I noticed this complaint at RC:
tamino says:
13 Sep 2011 at 3:34 PMOK guys. Having something “like†recaptcha to reduce spam is fine. But now it expects me to type (in the little box) the partial derivative of omega with respect to z. No, I don’t mean its description in words — I mean the partial derivative sign, and the Greek letter omega.
This is ludicrous. You do NOT serve your purpose by making it prohibitively difficult to post comments. It would serve you well to think more highly of your readers — because this kind of obstacle indicates otherwise.
I generally avoid commenting when captcha’s are required. But this intrigued me. I wondered: Just how many reloads to I need before I am asked to include a partial derivative of ω with respect to z? I reloaded many times– maybe 20. The captcha often requested punctuation. One seemed to want a subscript 4, like this: 4. Is there an keyboard entry for that?
Eventually, I got this request, which makes me laugh:
I think the first word ends in “s”.
That was pretty funny, so I refreshed a whole bunch of time and got this:
Are you supposed to enter “he nkiali” or “eh nkiali”? or figure out how to enter something to indicate the h and e are upside down? Listening didn’t help.
So, I refreshed and the next choice was:
This one is readable– but do you happen to know the keyboard entry for an a with a dash over it?
Yes, I’m reloading over and over picking the most difficult captchas, but I’m not making these up!
Tip: I find the nospamX plugin works like a charm. Foils spambots; doesn’t force visitors to match unreadable captchas or captchas including partial derivatives. As for rude people or trolls: I really don’t think unreadable captcha’s help on that issue. But if that works for RC, I guess they can go that route. I’m sticking with NoSpamX!
When commenting, please enter your guess for this:
Update: It took my a while, but Jim is fishing in WI. So, here’s a fun one:
and two refreshes later:
This part of the quote struck me as the best:
Really? This was his first clue on what Real Climate thinks of its readers?
bobalab–
Well, it’s Tamino posting that. He’s been a guest poster!
I suspect the guys at Real Climate have no idea how ridiculous those captcha’s are. They should really try NospamX. I was thrilled when I found it!
I recently learned that sometimes the captchas with two words are only using one word for validation. The other word is thrown in there so you can help them with OCR after a machine couldn’t figure it out. My guess is that the captchas with untypeable characters are there for blog readers to do manual translation after a computer failed. I’ll bet you never get two words with untypeable characters in the same captcha.
Learned about it here – http://homepageheroes.com/blog-doing-good-now-less-annoying/
“…reCAPTCHA gathers those unrecognized words and uses them (uses us actually). When you solve the two word distortion only one of them is being used to authenticate your “humanessâ€. The other is unknown to reCAPTCHA and is asking for your “human†translation of the word. If you answer the known word correctly, reCAPTCHA assumes some confidence in your answer of the unknown word. Your answers are then compared with other people given the same words. When enough people have answered in precisely the same way, reCAPTCHA concludes the word is now translated….”
scp–
If it’s true that the obnoxious nearly indescipherable stuff is to get free labor out of commentors, that’s very obnoxious! I’m even more glad I tend to avoid wasting my time on blogs with captchas and that I’ve found noSpamX which doesn’t force my commenters to donate their time to help do someone else’s deciphering for them.
I don’t comment anywhere enough to really care one way or another, except that I thought the idea was pretty clever. But… it’s not really free labor. It’s a voluntary exchange. The commenter does the labor in exchange for having his/her opinion indefinitely stored on the blogger’s servers and displayed on the blogger’s web page (at no monetary cost).
Some people might prefer to only decode one “real” word. Others might prefer the chance to participate in the effort to bring “dead” information back to life.
Having followed your blog for a while, I have noticed that ease of access is very important to you, so noSpamX definitely makes sense here. Maybe not everywhere, though…
scp–
But if your theory is true the commenter doesn’t do labor for the blogger, but for the person who wrote the reCaptcha.
I get that some people might be happy to translate, but other don’t and I prefer not to present them with a frustrating indecipherable 2nd word to translate for whoever it is who wrote recaptchah. I honestly don’t know why any blogger would present commenters with recaptcha if they knew how well NoSpamX works. But of course– I don’t run other blogs. Maybe some bloggers like the to present the hurdle.
I gave up on a comment at RC yesterday. I wanted to ask how valid the assertion is that replacing coal generation with natural gas would increase warming for a decade or two, because it also eliminates emissions of SO2 while reducing CO2 by give or take 35%. (This is assuming that emissions of unburned methane can be controlled, another unknown.) Does anyone here want to enlighten me on that?
Really!
Tamino’s definition of “purpose” clearly differs, because the ReCaptcha is by far the least obstacle in place for those wishing to comment on RC posts!
getting free labor via captchas is really slick. I love it. I suspect it’s another thing that’s too good to be true. As a frequent reader of
OCRd text, i find it continues to contain words that defy belief.
Maybe someone can invent a method for cleaning up machine translation using free labor.
Nothing I’ve seen to date is as funny as the line in the manual for my radar which refers to setting the cookstove. They meant range.
μιμεῖθαι
The iota with the wavy line doesn’t display correctly in the preview. Let’s see if it works in the main text.
Btw, the main Unicode chart index is here: http://www.unicode.org/charts/
The extended unicode character codes in the tables linked above are hexadecimal numbers. For decimal numbers one enters
& # 916 ; (with no spaces) for capital delta Δ
for hexadecimal codes one enters & # x 394; (no spaces) for Δ
The character codes are useful to enter characters that are special characters in HTML. I’m using the code & for the ampersand symbol when I don’t want the character codes translated.
There are entity codes as well as decimal and hexadecimal.
& lt; = <
& gt; = >
That still seems like too much work for entering a comment.
DeWitt, I don’t think the capcha code’s look at html, do they?
I think you have to enter the unicode directly:
ομφιλος
∂P(x,y)/∂x
etc.
On a pc you enter ALT-PLUS (alt and the plus key) followed by the hex code. The Mac has a builtin GUI that lets you select the characters. Not sure how to do it on LINUX.
Re: Carrick (Sep 17 08:44),
I don’t see how you enter A-F from hex numbers when holding down the ALT and PLUS keys. Also, the codes don’t translate correctly. I get è instead of Δ when I enter 394 holding down the ALT and PLUS keys.
A number of computer programs have emerged around the idea of getting people to help solve computer problems like that, called gwap (games with a purpose). I think http://images.google.com/imagelabeler/ was probably the first (more at http://en.wikipedia.org/wiki/ESP_game, also http://www.gwap.com/gwap/gamesPreview/espgame/).
I do agree that being forced to match 0 words is better than either 1 or 2, if it works as well at filtering spam and also creates no more effort for the owner/admin.
@J Ferguson – Not too good to be true. I just verified at RC’s site that they are using recaptcha, then went to recaptcha.net and found this – http://www.google.com/recaptcha/learnmore – “Each new word that cannot be read correctly by OCR is given to a user in conjunction with another word for which the answer is already known. The user is then asked to read both words. If they solve the one for which the answer is known, the system assumes their answer is correct for the new one. The system then gives the new image to a number of other people to determine, with higher confidence, whether the original answer was correct.”
Steve Funk (Comment #81634),
While this has nothing to do with captchas (sorry Lucia), I have been giving the question of aerosol effects versus GHG forcing some thought recently. It is a remarkably complicated question. The direct aerosol effects (increased albedo from particles in clear air) seems most likely to scale linearly with human emissions. That is, increase coal burning and you should get a comparable increase in direct aerosol contribution (unless you scrub out all the particulates and SO2 from the exhaust stream). The problem is that the indirect effects (cloud albedo, cloud lifetime), are claimed by the IPCC to be larger and much less certain than the direct effect, and these effects almost certainly do not scale linearly with emissions. The albedo effect on clouds is a function that depends on roughly the inverse of the total number of cloud droplets. However, it is unknown how much of cloud nucleation is “natural” and how much is due to man-made aerosols, so the potential contribution from a change in the number of man-made cloud nuclei is unknown. Worse, the albedo effects for any assumed change in the number of cloud droplets depends as well on the depth of the cloud…. the effect is potentially much greater for thin clouds than for thick ones. So it is an enigma inside a puzzle, all inside a black box; anybody who says they know the answer is probably mistaken.
One thing is certain: if the IPCC’s “most probable” aerosol effects said to be currently operating are accurate, and these effects were to scale linearly with future increases in human emissions, then we would see a quite dramatic fall in temperatures over the next few decades under a BAU scenario (2%/yr increase in emissions over the next 25 years) due to rapid increases in total albedo. The levels of warming projected by climate models (eg ~0.2 C per year) can only happen if net aerosol effects scale much less than linearly with increases in emissions. Something like a cube root scaling with total human emissions is needed for the climate model warming projections to be possible. Depending on how you assume aerosol effects to scale with emission rates, you can predict most any climate sensitivity and future rate of warming you find attractive.
Of course, since some models do not even include indirect aerosol effects, and since each modeling group uses a different assumed level of total aerosol effects (current and historical!), I think it is fair to say that aerosols are simply a kludge to ‘tune’ model hind-casts to historical warming. All the above is a long way of saying: nobody knows how much changing from coal to natural gas would change warming in the next few decades.
.
BTW, a reasonable alternative explanation is that the true current aerosol effects are close to the low end of the IPCC estimated range, the climate’s actual sensitivity is somewhere between 1.5 C and 1.8 C per doubling of CO2, and warming over the next few decades will be ~0.12C to ~0.14C per decade (plus or minus any natural cyclical changes in temperature). But you are not going to ever hear that at a politically motivated climate blog like RC, even if you do manage to get past the captchas and the moderation. Infidels are put to the sword. 😉
Ok– so the re-captcha thing is true.
Given current unemployment rates, I’d much rather whoever is trying to translate hard to digitize words in the NY Times hire people at minimum wage to do the job or hire people in India (or other lower wage places where people speak and read English) rather than put my commenters through the unnecessary evil of solving the human-detecting word and trying to solve the extra word.
Here’s a bit more on unicode and alt-code input:
http://en.wikipedia.org/wiki/Unicode_input
http://en.wikipedia.org/wiki/Alt_code
This allows you to enter the number as a decimal value.
I’m not a windoze person, so take anything I say on that with a grain of salt, it’s based on googling not personal experience.
I said above (#81657) “The albedo effect on clouds is a function that depends on roughly the inverse of the total number of cloud droplets.” That should have been: “… the inverse of the average cloud droplet diameter.”
I’m dyslexic and these things absolutely kill me. I have to change two passwords every three months that need quite specific types of characters, essentially information free. I write them in my bike helmet.
Captchas are pure PITA’s. The hellacious ones are doubly PITA’s.
I avoid them if at all possible. Commenting at RC has marginal value anyway; add double-strength PITA captchas and the value goes quickly negative.
Re: Carrick (Sep 17 10:06),
You can’t enter a-f with the alt key depressed in Windows because that brings up menus. There’s probably a way of defeating that, but then if your mouse gives up the ghost suddenly you’re SOL.
DeWitt, the ALT-+ is something you press and hold, release, then type the hex code. But you have to enable this via regedit.
If you hold the alt while typing the keys on numerical key pad, you have to enter the decimal value. I’ve succeeded in doing that in the past, but haven’t tried it in a while, so my memory is fuzzy (and if you didn’t guess, I’m typing on a Macintosh which doesn’t have this issue.)
So enough of my speculation, here’s another reference that may be of some use.
ΔRe: Carrick (Sep 17 14:21),
You can release the plus key, but you have to keep the ALT key depressed. I found a utility that opens when you press ALT and PLUS simultaneously to enter unicode characters that seems to work. In fact, it’s available from your link, which I had already found. His Method 5 sort of works, sometimes.
scp,
so it is real. I still think it’s wonderful. Pure genius.
next someone will put little generators on shopping carts with small batteries and then recover the energy while the store is closed overnight.
Maybe Lucia will entertain a contest for quatloos for other schemes to siphon off some intellectual effort a micro-thought at a time in support of some larger effort.
reCaptcha is a good idea, if it helps train OCR tools to be more useful. However, I did go look up the WP NoSpamNX plugin…
“PLEASE NOTE: Due to lack of time there will be no further development of NoSpamNX until further notice. The Plugin is ‘on hold’. (10.04.2011)”
AnonyMoose–
It works fine as is– best spam plugin I’ve ever used. Should problems begin to arise, then I can consider reCaptcha. Or something else. 🙂
At least they didn’t ask you to CALCULATE the partial derivative of omega with respect to z. That would be one way to keep me off their site for good.
Maybe the good folks at RC have decided they just don’t want readers any more. I gave up on them a long time ago. I know there is a certain amount of keeping up that I’m losing out on, but I just can’t rationalize giving their site my traffic because of their bad behavior.
On the subject of Captchas, I’ve had opthamic shingles this year [something I would not wish even on Gavin, and Hansen] and my vision in one eye is still a bit blurrrryy so captchas can be quite a challenge – and they do seem to be reaching the limits of legibility in general.
On the other hand Firefox gives access to all you ASCII and other “special characters” at all times under the Edit menu However programmers who force their users to resort to such measures should have their fingers bound the way Chinese ladies had their feet bound years ago – same goes for programmers who force us to input phone numbers without dashes because they are too lazy to write the lines of code to parse a phone number, something that was actually a first semester Fortran programming assignment for me at UCONN many years ago.
What’s that all about?
W^3
Indeed, you only need to enter the legible word. I like the human OCR spin, if you need to enter a captcha anyway it may as well be useful. It’s hardly theft of labour when the alternatives are comparably cumbersome to decipher but do no useful work (apart from the primary function, stopping bots).
Gee what happened to your ice posts… seems to be a huge decline in interest in this site over the past week LOL
http://ocean.dmi.dk/arctic/icecover.uk.php
Andrew (aka andrea,Annabelle Torres,Marie deschamps,Albert,Laura Gonzales,Rebecca, Fulton, and Albert)
1) Your comment showing the minimum has also been reached based on DMI belongs on the post discussing the minimum likely having been reached on JAXA.
2) Please pick one sockpuppet name. It may not be “Andrew”. If you use “Andrew”, you need to add letters to distinguish yourself from the other Andrews who used that name before you decided it was a good new sockpuppet name.
How can something that annoys Grant Foster be a bad thing?
It’s interesting that according to the video here, the whole recaptcha idea is based on a mistake in logic – http://code.google.com/apis/recaptcha/.
The guy was worried about the time being “wasted” on captchas, so he doubled the amount of time being spent decoding phrases. Near the end, he says – “people are usually very happy about this… these things are annoying, but at least my time is not wasted any more.”
Seems to me that the time spent answering the “real” captcha is unchanged, but now a new time cost is being added in order to answer the second captcha to manually decipher the old books. He’s increased the time cost of entering a comment without actually improving the spam filtering. There has been no reduction in “wasted” time. If anything, the “wasted” time is increased (if I don’t happen to think that deciphering old books so that google can profit from them is a valuable use of my time).
At any rate, I’m confident that the untypeable characters at RC can be safely ignored.
Julio–
.
That’s they way I look at it.
I think Google has come up with a system that wastes more human time. Maybe some in the know people can guess which word doesn’t need to be entered to get their comment through and not waste time on the other one. But many are going to try to figure out how to enter this one
Alternatively, they may reload or try to listen to the audio. Even though the commenter can just reload to get a new captcha, that option still consumes the commenters time.
Well: It’s especially not so when we consider that these are old books that google may consider too expensive to translate. It’s true some old things wouldn’t get translated– but some old translated things aren’t worth reading. I wouldn’t be surprised if the amount of time “volunteers” waste translating through re-captcha ends up exceeding the amount of time anyone will ever spend looking at the translation!
This is why I think if deciphering any particular old book is worthwhile, whoever wants them deciphered (presumably Google ) should hire people and pay them.
Now that I know how recaptcha works, and I’ve looked at a whole bunch of them, I think we can all guess that the open/filled letters are usually the real captcha. Those who know can decide whether
a) to bother to waste time digitizing books for google
b) to enter clearly fake choices to thwart google’s desire to get blog owners to “volunteer” their commenters to digitize old books for google.