I used to be a bit of a perfectionist when it came to language. I was known — infamous, I should say — for correcting everyone’s grammar, from the time I first learned what grammar was, all the way to the time I started learning about the differences between prescriptive and descriptive grammar. Nowadays, I’ve given up correcting people’s grammar — not because it doesn’t need correcting (!), but because I’ve learned that it is practical usage, even incorrect usage, that moves the wheels of language. If people are going to start spelling nightlight as nitelite, well, why not? If people want to use they in the singular, well, this has a long pedigree in English usage (if it was good enough for Shakespeare). And if people want to spell all right as one word, alright, let them.
For the record, I spell it all right, and consequently, it tends to catch my eye when people spell it alright. This is what happened while I was reading a political commentary by the well-known journalist, David Frum. The title of the column is, “In South Carolina, the kids are not alright” [link]. The title doubly caught my eye, and you might be able to guess why. Frum adopts the renegade spelling alright, referring to a song by The Who in which it is so spelled. But the same phrase is the title of a successful, award-winning film, The Kids Are All Right. See that? Thirty-five years more recently than the song, and thirty-one more recently that a documentary film about The Who with the same title, the newer film reverts to the “correct” spelling. My guess is that the 1965 song was the inspiration for the title of the 2010 film, but I don’t know that for a fact.
But although (all though) I write all right, it doesn’t bother me when others write alright. Why not? Apart from the fact that the linguistic winds are always (all ways) blowing change into the sails of the language, and apart from the fact that the compound form is already (all ready) deeply entrenched in popular usage, there is a long, legitimizing history of this form. The first edition of the Oxford English Dictionary has it, marked obsolete, and dates one of its earliest uses (c. 1230) to a passage in the Ancrene Riwle. The OED gives the passage thus, “And alriht so of þe oðre wittes”. In fairness, some editors of this work have transcribed it al riht (two words). For example, J.R.R. Tolkien , and much more recently, Bella Millett . In fact, most editions of the Ancrene Wisse have it as two distinct words, though I have seen one or two which have it as a single compound word, and it was obviously also (all so) the case with the edition consulted by the staff of the OED. How can we answer which is right, among all the extant manuscripts, and with all the editorial preferences imposed on them? 
But this is rather beside the point, or perhaps it even reinforces the point, as I hope some of the parenthetical phrases in the preceding paragraph suggest. All right comes down to us from the Middle English al riht(es) or alriht(es), in turn from Old English eal(l) riht or eal(l)riht. Certainly, the origin of this, and all of these similar collocations, is as two distinct words, but it has been written as one for nearly as long as it has two. In my view, there is no reason for anyone to feel guilty about writing alright, all right? It’s a perfectly good word, albeit (all be it) wrong by convention. But grammar is a fragile thing. Tomorrow, who knows?
And now, I am wondering: have today’s francophones started writing çava (as one word)? After all, the OED gives alamode as a perfectly good alternative for à la mode.
 J.R.R. Tolkien, ed. Ancrene Wisse: The English Text of the Ancrene Riwle, MS Corpus Christi College, Cambridge 402. Early English Text Society, No. 249. London: Oxford University Press, 1962. See, for example, p. 36, l. 21f., where Tolkien reads al riht al swa (“all right also”).
 Bella Millett, ed. Ancrene Wisse: A Corrected Edition of the Text in Cambridge, Corpus Christi College, MS 402, with Variants from Other Manuscripts. Early English Text Society, No. 325. Oxford: Oxford University Press, 2005. This is essentially an edition updated from Tolkien’s of forty years earlier. Compare to  the passage on p. 27, here read a bit differently, as al riht alswa. By “corrected edition”, does she mean to say she’s correcting previous editors or correcting the scribes? :)
 To give one notable example: Frances M. Mack and Arne Zettersten, eds. Ancrene Riwle: The English Text of the Ancrene Riwle, BM MS Cotton Titus D xviii. Early English Text Society, No. 252. London: Oxford University Press, 1963. Here, to compare to the same passage in the previous notes, we have yet a third transcription: alriht alswa. Arne Zettersten also discusses alriht, again giving it as a single word, in his Studies in the Dialect and Vocabulary of the Ancrene Riwle. Lund Studies in English, No. 34. Lund: C.W.K. Gleerup, 1965. This was the translation into English of Zettersten’s doctoral thesis, and it bears all the signs of influence and advice from Tolkien, a friend and colleague to Zettersten at the time. I am tempted to call Zettersten a protégé of Tolkien’s, but I am not quite sure it would be accurate. But Mack and Zettersten’s edition came just one year after Tolkien’s, as the Early English Text Society attempted to work its way through all of the extant manuscripts.