Stilgherrian (@stilgherrian)

Wentworth Falls NSW AU

The below is an off-site archive of all tweets posted by @stilgherrian ever

August 4th, 2015

@Steve_Lockstep @sgpulman I’ll read that in the morning, when my eyes aren’t falling out my head.

via TweetDeck in reply to Steve_Lockstep

@GordyPls I’d certainly give it a go, yes.

via TweetDeck in reply to GordyPls

@davispg @kcarruthers True, but then the people who specifically wanted clean material would be annoyed.

via TweetDeck in reply to davispg

@kcarruthers So, you know it can’t get to 100% Malcolm Tucker unless you get people to change their pledges, because arithmetic.

via TweetDeck in reply to kcarruthers

Very pleased to have been invited to @sgpulman’s masterclass on Sentiment Analysis tomorrow. adma.com.au/adma-fomma/mas… pic.twitter.com/kAZc7x9ygm

via TweetDeck

@joannejacobs Yeah I think it’ll have to be the weekend.

via TweetDeck in reply to joannejacobs

You know, 24 minutes of Malcolm Tucker grade ranting is going to be quite exhausting. Also, fun.

via TweetDeck

“The 9pm Urgent Hardware Refresh” now 72% funded, 47h left. Explicit / Tucker level 32%. pozible.com/edict2 pic.twitter.com/gLsCTxFnGv

via TweetDeck

Thank you, everyone, yeah look I know the traditional answer is “Yes”, but I’ve had a long day. I’ll save it for another time.

via TweetDeck

Finally back at the hotel, exhausted. Do I want to open the bottle of wine that has been left for me?

via TweetDeck

And the radio spot is done… recording to be posted in due course.

via Plume for Android

@aussiegoldy There’s several stories that I’ve been sure to brief myself on, but I dunno which ones we’ll do. Welcome to live chat radio.

via Tweetbot for iΟS in reply to aussiegoldy

@belindafarrelly True, I’ve been up since 0500 and caught the 0706 train.

via Plume for Android in reply to belindafarrelly

@semibogan @jdub I mean, gentlemen of a certain persuasion will remember Eagles BBS.

via Plume for Android in reply to semibogan

AFP Last surviving WWII Dambusters pilot dies in New Zealand u.afp.com/ZiS5

via Twitter Web Client (retweeted on 8:01 PM, Aug 4th, 2015 via Tweetbot for iΟS)

Seriously, you could sate me with a bucket of salmon roe and a funnel.

via Tweetbot for iΟS

@semibogan At the risk of ruining the joke, or making myself an object of ridicule… “I’m having fish.”

via Tweetbot for iΟS in reply to semibogan

phbarratt @Pollytics You have to admit HMAS Able Seaman Pyne has a certain ring to it.

via Twitter for iPad (retweeted on 7:53 PM, Aug 4th, 2015 via Tweetbot for iΟS)

@gnoll110 It’s OK, I’m talking about cybers.

via Tweetbot for iΟS in reply to gnoll110

Jesus, someone had better tell Cory about this. twitter.com/drearyclocks/s…

via TweetDeck

I suppose I should stop drinking, grab a bite to eat, and head over to the ABC.

via TweetDeck

I’ll be talking various tech stories on ABC @702sydney and Local Radio around NSW at 2030 AEST, so in an hour. abc.net.au/sydney/

via TweetDeck

@jonoabroad In what aspect of the job? I suppose I could frock up, I dunno.

via TweetDeck in reply to jonoabroad

@rtse999 Oh thank you. Well, I’m in Prof Pulman’s masterclass in the morning, lucnh dunno, then media disruption stream in the afternoon.

via TweetDeck in reply to rtse999

Well this puts a whole new perspective on things with respect to Soapy. urbandictionary.com/define.php?ter… HT @michaelneale pic.twitter.com/tfOPfXgvma

via TweetDeck

@jonoabroad I don’t think you mentioned it, but, with respect, it’s a terrible idea.

via TweetDeck in reply to jonoabroad

“The 9pm Urgent Hardware Refresh” hits 71% funded. pozible.com/edict2 A great new Public House Forum question too: “Craft beer, why?”

via TweetDeck

@jonoabroad Well someone has already pledged to pay me to explain why Soapy is an excellent Attorney-General. More on that later tonight.

via TweetDeck in reply to jonoabroad

“The 9pm Urgent Hardware Refresh” hits 71% funded. pozible.com/edict1 A great new Public House Forum question too: “Craft beer, why?”

via TweetDeck

@sxpert1 @mikko Well, it isn’t exactly the @FBI’s job to protect nations other than the US.

via TweetDeck in reply to sxpert1

@rtse999 Heh. I thought that was you, but I didn’t get a Polite Moment to ask. Will we see you tomorrow?

via TweetDeck in reply to rtse999

@will_ock Oh, I misunderstood the question. Now I’m confused too.

via TweetDeck in reply to will_ock

@will_ock You’re missing the fact that people are arsehats.

via TweetDeck in reply to will_ock

@NewtonMark Damn you’re good with computers.

via TweetDeck in reply to NewtonMark

I know, @ResignInShame, I’ve just re-watched “The Thick of It” series 1 to 3. I seem to not have series 4. twitter.com/ResignInShame/…

via TweetDeck

If you don’t know what “The 9pm Urgent Hardware Refresh” is about, this video may help. pozible.com/edict2 youtube.com/watch?v=QrCsVs…

via TweetDeck

@NewtonMark Did you try turning it off and back on again?

via TweetDeck in reply to NewtonMark

“The 9pm Urgent Hardware Refresh” now 67% funded, 48h left. Explicit / Tucker down to 82%. pozible.com/edict2 pic.twitter.com/tgdAEGMM61

via TweetDeck

@michaelneale ADMA was Aust Direct Marketing Assoc, now Assoc for Data-driven Marketing and Advertising. The industry association.

via TweetDeck in reply to michaelneale

Heading to Thaitown to prep for my 2030 spot on @702sydney and, I daresay, eat.

via Plume for Android

finicalgal …or you could just place the cup on the table. pic.twitter.com/tD6aUB6Uxk

via Photos on iOS (retweeted on 5:35 PM, Aug 4th, 2015 via Plume for Android)

RealNickHodge Stop arguing about stupid shit, people

via Twitter for Windows Phone (retweeted on 5:33 PM, Aug 4th, 2015 via Plume for Android)

NSWRFS Crews have made good progress towards containing the Fire. Details: rfs.nsw.gov.au/fire-informati… pic.twitter.com/vnlAaRTkqT

via TweetDeck (retweeted on 5:24 PM, Aug 4th, 2015 via Plume for Android)

@awelder Actually, you’re right, in 20 years these guys will have excellent longitudinal data. Will it reveal patterns?

via TweetDeck in reply to awelder

Mechanical Turk workers worked bette with polar Qs (cheap or expensive?) not shades (very vs moderately expensive)

via TweetDeck

Remember, all your data is based on what happened in the past. In fashion, that doesn’t always work so well.

via TweetDeck

Elbert: As data scientists, we fall in love with our own models… Are they actually useful to anyoneMcs

via TweetDeck

jturner_ibrs thought (bubble) leadership

via Twitter Web Client (retweeted on 4:59 PM, Aug 4th, 2015 via TweetDeck)

Gilt’s process also factors in store-wide data. instagram.com/p/588YkBCFrh/

via Instagram

So Gilt’s process is a mixture of machine learning and the wisdom of strangers in the cloud.

via TweetDeck

It turns out that one of those answers is very, very predictive of whether a dress will sell.

via TweetDeck

Gilt has also been asking Mechanical Turk to provide data about the dress. instagram.com/p/58754cCFq8/

via Instagram

Sometimes metadata about the product is predictive about how it’ll sell where. instagram.com/p/587l0rCFqo/

via Instagram

Remember to clean your data before putting it into the machine learning. instagram.com/p/586pwiiFpt/

via Instagram

rtse999 Data is king - 90% of time is spent on cleaning and preparing data - @ielbert

via Drafts for iPad (retweeted on 4:42 PM, Aug 4th, 2015 via TweetDeck)

@garystark Yeah, is North Dakota the flat one? A bit of height might well help.

via TweetDeck in reply to garystark

@reselsnark No, that’s her point. That this sort of thing is what flags stuff as wrong.

via TweetDeck in reply to reselsnark

Why do they wear higher heels in North Dakota than in Canada? instagram.com/p/585QZNiFob/

via Instagram

Now: Pre-emptive shipping: Igor Elbert, Principal Data Scientist, Gilt.com

via TweetDeck

5at5 number 208: Windwheels, Elon Musk, why Facebook rules, 1980s kung fu kitsch, and why the US ain’t a democracy. tinyletter.com/5at5/letters/5…

via TweetDeck

So overall, maybe 5 to 10% of traffic is actually real.

via TweetDeck

In one example, 50% of clicks came from countries where the ad wasn’t scheduled to be shown.

via TweetDeck

Perlich: 10 to 40% of internet traffic is not explicable by “normal” traffic, another 40% is accidental.

via TweetDeck

And here’s another point about the prevalence of click fraud. twitter.com/MylesPeterson/…

via TweetDeck

@bastardsheep With Perlich’s German accent, I made that same mistake.

via TweetDeck in reply to bastardsheep

@hughstephens SHUT UP I AM SCIENCING BLOODY HARD HERE.

via Twitter Web Client in reply to hughstephens

It all comes back to that first question. What are you REALLY wanting to optimise for?

via TweetDeck

Want top click-thru rates? Show your ads on a flashlight app. Your audience is now people fumbling in the dark.

via TweetDeck

Bots are easier to predict than humans, which is how they found all this ad click fraud.

via TweetDeck

Perlich is astounding the audience with facts and stats on how botnets and other bad actors skew all this data.

via TweetDeck

Geolocation data isn’t as accurate as you might think. instagram.com/p/582gkTCFkz/

via Instagram

Being able to predict which creative will most likely covert a prospect is not quite current tech, but soon.

via TweetDeck

Different products have different time to purchase characteristics. instagram.com/p/581tNZCFj3/

via Instagram

… because we have multiple personas. Don’t try to sell me a holiday when I’m Googling SQL syntax for workMcs

via TweetDeck

For real-time bidding, the audience is less than one… instagram.com/p/581SS0iFjO/

via Instagram

Every product has its own data model.

via TweetDeck

The big data approach is just like that 2D chart before, and the same equations, but in 10 million dimensions.

via TweetDeck

We don’t care who you are or what it means. instagram.com/p/580geTCFiR/

via Instagram

“Let the data speak.” Down the rabbit hole we go! instagram.com/p/580NG7iFh_/

via Instagram

Perlich: “I can optimise to whatever you want,” so what is the question you want me to optimise to?

via TweetDeck

Start with this question: “Who do you want to show an ad to?” Because..Mcs

via TweetDeck

Perlich is attracted to this field because advertising has vast quantities of interesting data.

via TweetDeck

Perlich: I grew up in East Germany, so until I was 15yo I’d never even seen an advertisement.

via TweetDeck

Now: Machine learning for smart, targeted advertising: Claudia Perlich, Chief Scientist, Dstillery

via TweetDeck

Next in analytics stream: machine learning for targeted advertising; pre-emptive logistics. pic.twitter.com/Z8oqHYA6p8

via TweetDeck

bandt Women are more likely to lie on Twitter, but the blokes lie more on Facebook @adma pic.twitter.com/2swMvQ8u1e

via Twitter for iPhone (retweeted on 2:31 PM, Aug 4th, 2015 via TweetDeck)

“The 9pm Urgent Hardware Refresh” is now 64% funded, and 86% explicit or Malcolm Tucker. pozible.com/edict2 pic.twitter.com/UM8JvamA52

via TweetDeck

@iain_chalmers That, Sir, is the great joy* of it all, right?

via TweetDeck in reply to iain_chalmers

I have a “Now what?” that works for this context, @jturner_ibrs. Now we all just give up and open the gin. twitter.com/jturner_ibrs/s…

via TweetDeck

Eleventy billion points to @adma for equipping the media work room with its own cheese platter.

via TweetDeck

CSIRO has its Data Access Portal (DAP) with some petabytes of data that you can use, apparently.

via TweetDeck

This room seems to be running 20 minutes behind schedule. This does make jumping between streams… problematicMcs

via TweetDeck

CORRECTION: Square Kilometre Array (SKA) generates 30PB of data a day, not 20PB. twitter.com/stilgherrian/s…

via TweetDeck

Related: “IBM pushes Tivoli for cows and guns” (2011) zdnet.com.au/ibm-pushes-tiv…

via TweetDeck

Related: Pay heed to the internet’s Third Wave Cows of Disruption” (2013) zdnet.com/pay-heed-to-th…

via TweetDeck

Ethical issues? Based on its growth data, you can predict 6mo out which cow will be your steak.

via TweetDeck

“Each cow generates around 200MB of data a year,” and there’s a lot of cows.

via TweetDeck

@rtse999 Oh yes, I will fact-check all this stuff later anyway. Cheers.

via TweetDeck in reply to rtse999

rtse999 - when your house plants SMS you to water them - Arkady Zaslavsky

via Drafts for iPad (retweeted on 12:46 PM, Aug 4th, 2015 via TweetDeck)

@BXGD There is that, yes. It’s called a botnet, obviously.

via TweetDeck in reply to BXGD

I won’t tweet the details of these CSIRO projects. Use the internet and watch the videos, you lazy sods.

via TweetDeck

@CosmicRami There are live streams too, but an A380 has…oh, is it 100,000 sensors? 250,000 sensors? I forget which of the two now.

via TweetDeck in reply to CosmicRami

@CosmicRami It is taken from the engine upon landing and goes to the manufacturer (Rolls-Royce) for maintenance planning etc.

via TweetDeck in reply to CosmicRami

The main message here is that the overall challenge is real-time analysis of multiple IoT data streams

via TweetDeck

What if all the computers in smart cars parked in the city could be turned into cloud computing and rented out?

via TweetDeck

@rtse999 Oh, I heard 20PB/day. Is my hearing going?

via TweetDeck in reply to rtse999

Square Kilometre Array (SKA), and its 20PB of data a day, gets a shout-out.

via TweetDeck

An Airbus jet engine generates 10TB of data for every 30 minutes of running. What if we could analyse in flight?

via TweetDeck

Now: The effects of the Internet of Things (IoT) data explosion, Arkady Zaslavsky, CSIRO

via TweetDeck

CMU has open-source tools for processing and analysing tweets. ark.cs.cmu.edu/TweetNLP/ Ping @leslienassar

via TweetDeck

SA is domain-specific. “Earnings broadly in line with expectations” actually means they weren’t in line.

via TweetDeck

Further examples of sentiment analysis in action. instagram.com/p/58bav8iFgQ/

via Instagram

That Live Journal work was done 10 years ago.

via TweetDeck

Live Journal was good for training an algorithm, because each post is already marked with a mood indicator.

via TweetDeck

The upper limit to all this is the having an annotated corpus of text on which to train the algorithm.

via TweetDeck

SA version 4 is multi-dimension analysis, not just +ve/-ve, but strong vs weak, happy vs sad etc etc.

via TweetDeck

Some great examples in there of how this version-3 SA works in practice.

via TweetDeck

So, they ran this discriminator over the text of speech of politicians in the lead-up to the election, and… hehMcs

via TweetDeck

Prof Stephen Pulman explains how the computers can tell when you’re lying. instagram.com/p/58ZxR1CFtm/

via Instagram

Machine learning could be train up a discriminator to detect those trying to fake it with 80% accuracy.

via TweetDeck

In an experiment, people were given fake or real passports, and told to try getting through border control…

via TweetDeck

It also missed cues like gesture and so on.

via TweetDeck

SA can also be used to look for deception. There are reliable indicators, but textual analysis missed vocal cues.

via TweetDeck

Pattern-matching, but here’s only 150-200 ways of expressing future predictions in a language like English.

via TweetDeck

SA can be used to look for forward-looking language and risk-description language in company reports. Hard!

via TweetDeck

Also, +ve/-ve of words like “wicked” or “sick” depends on who the speaker is. “A sick night out.”

via TweetDeck

Each of those combines words all negative by themselves, but +ve or -ve depending on composition.

via TweetDeck

Example, compare “bacteria”, “kill bacteria”, “fail to kill bacteria”, “never fail to kill bacteria”.

via TweetDeck

This is still a very hard problem, but current systems get it right 85% of the time.

via TweetDeck

SA version 3 is a full linguistic analysis, using a parse tree and “compositional sentiment”.

via TweetDeck

Problems? “Clever” is +ve, “too clever” is -ve, “not too clever” is also -ve.

via TweetDeck

SA version two is a bundle-of-words classifier, on groups of words, ignoring syntax.

via TweetDeck

Also, “cold” is positive for beer, but negative for coffee. Must do more than count words.

via TweetDeck

Problems are that +ve / -ve don’t cancel out. “Not wonderfully interesting: is 2+ 1-, but still overall negative.

via TweetDeck

You’d be surprised how many commercial SA systems are nothing more than this, says Pulman.

via TweetDeck

You have a work list, classify them, then could the number of +v and -ve words in a text.

via TweetDeck

Sentiment analysis (SA) version 1 is a cheap and cheerful lexicon-based approach you can build in an afternoon..Mcs

via TweetDeck

Most common is classifying text as a positive, neutral, or negative reaction, but it’s difficult to do.

via TweetDeck

The term “sentiment analysis” has been around since 2000ish and has been used in several contexts.

via TweetDeck

EvanStubbs Common trend across the talks - the theory isn’t new, it’s that the hardware is finally making the theory possible

via Twitter for iPad (retweeted on 11:36 AM, Aug 4th, 2015 via TweetDeck)

Many techniques were known 20 years ago, but we now have the computer power and data to actually do it.

via TweetDeck

Pulman says this sort of thing isn’t pure machine learning, because we already know a lot about language.

via TweetDeck

Also, I majord in both computing science and linguistics, so for me this is utter pig-in-heaven stuff.

via TweetDeck

There’s been HUGE advances in computational linguistics in recent years, so this should be interesting.

via TweetDeck

Now: Sentiment analytics: detecting emotions in textual customer data
Prof Stephen Pulman,
Oxford University

via TweetDeck

Next up at , computational linguistics, sentiment analysis; big data of Internet of Things. pic.twitter.com/7FVDNT80Qb

via TweetDeck

Shit. I just realised I’ve only looked at 1/4 of the event program @Amda. There’s three other streams apart from . Idiot me.

via TweetDeck

paulwallbank “Marketers want to be intimate with you.” It’s moments like these I’m glad I’m not sitting next to @stilgherrian

via TweetDeck (retweeted on 10:20 AM, Aug 4th, 2015 via TweetDeck)

So @linode tell me they’re getting further reports of card failures, so the problem with my @Westpac card isn’t me. twitter.com/stilgherrian/s…

via TweetDeck

Let’s hashtag this. I’m at the @adma Global Forum in Sydney Tue and Wed. adma.com.au/adma-fomma/pro… twitter.com/stilgherrian/s…

via TweetDeck

And here we are at the adma Global Forum… Already I can see that geo and location based services..instagram.com/p/58LFnWiFlB/mH

via Instagram

Walking through Thaitown towards the Hilton for day one of the @adma Global Forum.

via Plume for Android

“Good morning, welcome back.” Recognised, checked in, upgraded, exchanged gossip, given drinks vouchers. Elapsed time three minutes.

via Plume for Android

@haruki_zaemon Howard’s tracksuit was used while walking, had AU branding. Abbott’s sporting gear has drug company logos. So that.

via Plume for Android in reply to haruki_zaemon

Arriving at Sydney Central station instagram.com/p/58EYSZiFoV/

via Instagram

Crusader Rabbit loves his uniform dress-ups, doesn’t he. Can our next PM be psychologically older than 15, please? twitter.com/NewtonMark/sta…

via TweetDeck

“Princeton Study: U.S. No Longer An Actual Democracy” talkingpointsmemo.com/livewire/princ…

via TweetDeck

@joannejacobs @kcarruthers MEL doesn’t have a rail connection either. Or BNE. Or ADE. Or PER. Or CBR.

via TweetDeck in reply to joannejacobs

@semibogan It turns out our police can’t shoot the merely annoying. But building a little rapport with anyone carrying a gun seems sensible.

via TweetDeck in reply to semibogan

@semibogan True. But you could do what I do and ask first. “‘Scuse me, officers, would you be allowed to shoot these annoying people?”

via TweetDeck in reply to semibogan

@juhasaarinen @mgherman Oh it’d more that I happened to have had briefings from those two bank and they’re on much the same page.

via TweetDeck in reply to juhasaarinen

dr_teng pictured : konami employee going to lunch pic.twitter.com/GQ0jiaf5NG

via TweetDeck (retweeted on 8:12 AM, Aug 4th, 2015 via TweetDeck)

@juhasaarinen @mgherman Was this the yellow one, or the red and white one?

via TweetDeck in reply to juhasaarinen

trinnybopper My mate got bitten by a pig yesterday. I don’t know why I think it’s so funny but I do

via Twitter Web Client (retweeted on 8:08 AM, Aug 4th, 2015 via TweetDeck)

Maybe @BuzzFeedOz needs such signage for a range of public figures. twitter.com/Rob_Stott/stat…

via TweetDeck

jonrowlandson Pac Man is in a bad place these days. pic.twitter.com/ZXite8thGQ

via Twitter for iPhone (retweeted on 8:06 AM, Aug 4th, 2015 via TweetDeck)

@jeffsonstein I am so going to borrow that analogy. With due credit, of course.

via TweetDeck in reply to jeffsonstein

@mgherman @juhasaarinen Sure. I’m thinking of Coles/Woolworths are huge supermarket chains, which are also, really, logistics operations.

via TweetDeck in reply to mgherman

@mgherman @juhasaarinen Otherwise you end up in the daft world of Aussie Home Loans as a telecommunications company, not financial services.

via TweetDeck in reply to mgherman

@mgherman @juhasaarinen The big-pciture issue is that I’m pushing back against the idea that working on the internet makes you “tech”.

via TweetDeck in reply to mgherman

@mgherman @juhasaarinen You’re starting to get it. Amazon is a retail operation, a shop. Netflix is a media company. Yes, online.

via TweetDeck in reply to mgherman

@michaelneale @juhasaarinen @mgherman We’ve probably got three categories. Sci/tech; startups and business innovation; consumer electronics.

via TweetDeck in reply to michaelneale

@michaelneale @mgherman @juhasaarinen Ah yes, that post sums up a lot of my thoughts. Maybe I should do my own version.

via TweetDeck in reply to michaelneale

@1Password Oh. Thanks again. [Shrug.] Well, @Westpac is refusing transactions from @linode, so some deeper glitch is at play here.

via TweetDeck in reply to 1Password

@1Password @Westpac @linode Thanks. I hadn’t seen this kind of warning before. A relatively new feature?

via TweetDeck in reply to 1Password

It’s this daft categorisation that allows Uber to run the spin that they’re not a transport company. @michaelneale @mgherman @juhasaarinen.

via TweetDeck

@michaelneale Just like Network Ten is an incredibly heavy tech-using media company, yes.

via TweetDeck in reply to michaelneale

@mgherman @juhasaarinen Well, how would be define “an X company” for any other value of X? All companies use, and many improve “tech”.

via TweetDeck in reply to mgherman

Hmmm. @Westpac is still refusing to let me use my card at @linode. And @1Password had a “change password now” warning.

via TweetDeck

@juhasaarinen As internal tools. What they SELL is advertising. What Arnott’s sells is biscuits, not production-line machines.

via TweetDeck in reply to juhasaarinen

I disagree with @mgherman’s point. Many companies build their own tools. But what Facebook sells is advertising. twitter.com/mgherman/statu…

via TweetDeck

Facebook is a much a tech company as Arnott’s Biscuits is a mechanical engineering company. “Using computers” doesn’t make you “tech”. (2/2)

via TweetDeck

“Facebook is the undisputed king of the tech industry”, reports @BIAUS. businessinsider.com.au/facebook-rules… Except… (1/2)

via TweetDeck

Tue plan; 0706 train to Sydney (now); @adma Global Forum adma.com.au/adma-fomma/pro…; various writing tasks; 2030 ABC @702sydney; collapse.

via Plume for Android

rstevens Soylent CEO walks into a bar-

“Eww- nuts? Food is inefficient. I want my beer half price because I did not consent to the cost of your TV”

via Tweetbot for Mac (retweeted on 7:01 AM, Aug 4th, 2015 via Plume for Android)

Current temperature at Wentworth Falls station: -2.2C, apparent -6.2C. instagram.com/p/573T7lCFu-/

via Instagram

The morning rush begins, Wentworth Falls. instagram.com/p/573GXpCFuo/

via Instagram

The morning rush begins, Wentworth Falls. pic.twitter.com/zssWJlRfhl

via Plume for Android

AshiLabouisse Day 19. I have successfully conditioned my master to clap and write in his book every time I drool. What an idiot.
~ Pavlov’s Dog

via Hootsuite (retweeted on 6:26 AM, Aug 4th, 2015 via TweetDeck)

@manicode @csoandy @JetBlue I’m happy for the vendor of magic flying armchairs with bar service to make whatever rules they like.

via TweetDeck in reply to manicode

We’ve just gone to 85% Malcolm Tucker on 43 minutes of your content. pozible.com/edict2 (2/2) pic.twitter.com/BKpbj4RLgI

via TweetDeck

Less than 2.5 days remain in “The 9pm *Urgent Hardware Refresh” @Pozible campaign. pozible.com/edict2 pic.twitter.com/ABhdMP2dZB (1/2)

via TweetDeck

@manicode @csoandy @JetBlue It was a LAX to SYD flight, though the smell finally subsided about 3 or 4 hours into the 15-hour flight.

via TweetDeck in reply to manicode

@manicode @csoandy @JetBlue Exactly. I’ve seen a family removed from an aircraft before it departed ‘cos a kid threw up. Plane still stank.

via TweetDeck in reply to manicode

@csoandy @manicode @JetBlue Pretty much, yes. Though the passenger did say it was an upset stomach right? HTFU and follow procedure! ;)

via TweetDeck in reply to csoandy

@csoandy @manicode @JetBlue Sounds like standard procedure to me. Passenger reports potential medical problem, so won’t be flying.

via TweetDeck in reply to csoandy

Tue plan, draft; 0706 train to Sydney; @adma Global Forum adma.com.au/adma-fomma/pro…; various writing tasks; 2030 ABC @702sydney; collapse.

via TweetDeck

Tuesday. This time, They have assessed the risk correctly. They press the Button. Your blood freezes. You scream. They laugh. Tuesday.

via TweetDeck

Monday. This time, They have assessed the risk correctly. They press the Button. Your blood freezes. You scream. They laugh. Monday.

via TweetDeck