Design 2: From Relational to Distributed

Theme

The previous puzzle Open Source Basics was meant to be approachable to anyone with programming experience. With this puzzle, I wanted to explore database history, which is a topic I find really interesting. It's fun reading about the companies and people involved in shaping modern databases.

However, this is not a topic that most people encounter on a daily basis, so it is less accessible than the previous puzzle. We use databases all the time, and a lot of how we think about them is shaped by foundational ideas created by Codd and early database research.

Later, Google popularized large-scale distributed systems to deal with high-volume data at a global scale. We use many of these tools today, but the historical elements are not often discussed. Databases have a rich history in computer science, so I wanted to focus on them.

Grid

Similarly to the first puzzle, creating the grid for this puzzle was a bit difficult.

There was a lot of trial and error in designing the layout. It's difficult to predict how a puzzle will look until all the clues are connected. Often I would be close to finishing a layout, then struggle to add the last couple of words. Other times I would finish a layout but the final result did not look how I wanted. In those cases I would start over to find better intersections.

I use Phil to draft my grid layouts. The process involves manually writing the clues and intersecting them.

Finding intersections themselves was not the hard part since the answers were not too long and shared many common vowels. The main complexity was making the grid look balanced. Since this topic is even more technical than the last, it was important that the grid still felt approachable.

Originally I was worried the grid would feel too sparse since this puzzle contains only 11 clues compared to the previous puzzle’s 14.

Clues

Coming up with clues for this puzzle was also challenging. I ran into creative trouble around 8 clues. I originally wanted to reach 14 clues, but after thinking about it for quite some time I couldn't find additional clues that fit the theme.

SQL, Oracle, IBM, and Codd formed the main 1970s historical theme. Bigtable, Hadoop, and Mapreduce represented the early 2000s era of distributed data systems.

Clues like Memcached, Cap, and Sharding were later additions that helped expand the clue list while still preserving the database theme.

Stonebraker ended up being the final addition. His work on Postgres fit the theme well and made him a strong inclusion, but I was initially hesitant because PostgreSQL appeared in the previous puzzle. I didn't want repetition fatigue, but I ultimately decided to still include him.

Tradeoffs

The historical connections between IBM, SQL, Codd, and Oracle are not as prominent in the puzzle as I would have liked. This is mostly due to changing the clue ordering while designing the grid, as well as simplifying clues so they were easier to read.

Notes

I did learn to be more aware of clue length when writing clues for this puzzle. Originally I increased the character limit from 150 characters to 250 characters for clues, but quickly realized this made clues dense and difficult to scan. I reverted the change and made clues more concise, even though it meant removing interesting details.