[jdtig]

You think that /k/ sounds like /g/, because you speak a language (English) that doesn't distinguish them. However, I imagine that writing "k" for /k/ makes perfect sense to speakers of languages that have /g/, /k/, and /kh/.

And your solution works for /k/ and /kh/, but runs into problems when you have a three-way contrast (/b/p/ph/ and /d/t/th). You can't write "buu" for "crab" just because you think it sounds like an English "b"; you need "b" for the actual /b/ sound.

[smallnamespace]

A romanization that’s good for writing an entire language accurately is usually not going to come up with spellings that lead an English speaker to a good initial approximation for exactly this reason.

I would argue that since food names are usually pronounced by non-speakers who will be unfamiliar with the romanization (or often don’t even know which language it’s coming from), the latter sort of transliteration is more useful. Egg foo young, not egg fuyung, egg furong, or daan fuyung.

[CitizenKane]

Not sure what the issue would be here. I don't think it would be accurate to transliterate ปู as "buu" but instead "bpuu". Same with /d/t/th, as these can be written just fine like ใด - dai and ไต - dtai for instance. It can't capture everything, but it's impossible to anyhow because English doesn't have consonant classes like Thai does.

[wingerlang]

I just think that ไก่ is miles closer to Gai than Kai, that's all. Another hill would be that ปู is best written as bpuu/bpoo.

I don't really think it matters, no one is consistent, no transliteration covers all cases. Unless maybe ipa? (which no one uses to communicate anyway).

[jdtig]

I can agree that the Paiboon romanization (g/k, b/bp/b, d/dt/t) is probably the most intuitive for English-speakers.

But it's silly to say that it's the best system just because it is easiest for people like yourself.

I do think something IPA-inspired (k/kh, b/p/ph, d/t/th) is consistent and still usable. It will also encourage better pronunciation (at least for somewhat serious students) if they are trying to say the correct sound (unaspirated /k/) instead of approximating it with an English phoneme that isn't used in Thai (voiced /g/).

[wingerlang]

I didn't say it is the best system though, I said that I think some characters are more accurate than others. The implied context being, casually writing Thai in English, to other English speakers.

But I do think that an intuitive system is better than an IPA inspired system, and that's purely because of the fact that only 1% of people will be using it and understanding it. The other 99% will be using other, random, variations. However an intuitive system will at least catch most people who speaks English.

Any romanization that is more than "close enough" is a waste of time though, when compared to simply learning to write in Thai.