262 karma · joined January 22, 2015
> Three Japanese words, three Hepburn transcriptions. The letters you already know, telling you what they sound like.
I mean come on.
> Order food
> Food comes out as raw, unprepared ingredients
> Complain to chef
> Tells me to go cook it myself
> wtf, I'm not here to do work for free
> "Is doing something for yourself really working for free?"
The solution here is to fix the sample text itself to use U+202F (NARROW NO-BREAK SPACE,  ). As opposed to the standard width no-break space it currently has for "No. 12" and "Fig. 7".
I added the “Alice in Wonderland (long)” text sample to the demo as a stress test, but its paragraphs are still relatively short. I should add a sample that’s one really long paragraph and see what happens. I may need to add a bail-out after X items option if it calls for it.
And yes, I did use LLMs for the implementation, both Fable 5 and 5.6 Sol specifically. I would not have been able to achieve anywhere close to the current result unaided. My patience and expertise would’ve been exhausted long before getting anywhere especially interesting.
My philosophy with LLMs is to try to use them to achieve levels of quality and polish that would’ve been previously unreasonable. I hope this project is a demonstration of that.
Feel free to make a GitHub issue with details and I can take a look at supporting it. For now the workaround would be falling back to native justification by excluding the paragraph from justif, and using css float/shape-outside to get the desired edge.
Narrow columns: open up the controls on the demo and try adjusting the width slider :)
Long documents: try the “Alice in Wonderland (long)” sample text in the demo to get a feel for how it handles long documents. In general it performs well enough I think, though Safari is a little slower than other platforms. It doesn’t do it incrementally (not sure that’s possible), but I spent quite a while trying to make sure it was fast enough to re-layout in real time even on long documents.
I haven’t thought about coexisting. I’m not sure how that could work but I’m open to ideas.
But there factors that counteract this. Merchants pay different average fees depending on the mix of cards that they accept. A dollar store probably sees fewer Chase Sapphire Reserved or Amex Platinum cards than a Gucci store does. Accordingly, the average interchange fee they pay is probably lower. [0]
Also fee surcharging is becoming more common in the US these days, and is already common is some other countries. Visa et al. still don't allow merchants to pass on the actual interchange they'll pay, but merchants can charge a fee to credit card transactions that they don't on debit or cash.
But my original point is that giving rich people credit card rewards are not a transfer from the poor to the rich. It's a rational business decision by card companies that does not require any outside funding source, from poor people or otherwise, to make work financially.
0: There are some other factors that complicate this, such as the risk profile of the business. Riskier generally means higher fees.
Credit card companies do make a lot of money off of interest. But they also make a lot of money off of interchange fees.
Businesses want wealthy customers because they spend more money, so they're willing to pay a higher interchange fee to access those customers. Higher interchange fees mean card companies can offer better rewards, which in turn attract more wealthy customers to their cards.
So even if credit card debt was not a thing, it would still be incredibly profitable for card companies to sell access to their rich cardholder clientele, and to in turn provide rewards to those cardholders.
And I did struggle a bit with deciding how best to do the comparison to native browser rendering. I wanted it to be an apples-to-apples comparison, but also didn't want to unfairly handicap the browser by disabling something like hyphenation. It's already bad enough as is...
0: http://www.youtube.com/playlist?list=PLbgaMIhjbmEnaH_LTkxLI7...
If your name is on there it’s probably included in namex.
I could also manually check the data of course but that would require knowing your name :)
In an ideal world maybe learners could focus exclusively on listening and speaking first, then move on to kanji later. But writing is a very useful tool in learning, and having access to that tool can help speed things up.
Like most things in life, a balanced approach is probably the right one. But you have to know what your goal is. Our brains are lazy, they only get better at what we make them get better at. If your goal is to just read kanji, practice reading kanji. If your goal is to understand and speak the language, practice listening to and speaking the language. But if you want to have a balanced language ability, you'll need to practice it all.
Overall it's been a successful approach, and I recommend it to new learners unless they have a particular interest in being able to write by hand or they feel strongly that writing the characters helps them remember them.
It's only rarely that I have to write anything other than my own name in Japanese. I've practiced my address but writing it in English is fine in 99% of situations. Being able to write properly would save a little embarrassment, but I still believe my language learning time would have a much higher ROI in other areas.
HN discussion: https://news.ycombinator.com/item?id=25023673
One other limitation in place is that these transit cards have a limit of ¥20,000 (~140 USD) max that can be loaded on to them. So any transaction larger than that is out of the question.
So to answer your question, no this isn’t really a person-to-person cash replacement. It’s a transit card that happens to be able to be used as an offline payment method, but it’s got various limitations and weirdness that prevent it from being something more.