Speros Kokenes
@sperosck
Followers
1K
Following
9K
Media
429
Statuses
4K
@MalloyDev. Maker of @knarr (acq by Qlik). Ride or die with @PolicyPeach.
Atl
Joined July 2013
4 months ago I joined the Malloy team to work on an interesting data challenge: how to visualize complex, deeply nested data structures in a consumable way. Here is where we are at today, and where we are going 🧵
4
10
163
Incredible piece of writing
My latest newsletter is about why the “male loneliness crisis” isn’t about loneliness, how casino capitalism’s conquest of media and finance is just getting started, and why those two things are connected. On America’s monks in the casino: https://t.co/ijNwLcz1fh
0
0
2
probably my most deeply-held belief: it is not possible for a slide deck to be both (1) a good presentation aide; and (2) work well as a standalone document that can be read and understood in isolation.
35
44
682
Whats up with cursor agents not making commits lately. Happens to me all the time, I try to open a PR and then have to go back and remind them what to do
0
0
0
you know you're in hot water when claude starts writing debug scripts for its debug scripts
0
0
1
I worked with a designer from the Jobs era, & he was adamantly opposed to any kind of design system. Their workflow was he'd design an app (he did Keynote), then there was a graphic designer who would produce a sticker sheet of the pixel perfect UI els for it. always contextual
stay true to the medium the “liquid glass” approach to design is peak “design system” gone wrong — when you optimize for visual consistency over interaction fidelity, you lose the soul in each medium. each input method has its own physics and affordances. touch needs bigger
0
0
2
Forbes 30 under 30 has been doing this for years
642
3K
32K
Is that dude holstering a loaded gun, lmao. Genuinely curious what the legal implications are of one of these guys murking someone on your property
Introducing Patrol. Homeowners in LA can now book off-duty police officers to help protect their homes. Now available in Brentwood, Beverly Hills, Bel Air, Holmby Hills, and Malibu — more areas coming soon.
0
0
0
Lately I've been playing around with building my dream analytics table. I'm currently working on a feature I've wanted forever in a BI tool but have never seen taken all the way: natively using "sets" as first class dimensions
1
2
18
There's so much more to do with sets: - using them as first class filters for any dimension/measure - flexible set membership at table vs. group by level - derived sets through union, intersection, diff, or symmetric difference operations
0
0
2
And nest them within other set lists, analyzing set intersections. Here we analyze deal hunters vs. entire business, then break out by marketing campaign type within each set
1
0
2
Back to ecommerce: imagine deciding a marketing campaign target. You might consider metrics across different segments: - repeat guests w/ multiple purchases; - women who never bought fashion; - products never offered at significant discounts. Group them like any dimension:
1
0
0
This makes true set analysis difficult when comparing various segments. To make this easy, we need "synthetic" dimensions representing lists of sets. These set lists should work exactly like natural dimensions: in grouping, nesting, pivoting, filtering
1
0
0
Natural dimensions are columns in a table. Tools let you aggregate and filter with these columns. By definition, when aggregating by a dimension, a record can only exist in one column value at a time
1
0
0
The key thing: these sets aren't mutually exclusive. Customers and products can exist in multiple sets simultaneously. This presents a problem in typical data tools that only aggregate based on what I call "natural dimensions"
1
0
1