? A 25% chance of missing the connection? How much of a problem are we actually talking about? (Railroad analysts might call what I'm asking for a p-make analysis of connections at Chicago:
https://pubsonline.informs.org/doi/abs/ ... lCode=trsc). I am most experienced with CZ-Illini/CONO and CZ-LSL, and mostly a few years ago, so I may not have the right experience. (Back in 1988 we missed our LSL-Builder connections in both directions, but I have never been stranded since then and I don't know which is typical these days.) So, sure, lots of us love trains and have starry-eyed, kind of impractical hopes for what the train system could be, but just pointing that out doesn't make your suggestion a good idea. Do you have some numbers to give us an idea of what we're looking at?