Ask the CTO

Nate Glissmeyer is Chief Product Officer (CPO) and Chief Technology Officer (CTO) at HolidayCheck Group AG and explains in the following interview, among other things, how artificial intelligence is used in the holiday rating portal and why only five lines of code in the Python programming language can cause goose bumps.

Please explain briefly what you do as CTO: What are your tasks and what is the vision of Holidaycheck?

Our vision is to become the most holiday-friendly company in the world. I always say, as CPO and CTO, I help decide what we are going to do for our holidaymakers and how we are going to do it. The CTO role is very strongly about how are we building & running the business every day. Is our vision clear enough that every person in engineering knows how to get started right now realizing that vision? If not, I haven’t done my job yet.

It is very heady to be responsible for everything from spearfishing attacks to our own APIs to Mergers and Acquisitions of tech in other companies. I am often humbled by the mix and the level of breadth and complexity. My team does a great job of giving constant feedback and we do “tea time” sessions where we can sit together and very openly talk about how to improve things, which gives me lots of access to what it feels like to work here. I love that part, too.

GDPR's coming into force in 2018 brought the classic IT topics of security and compliance into focus. In your opinion, what are the advantages or disadvantages of this law?

Clearly, Germany leads the world in privacy and the right to privacy. The inalienable rights under GDPR are courageous and ensure people can unwind future, unforeseen loss of rights. I love that. I am also proud that German DSG was such a blueprint here.

But, like most laws meant to govern technology, the real questions about how things work and the real insights, the really elegant solutions for managing tricky questions, do not seem to come from lawmakers. GDPR offers little to no real advice and the more you know about technology and data usage, the less useful the statements in there are and the more you have to go your own way. I hope that “use makes master” and running tech in a post-GDPR world will lead to practices that work. As a board member, I am just nervous that I am going to have to pay this price for others.

Artificial intelligence is on everyone's lips and is being used in more and more areas these days. Can you give us an example of how you are implementing this technology in your company?

Holidaycheck has 20 years of long-form reviews. Not stars or thumbs pointing up and down, but actual narratives, with real insights into the hotel, the vacation, and the person writing. This evocative, deep material is starting to be digestible to machines, which just blows my mind. We have a model in beta right now that can summarize a review in the person’s own words and highlighting a poignant quote. That is wild! Imagine a good editor picking out the essential point of an argument and then realizing that impressive editor is five lines of python. Gives me goosebumps!

One question at the end: Why did you choose your profession and can you remember the first project you ever programmed?

I never picked my profession; I just started doing it. That has been a bit of a surprise, but it has always been my way. Tech was the best tool for the job (teaching German at the University of Utah), so I used tech.

Yes, my first substantial program was modeling the expected order volume for Amazon so that we could have a “performance against expectation” baseline for monitoring. I started working on it in May, 1997. Until then, people would just page my pager when the site felt slow (Jeff and Mackenzie Bezos included).

I forgot to empty the array of values that modeled different days of the week before modeling the values of subsequent days and my first test run, which led to seven astonishingly similar charts. I rushed into my boss’s office to tell him we had a real problem. He stacked all seven on top of one another, held it up against the light, and said, “you have a bug, not a problem.” It was then that it was clear I was cut out for management, not writing lots of software!”