Gents, here is my write-up on twitter, I am making assumptions, which Daryl needs to confirm or correct. Red text are assumptions, not the color of the text shown in the room. Blue text represents what is seen in the chat room window.
Proposal for adding “Twitter” capability within NWSChat.
Goal: Reduce the number of systems forecasters need to use for communicating with general public by providing capability to send Twitter ‘tweets’ right from their NWSChat WFO chat room.
Concept: From within their ‘public’ WFO chatroom, forecasters can send a ‘tweet’ by starting a message with the key phrase “#twitter“, then a single space, followed by no more than 140 characters. Others in the room WILL see the message, this is helpful for media and EM partners, so they are aware of the NWS originated tweet without having to be in twitter at the time. The ‘tweet’ will be ‘sent’ from their corresponding WFO twitter account. For example: if in chat room nwsepz (El Paso), the twitter would be sent from “@NWS_ElPaso”.
In order to ensure that: (1) Non-NWS cannot send tweets from the WFO chatroom, and (2) that NWS forecasters do not inadvertently send a tweet from the wrong WFO, the system will only send tweets from NWS users who have “Admin” privileges for their WFO room, from their WFO chat room.
For a successful send (from a WFO room admin, in their room, less than 141 characters, and OpenAuth password is correct), the message would look like this in the chatroom as echoed: (14:53 PM) NWS ElPaso SOO: #twitter posted @NWS_ElPaso: “This would be the actual text typed by the forecaster”
Note the echoed format: Time and user handle (as with any other text typed in a room), followed by “#Twitter posted” indicating success, followed by the NWS WFO Twitter account name, followed by the typed message.
System Limitations: NWSChat will NOT provide a character count as the message is typed, nor check for any other quality. The ONLY capability which will be implemented for QC is as follows: IF a message is typed which is longer than 140 characters, the bot will reply in the chatroom with an error message, such as: (14:53 PM) NWS ElPaso SOO: #twitter NOT posted – TOO Long: “This would be the actual text typed by the forecaster in the example where the message exceeds one-hundred-fourty characters, it is not sent but still echoed. “ Note that it does NOT include an issuing WFO Twitter name, only the alias for the person who tried to send it; this is very important, as another distinguishing feature indicating no NWSWFO tweet went out. The twitter message will NOT be truncated and will NOT be sent in such cases, but it is shown as echoed text in the room along with the “#twitter NOT posted – TOO Long:” indicator.
We need to address the following:
(1) A forecaster is in multiple rooms, and tries to Tweet in the wrong room: Need to provide feedback so they know that no tweet was sent. If the message is NOT typed by an Admin for the given room, the echoed message would be something like: (14:53 PM) NWS ElPaso SOO: #twitter NOT posted: Invalid room access or Twitter password – “Message text goes here”
(2) Invalid OpenAuth account or password has been changed would result in the same message: (14:53 PM) NWS ElPaso SOO: #twitter NOT posted: Invalid room access or Twitter password – “Message text goes here”
(3) The sample message above would also apply for any non authenticated attempt, such as a media-, ham-, or em- trying to tweet. For example: (14:53 PM) em-john.smith: #twitter NOT posted: Invalid room access or Twitter password – “Message text goes here”
(4) How do we deal with a forecaster trying to tweet using “All chats”?
- Option 1 (preferred): Echo the following message in the All Chats window. (14:53 PM) NWS ElPaso SOO: #twitter NOT posted – send from Local WFO room. “Message text goes here”.
- Option 2: We could just send the message from the room for which they do have admin access, but I DON’T recommend this approach, because they may be in other rooms for which they have admin access.
Other issues to address:
(1) What rooms can tweets be sent from? Only WFO public rooms, e.g. epzchat, but not epzemachat.
(2) Will tweets be sent if #twitter is typed anywhere in the message? It would be best if ONLY send when the message starts with “#twitter “. (Otherwise someone mentioning twitter or quoting a message can result in an unintentional tweet).
(3) Can / should a WFO select NOT to have tweets sent? (Besides deleting admin access to all their employees).
(4) Should we deal with the bot sending any type of automated tweets for non-short term information? I need a lot of help in understanding this one. My gut reaction is NO. But I realize this may be water under the bridge, as we have set precedence. This one needs more vetting.