The kanji stumbling-block
The inability to read kanji becomes a great block to self-development. 漢字 appear like potholes on a page. There they are, not so complex, but too complex for you to be able to remember correctly enough tomorrow if you had to compare it to a similar one, or write it by hand. There they are, without sound, for the phonetic readings of kanji are entirely unsignalled. Even if you've used the word a hundred times in conversation it will be fully anonymous there on Page 18, first column, line number 7.
The effect of hitting kanji above your level is pretty much like this: Imagine you are an intermediate student of English and you saw a newspaper article like the one below.
Qantas pilot 축소 in 조종석
by Dylan Welch
The 선장 of a Qantas jet flying over the middle of the Pacific Ocean 축소, causing an 알림 to be sent to Australia and 설정 emergency crews at Cairns on standby, the Transport Safety 관리국 stated in a report released today.
The 선장 was 담당 of a Qantas 767-300 overnight flight from Nagoya, Japan, to Cairns on July 9 this year. He had just handed over the controls to his co-pilot at 4pm (조율된 유니버설Time), when he 축소.
"The co-pilot heard a bang and turned to see the pilot-담당 had 축소 on the 조종석 floor," the report stated.
Now, as an intermediate reader of English, what help has it been to you to have read that article? You wont know how to pronounce any of those mystery words. You can't guess the meaning of them. ("Qantas pilot 축소 in 조종석": = Qantas pilot thrown in jail?; = Qantas pilot basks in success?) You wont remember those characters tomorrow. Even at this moment consider the words "선장" and "섣장". Which of those was used above?
Not only will you never learn that a Qantas jet captain had collapsed in his cockpit (the actual message of the article) there's little to no educational value in this sort of reading practice. It's simply a waste of time.
Provide the furigana
The student of any languages other than those that using chinese ideographs has the advantage of being able to remember and reproduce unknown words much easier. Do you know the english word "martext"? As I just pulled it out of dictionary of obscure words I bet you don't. But I bet you could remember how to spell that tomorrow if you wanted to use it. You also know how to pronounce it, right?
Displaying furigana with a kanji does not, in one go, make it easy to remember the ideograph. But you can read it phonetically and thus you can say it. You might recall it's meaning/context from spoken conversation. If you know the phonetic reading you can look it up in a dictionary. And you can write it yourself if using a computer (as we do pretty much exclusively in this era). A network of connections between the kanji word and it's meaning opens up, and you gain the means to speak the word/kanji.
So...
The extension that I would love to have in my web browser and email client is an agent that attaches furigana to kanji words. In the usual way, as tiny-fonted hiragana font beneath the word.
I would like it to not display the furigana for all the simple words I already know, just those above my level. To initialize this level at installation of the extension I would choose a preset such as "500 kanji". The 500 highest frequency-of-use kanji would become the kanji exclusion list.
As I use the extension over time I would indicate the words that I do and don't know by a context-menu selection or similar action over the word on the page. That would be cached in user preferences. And for ease of use, words that I have already viewed, say, fifty times would be automatically assumed to be learnt vocabulary.
The main point is as above, but it would also be excellent to have context-menu (or likewise) bilingual dictionary lookup for words, as the RikaiChan extension does.
"That's not the way to learn kanji"
Keyboard input has supplanted handwriting in all developed societies. History is history.
The generations of japanese who finished their schooling in handwriting but joined a workforce using PCs typically think their kanji recognition is notably below the level they believe is normal. They are measuring themselves against normal place to get standards, the previous generation, but at this time it also means the pre-computer generation. What I am saying, in summary, is that the prevailing opinion that rote-learning is the one true method of kanji learning is an opinion that is shortly going to be out-of-date.
"You wuss. Show some grit and do it like the Japanese do."
To overcome the 'great kanji leap' the native japanese reader has had two advantages and one absolute incentive. The two advantages are being exposed as child (more allocatable neurons on hand) and being exposed 24/7 to a japanese language environment. The absolute incentive is that there is no alternate option for them if they're to graduate from school, get a job, communicate intelligently with friends, and so on.
Having spelt out the native japanese reader's past advantages, we can see that adult learners of japanese as a second language have no opportunity to obtain the first advantage. The second advantage can't be obtained without sacrificing career and a serious amount of income. And the incentive is not an absolute one for those who already have a different mother-tongue.
"You'll get addicted to the hiragana and ignore the kanji"
By reading kids novels and comics, which have furigana, I did learn how to read the kanji at that level. It has been effective enough that I can't imagine doing any more of it ... I need higher-level kanji now.
This is exactly my philosophy. I developed a similar application using PHP, JDict and Kakasi. This browser extension is a much better solution though. I agree with your conclusions for deciding how and why to do this project this way.