Below you write 

". . . let’s say every X number of each typists' keystrokes + a unique and viewable random series of digits/letters are hashed using Sha-1. The result will be a unique 40-digit hexadecimal number. That number could then be fed automatically into bitaddress.org's brainwallet function to produce a public and secret key. (They would be used only as a time/date stamp.) . . ."

Generating an address alone won't generate a time/day stamp. You have to put bitcoins into the address.  So the typists would themselves have to be governed by a smart contract that would move a minimum amount of their own coin into an address with every X number of keystrokes, right? And since others might see the addresses generated, the coin would have to be moved right back out again.

First, I want to say thanks to Luckybit for posting this topic here. A while back I tried to start a discussion on ethereum.org's general discussion forum. It has so far generated all of 40 or so views and not a single comment.

Regarding your most recent reply, I take it that your reservations are two-fold:

1) how to establish whether actual people and not bots are participating as typists -- bots that would presumably try to swamp the system with falsely group-validated keystrokes.

2) fitting the lotteries into Bitshares PLAY in order to generate truly random winners without relying on a trusted central authority.

My comments:

1) Where the material uploaded to be typed/translated is handwritten, the work itself would prove that a human is participating and not a bot because software can't yet read handwriting nearly well enough. There is no way a machine could produce keystrokes that would be validated by the majority of (human) typists. As an additional form of protection, Captchas could be randomly required to continue typing, each Captcha being unique to each typist. And only keystrokes in real time would be acceptable.

In the case of a Word file uploaded for, say, group translation, Captchas could be randomly required here too. (The entire document might even be converted into a Captcha-looking format.)

[Edit: I'm not really sure all those Capcha tricks would really be necessary. If there were software programs out there that could fake keystrokes (or even roughly mimic handwritten text) for the purpose of swamping the system, there could also be software programs out there that could catch such programs, and there is no way such programs could be validated by typists doing real work. The fake stuff would stand out. And why would anyone use such a program anyway? Remember that chance would still rule. Easier to just type right along with everyone else for the purpose of achieving consensus.

[As for Word files uploaded for translations, any one or group of people who tried to copy/paste the text into a Web translator would still have to type the translated result in manually. But even so, the system could in theory be swamped by billions of Beavis and Buttheads using the same online translator. The solution would probably be to just to automatically run such a file through an identified online translator as a matter of course before submitting it to the typists -- and then have those typists correct the machine-made result. Here the work of real people working together to achieve consensus would defeat the copy/pasters.]

2) For those willing to pre-register with their real names and submit SSNs, etc. to firmly establish their identities (and pay taxes on any substantial winnings) I think it would be routine to establish a trustworthy way of randomly finding a qualified winner. These lotteries might even be government supervised, like a state Lotto. Libraries and archives seeking to digitize their handwritten material could fund such lotteries. Of course, such lotteries need not be bitcoin-based. Dollars and Paypal would do.

Even confidential material -- lawyers’ handwritten notes and diaries -- could be processed. It should be fairly easy to develop a smartphone app that would take a picture of a handwritten page (or of a computer screen of text) and then allow that picture to be parsed into separate files which would then be uploaded to separate lotteries for typing, thus avoiding any one virtual pool of typists from knowing too much about it. Looked at this way, it might not even be necessary after all to use smart contracts.

But competition for typists might induce lottery funders to go via Tor to the Dark Web, thus allowing tax-free payouts and no identity registration. Here is where smart contracts would be required. But how difficult would it really be to have transparency? We want typists with group-validated keystrokes to earn tickets proportional to the work they do. I’m no mathematician (which may be obvious), but let’s say every X number of each typists' keystrokes + a unique and viewable random series of digits/letters are hashed using Sha-1. The result will be a unique 40-digit hexadecimal number. That number could then be fed automatically into bitaddress.org's brainwallet function to produce a public and secret key. (They would be used only as a time/date stamp.) The hash-input and public key would constitute a lottery ticket. Each lottery in advance could state what its own "goal number" hash/public key will be. It might be the previous days’ Dow Jones closing fed into the hash-bitaddress generator, the point being that it is something transparent and anyone can confirm it. Now as people type, they could watch their own tickets being generated and consensus-validated in real time. The ticket-hex number closest to (or perhaps furthest from, see below) the goal-number (again, stated in advance) would win. In this way, typists could see as they type how close to or far away from they are from likely winning anything. Of course, mere submission of a "winning" hex number by anybody on the Web would not be enough. Only logged-on typists with the actual recorded keystrokes would qualify because only they would have the numbers that produced the hashed result.

(Presumably the calculations to produce the numbers would be strictly client-side, to prevent other typists knowing anyone else's tickets, although some lotteries might allow everyone to see others' tickets as a reflection of typists' ability to communicate with each other anyway. In that case, the winning ticket would probably have to be the one furthest from a lottery's goal-number announced in advance, otherwise premature public knowledge of a generated random number very close to the goal number would discourage further participation. Even with that arrangement, a given lottery might have to offer rounds of payouts as knowledge of a generated number vastly far from the goal-number would encourage typists to quit and move to a more favorable "game". In fact, tracking of the numbers might generate a kind of side-game in which non-typists bet on whether a second round of payouts will be required. After all, true gamblers will bet on anything -- think of the famous example of betting on which drop of rain will reach the window sill first.

(Note that the above describes a process not unlike bitcoin mining, in which people are generating random numbers in a kind of "proof of work".)

Admittedly this is all a crude outline, but my point is that it shouldn’t be too hard to create a smart contract that can automatically collect the needed information (Dow Jones closing figure, etc.); create typists’ lottery tickets and the winning ticket in a transparent manner; and then send the winner the bitcoin winnings.

Or maybe I just misunderstood your comments. If so . . . never mind . . .