Tech

Book review number 250

This is my two hundred fiftieth review on Goodreads since 2010. You can read them all, fiction, non-fiction, and all sorts of other stuff, at this link:
Rich Magahiz’s reviews on Goodreads


You Look Like a Thing and I Love You: How Artificial Intelligence Works and Why It's Making the World a Weirder Place

You Look Like a Thing and I Love You: How Artificial Intelligence Works and Why It’s Making the World a Weirder Place by Janelle Shane
My rating: 4 of 5 stars

This was a solid piece of work on the state of machine learning which is so rapidly changing that one of the main platforms she uses in her examples, GPT-2, has already been replaced by another one with capabilities a hundred or more times that of its predecessor. So while it gives a good account of the challenges in applying the technology, it will only be a snapshot in time of how far researchers have been able to go (until it is revised perhaps some day). It draws on her popular blog with examples of her own experiments along with descriptions of what other researchers have been working on as broadly as can be accommodated in a work aimed at a general audience. Although I do have some involvement with deep learning in my job, my own interest is that of someone who is interested in the technology in a general way.

See something, say something

High speed highway
© creativecommonsstockphotos ID 92160937 | Dreamstime Stock Photos

I’ve been keeping an eye on the some items that

  • Something on a little bit of fossilized legal speech you have probably heard at some time.
  • In my past life I did a few years of work on neutrinos, enough to find it that something as small as the Earth could manage to absorb them when conditions are right.
  • True life crime can be more crazy than what thriller writers think up. The story of Susan Kuhnhausen.
  • You can search for any term you like in the Trump Twitter Archive to find out what our leader has opined on the subject.
  • Some of these trilobites look ready to get up and swim around.
  • I could watch these traffic intersection simulations
for far too long. * Here is an exhaustive set of metrics on where to spend your [retirement](https://wallethub.com/edu/best-and-worst-states-to-retire/18592/#main-findings) around the U.S. Some of the scores I find disappointing. * I am of the right age to know about the [B-sides](http://www.slate.com/articles/podcasts/hit_parade/2018/01/how_does_a_b_side_become_a_chart_topping_hit.html) they're talking about in this podcast, but not obsessed enough to know the right answers to the trivia questions.

I actually had a few more, but these are the best.

Kind of great

Tic tac toe beads
© creativecommonsstockphotos ID 89250373 | Dreamstime Stock Photos

Instead of passively consuming the nearly endless stream of content coming my way I thought I’d gather together a few items to share with everyone here.

  • Forty-nine years ago this happened:
* I particularly like this polyphonic visualization * We were moved when we saw the remnants of the Berlin Wall last October, so I was pleased to see the site for the online [memorial](https://www.berliner-mauer-gedenkstaette.de/en/) * It's been over two decades since I was passed over for tenure but the feelings this left behind are still pretty much as fresh as what I see in [this account](https://www.chronicle.com/article/a-professor-and-his-wife-on-absorbing-the-shock-of-tenure-denial/) of the experience. I wrote an account of my own experience [on Quora](https://www.quora.com/What-is-it-like-to-be-denied-tenure-as-a-professor/answer/Richard-Magahiz?srid=XXIY). * Along the same lines as the tongue in cheek stories of [The Codeless code](http://thecodelesscode.com/contents) (unfortunately no longer live on the web) are [The Unix Koans of Master Foo](http://web.archive.org/web/20180305224708/http://rationalfiction.io/story/the-unix-koans-of-master-foo) * [Psychological safety in operations teams](https://www.usenix.org/system/files/login/articles/login_winter17_09_looney.pdf) * This blog post was inspired by [52 things I learned in 2017](https://medium.com/magnetic/52-things-i-learned-in-2017-d9fb0040bdcb)

A cryptocurrency windfall

Bitcoin
© Publicdomainphotos ID 102950205 | Dreamstime Stock Photos

A couple of years ago I was working for a company that allowed me to sit at a desk in their office in midtown Manhattan in lieu of relocating all the way across the country. It was still a kind of shock to adjust to the New York style of life after so many years working for sedate firms in northern New Jersey, with the commute by bus and the walk across Times Square and the standing in line for lunch at one of the dozens upon dozens of eateries within a short distance of the office, and the tech meetups around town. Now that the economy was picking up after the shock of the late 00s there were plenty of exciting things going on in the tech world, many of them right there in the country’s heart of finance. Around then is when I first started hearing about bitcoin when companies like Adafruit and Overstock were beginning to accept this cryptocurrency as payment for their merchandise. People had even written phone apps which allow you to conduct transactions right from your smartphone by then, which sounded both great and worrying considering the rise of mobile phone dependence right around the same time. There were articles describing how blockchain-based currencies could decentralize the global monetary system independent of central banks, and help eliminate middlemen which hinder micropayments. So, I decided I’d take a small amount of money and find out where I could get my own stake in bitcoin.

What I've done

Jelly Boats

Here is a list of what I have put together or worked on for some significant amount over the last couple of years on the job.

  • Monitoring will get its own post eventually.
  • We migrated from a single service which ran on a single instance in Classic EC2 to a half dozen services running in separate VPCs, each with autoscaling instances behind an elastic load balancer to provide high availability. Getting all the plumbing on this right was a major effort.
  • We also shifted our configuration management system from exclusively Puppet at first to Chef, where we could take better advantage of community-written cookbooks, and then brought
  • Wrote an automation script which packaged service config files into a Debian package which would simply be pulled into a new instance like any other versioned package.
  • Worked on a couple iterations of an ELK stack in order to accept logs and structured data
  • Prototyped a stack which used Amazon IoT, Kinesis Firehose, DynamoDB, and S3 to accept mobile data streams and store them for business intelligence. The components were held together with Simple Notification Service and Lambda functions in a serverless fashion.
  • Hacked out numerous one-off scripts in bash and Ruby to automate the recurring maintenance chores in our environments. For instance we needed something to clean up user logins which were not needed any more, scripts to take reference snapshots of volumes for safekeeping, and so on.
  • Worked on some scripts to transfer information from our services to a data warehouse running on Amazon Redshift (so-called ETL scripts) so we would have an analytics service for our own use.
  • Migrated some administrative tools off of dedicated virtual machines onto containers which would run under Rancher. The idea was that we could pack a bunch of light-duty tasks together on a cluster of instances which would run the same workloads with much higher density, reducing costs and maintenance.
  • Wrote a simple Sinatra app which would gather information from our environments (addresses, instance IDs, running state, and so on) and display them in a tabular format. This saved me a lot of time switching between different dashboards and running things through the command line interface just to keep things oriented.
  • Kept track of our hosting costs in order to find out where we could benefit from terminating instances or phasing out legacy services.

Snapping pics over people's heads

In 2016 I went to a bunch of technical talks, none of which I intend to discuss right now, and all of which armed with a cameraphone along with everybody else in the audience. In this day of Slideshare and official corporate tech blogs many of the presentations will make it up to the web in pristine form, so why would anyone want to take a crooked, out of focus, keystone distorted, and sometimes half second too late picture of the screen?