08/20 - Super Sunday!

Title: Scenario study(~ 5 minutes) | PANDA Worker: Preview | Accept | Requester Requester: Simone Tang [AQ74Q2Z0O5CPI] (Contact) TO 1: [Pay: 4.54] [Fast: 4.79] [Comm: 4.80] [Fair: 4.90] [Reviews: 102] [ToS: 0] TO 2: [Rate: $6.11/hr] [Pen: -- days] [Res: -- of 0] [Rec: 100% of 1] [Rej: 0] [ToS: 0] [Brk: 0] Reward: $0.50 Duration: 20 minutes Available: 1 Description: $.50 for 5 minutes Requirements: Exc: [53880484-54440] DoesNotExist ; Location In US;

HIT exported from Mturk Suite v1.25.5​

 

So, these "broken" nlp's (the one's that don't display any text to rate, not the one's that don't let you finish one) all seem to have the same issue: whitespace. Or, at least how the code they wrote handles it.

Not Working:

Code:

var text = "� 31 games
� 12 clean sheets
� 6 goals
� 1 last-minute equaliser

� Thanks for the memories, Clint. � https://t.co/thwUVTuVzo";

var text = "Bkchat question: \"Who is better Ronaldo or Messi?\"

Esther: \" from my experience, I used to work at Real Madrid and Barcelona..\" #BKChatLDN";

var text = "@VenomSureiya

【The Battle of W I L L P O W E R】

His first fight went well… But how would this fight fair? This was the brother of —";

Working:

Code:

var text = "Just entered for a chance to win 2 Amcrest Baofeng ATR-22 Two-Way Radios! Thanks @amcrestsecurity! https://t.co/H3CPZsLTSx #giveaway";

var text = "���PLEASE SEE MY LIL POTATO FACE��� @GraysonDolan @EthanDolan #DolanTwinsNewVideo 8 https://t.co/JNEm7jL6Sk";

var text = "Loved your new video Oliver as always it was amazing! Keep up the great work <3 @intravurt";

Tried messing with .trim(), but I don't think I'm using it right. So if anyone has any ideas, post 'em!

 

Twas a wonderful evening for flying, until the right mag went out and the cub made a super fun sound.

 

Title: 30 minute survey - Facebook, Twitter, or Instagram Use | PANDA Requester: Robert J Thomson [A3JIN64DHIFZNQ] (Contact) (TO): [Pay: 4.93] [Fair: 5.00] [Comm: 5.00] [Fast: 4.83] Description: For users of either Facebook, Twitter, or Instagram - Answer this 30 minute anonymous survey about behavior on social network sites Time: 5 hour(s) HITs Available: 1 Reward: $3.50 Qualifications: Masters Exists; HIT approval rate (%) GreaterThanOrEqualTo 90; Location EqualTo US;

 

Title: Name our Shopping Game app | PANDA Requester: Gopinath Subbaraj [A1WXSA3NAD9CYW] (Contact) (TO): [Pay: 5.00] [Fair: 5.00] [Comm: 0.00] [Fast: 5.00] Description: Help name our Price Guessing Game... Select from 2 options. Time: 5 minutes(s) HITs Available: 1 Reward: $0.10 Qualifications: HIT approval rate (%) GreaterThan 90; Location EqualTo US;

 

Title: Short survey about internet usage (average completion around 2.5 minutes) | PANDA Worker: Preview | Accept | Requester Requester: Mindspark [A3PTG4CPUCAV4C] (Contact) TO 1: [Pay: 4.33] [Fast: 4.86] [Comm: 1.80] [Fair: 4.33] [Reviews: 11] [ToS: 0] TO 2: [Rate: --/hr] [Pen: -- days] [Res: -- of 0] [Rec: -- of 0] [Rej: 0] [ToS: 0] [Brk: 0] Description: Answer 6 questions about how you use the internet to search for things. Must be in the US. Time: 60 minutes HITs Available: 1 Reward: $0.50 Qualifications: Total approved HITs is not less than 100; HIT approval rate (%) is not less than 98; Location is US;

 

Title: Search result analysis! | PANDA Requester: Frank Castle [A7FCHOMFPG7DM] (Contact) (TO): [Pay: 5.00] [Fair: 5.00] [Comm: 0.00] [Fast: 5.00] Description: Help figure out sentiment and composition of search results Time: 1 hour(s) HITs Available: 3 Reward: $0.25 Qualifications: Total approved HITs GreaterThan 1000; Masters Exists; HIT approval rate (%) GreaterThan 98; Location EqualTo US;

 

Lol, 10 second glance looks like they didn't properly handle the strings being fed into the HITs. I would not attempt to correct that because you have absolutely no way of knowing if the entire string (from their end) is making it through.

.trim() simply removes leading/trailing whitespace - I don't believe its going to fix hard line break issues (which is what's going on in the HIT I looked at). Even if it did at a base level, its not going to fix the HIT because the line-break is destroying the entire execution of the script. Line breaks will break the HIT in a fairly benign way (you'll probably get the rest in the source code) but characters like `/'/" will really fuck things up and you may lose chunks of data.

Interestingly, if someone tweeted:
#nlpdoesnthandleinputproperly
<script>alert('frack turkers')</script>

and it gets fed through this HIT your browser would throw an alert saying "frack turkers" The line break would end the variable, and then the alert line would get compiled like JS in the HIT and cause the alert to triiger

Edit: languages
Edit 2: I didn't look very hard, you may not even need the <script> tags.

 

No worries, the mags are meant to be redundant so we only need both to function for full power and can safely fly with 1. That being said, there still is a pants shitting moment when the engine makes a crazy sound and you decide screw that, let's head back to the airport!

That's why you always wear brown pants in a plane folks!

 

I thought you were shooting the cubs from the plane and you finished the right magazine of your weapon.