Profile picture
Patrick McKenzie @patio11
, 3 tweets, 1 min read Read on Twitter
I use this app which is the Japanese equivalent of Mint to aggregate and categorize transactions from all the cards/accounts/etc, and because machine learning, 30 transactions for “Starbucks Harajuku” got flagged as “fashion”, to the great amusement of my wife.
Harajuku’s four industries: fashion for young Japanese women, tourism for 20-something foreign tourists, extremely sugary snacks, and tech startups.

(No really; that last one is legitimately why we put our office there.)
I wonder if I’m second guessing a computer which would say huffily “OK human it is literally the case that more receipts I see with Harajuku on it are fashion that receipts I see with Starbucks are for coffee can you even math human they are a diversified lifestyle brand.”
Missing some Tweet in this thread?
You can try to force a refresh.

Like this thread? Get email updates or save it to PDF!

Subscribe to Patrick McKenzie
Profile picture

Get real-time email alerts when new unrolls are available from this author!

This content may be removed anytime!

Twitter may remove this content at anytime, convert it as a PDF, save and print for later use!

Try unrolling a thread yourself!

how to unroll video

1) Follow Thread Reader App on Twitter so you can easily mention us!

2) Go to a Twitter thread (series of Tweets by the same owner) and mention us with a keyword "unroll" @threadreaderapp unroll

You can practice here first or read more on our help page!

More from @patio11 see all

Profile picture
I’m often surprised when reading about startups that critics move to envisioning a world where the startup has won so totally that there is no alternative to their model, which is not what the victory case looks like.
Thought prompted by (paraphrase) “Who will cross-subsidize the education of non-renumerative-occupation-of-interest after Lambda School is educating every professional in capitalism?”
And as much as I share enthusiasm for living in that hypothetical universe good golly there is a heck of a lot of work between the status quo and the first 1% of that outcome.
Read 4 tweets
Profile picture
Using a Chromebook for the first time and reflecting that given how much of my workday is in Google applications and other generic web apps this might be a perfectly adequate main computer if one isn’t doing development (on which I have no read on it).
I always thought “You’ll pry MS Office from my cold dead hands” but these days 98% of what I use it for is handled by Google Docs. (It helps that almost all docs I care about are internal and began life in Google Docs.)
Also Google seems to have discovered a revolutionary technology called a “keyboard” where you mechanically actuate a switch and it results in a character of input to your machine.

Apple should steal that.
Read 4 tweets
Profile picture
One of the least heralded and most important technological advances of the last 20 years is a battle getting fought one piece of paper at a time to replace workflows with web applications. It’s slow, boring, and enormously impactful.
It’s surprising how much you can get done these days *without* needing to call anyone, line up, or make a visit to an office to figure out the correct form to file something.
Random example for you: the IRS hires 10k people every year to type numbers from returns into the computer. This is a decrease from earlier, since e-filed returns don’t need to be redundantly retyped.
Read 4 tweets

Related threads

Profile picture
Because of the currently ongoing potential 51% attack on #ETC we have temporarily increased block confirmations on our ETC mining pool. We will notify you once the network has stabilized again
We can confirm that there was a successful 51% attack on the Ethereum Classic (#ETC) network with multiple 100+ block reorganization. We recommend all services to closely monitored the chain and significantly increase required confirmations.
Take care, the Ethereum Classic #ETC network is still being targeted by a 51% attack. Today we saw again 3 deep chain reorganizations: pastebin.com/xtu1a1Kg
Read 3 tweets
Profile picture
Learning about surface vs deep learners. @lsu Teaching Colloquium. Thinking about implications and applications of learned information.
Learner needs to regard info as important, intriguing, or beautiful. Then deep learning occurs. Learner needs to be in charge of the question. But, few students will not take a deep approach to learning.
Students are conditioned to take a surface approach to learning. School is the biggest place that conditions surface learning. #KenBains #learning #whatTheBestCollegeTeachersDo
Read 22 tweets
Profile picture
1/
 $DRGN shameless speculative shill thread 🧐

⁉️Starbucks first Fortune 500 adopter of Dragonchain⁉️

#starbucks
#bakkt
#btc

Today ICE announced Bakkt, with big partners including Starbucks.
2/ Maria Smith, VP and Partnerships and Payment Strategy at Starbucks, who also happens to be the Financial Executive at @dragonchaingang is involved. Starbucks will provide seamless BTC to USD conversion to pay for items at Starbucks.
3/ What does Starbucks need to develop this in a decentralized scalable way? A scalable, hybrid private/public Blockchain solution with BTC smart contract and interchain capabilities.

What is on the roadmap of @dragonchaingang ? Right:

BTC Interchain Smart Contract Library
Read 6 tweets
Profile picture
My experience with learning modes - Text vs Video content

- Due to concentration lapse, it's easier to drop off for a few minutes from a video content (> 5 mins)

- Concentration lapses happen in text too, but its easier to re-read the para vs rewinding the video

#learning
- For consuming the same amount of content, text form is less time consuming than videos.

- To explain complex topics/situations, texts usually take more time building up the context than videos.
- We try to understand things by building mental imagery, that way videos are more efficient than text because it takes a bit of time to build the mental imagery from interpreting the text and yet you might not be building the right imagery.
Read 5 tweets
Profile picture
I suspect automated translation between s/w standards that were not built with awareness of each other is harder than translating between 2 languages.
Ie people arguing that 2 Google-duplex-class AIs talking to each other would switch to more “efficient” machine talk once they detect that the counterparty is also a machine are wrong.

This is only true of 2 agents that speak the same language and know that the other does too.
Consider how humans communicate across a language barrier. Me and a Chinese speaker for example (I don’t know Chinese). Only 2 outcomes:

One of us asks “do you speak English?” and we switch to that if yes... OR

We do painful pointing-and-miming to develop minimum common ground.
Read 13 tweets
Profile picture
Machine learning weirdness: a youtube video about a group of Moroccan children living on the streets (fr)
then..
followed by one about a Dubaï rich++ kid who spends 100s of k$ for shoes, his own personal zoo, his youtube chanel..
I really wonder what kind of features led the rich kid video to have been ranked as relevant when I watched the first one.
Read 3 tweets

Trending hashtags

#SafeRouteNow #whatTheBestCollegeTeachersDo #smarticities #SafetyOverSpeed #BanDigitalElections #accessibility #infrastructure #smartcitirs #HappyCities #ChristmasMorning #HouseOfCardsQ #rstats #TFLite #TensorBoard #bikeNOLA #unroll #OSS #DriverNotCar #VR #bikes4climate #AlphaFold #sustainability #DataValidation #ETC #JavaScript

Did Thread Reader help you today?

Support us! We are indie developers!

This site is made by just three indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member and get exclusive features!

Premium member ($3.00/month or $30.00/year)

Too expensive? Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal Become our Patreon

Thank you for your support!