It seems like this would be a great case for doing a machine learning approach. Something good at anomaly detection like single class support vector machines, perhaps.
I wonder how well that would translate for a client services person, though. "Why do we think this client is off?" "Because black box."
I would argue that you are correct, but I wouldn't limit it to Twitter. There are many crackhouses -- even hacker news can be one (though thankfully, our junkies remain civil and politely die in a corner where no one sees them). It is never as easy to be inhuman to someone as when hiding behind the veil of apparent anonymity.
I seem to recall, a while back the creator of RVM suffered a similarly disheartening setback when rbenv came out. RVM is still going strong. The community has not decided on lightweight frameworks. The choices you make won't be the same as lotus. You will no doubt attract a group of like-minded people who will be grateful for your contributions.
There isn't even much technical ingenuity here. I'm not sure how this is even noteworthy. You save $1 between two people? So you split it for a lousy $0.50 or maybe $100 per year? Just stop drinking lattes and you'll save that in a month.
Not to mention if you live further away, you deserve to pay more to make up for the extra resources needed to move you around, as nchuhoai said.
Though who knows if anyone actually takes these routes. And remember many of the places BART goes aren't exactly filled with people with large disposable incomes who can stop drinking "lattes" to make up the difference.
I must not understand his methodology, because if you're willing to meet at an interim stop (as in the examples above), then it's easy to save quite a bit more than $2.05 just by swapping with someone who is taking almost exactly the opposite route as you.
For example, if you're making the long trip from San Bruno to North Concord / Martinez, that'll cost you $6.95. If someone is traveling from Pittsburg to South San Francisco (almost exactly the opposite route), then that would normally cost them $6.70. But if you notice that their destination is one stop from your starting point, and vice versa, then you can meet up at any station in between and swap tickets, so the ticket you started with travels from San Bruno to South SF ($3.10) and their ticket travels from Pittsburg to North Concord / Martinez ($1.85). Instead of paying a combined $13.65, you pay a combined $4.95.
What am I missing? Are those twitter data scientists just not that bright? ;)
Orinda => Pittsburg is going the same way and on the same line as Walnut-Creek => Pleasant-Hill, Walnut-Creek => Concord and Pleasant-Hill => Concord (the East-bound Pittsburg Bay Point train). There is no need to get off the train in order to swap tickets.
You are of course right though that you can save far more money if you do get off and swap with someone going the opposite way, but then you have to wait for another train. If you already get off the train in order to transfer (say you commute from Pittsburg to Fremont), then swapping cards with someone going the opposite direction is easy.
