Numi looks incredibly fresh

Numi, is calling itself a Beautiful calculator app for Mac, and it caught my attention the other day so I thought I'd share it with you guys. I tried playing around with the app, and while it makes for extremely simple and impressive demo on it's website, I quickly ran up against limitations.

Natural language is a really hard thing to parse, it often requires wither complex algorithms and machine learning like what you find in Siri or Google Assistant for example (both services not always known for nailing exactly what you mean), or solutions require a seemingly endless database of brute-force rules to parse, making things pretty slow, and easy to break.

My Numi experience started very well. I downloaded the app from it's website, with no ask for any information from me, and within seconds it was in my downloads folder. Then upon first boot, the app offered to move itself for me. I really can't say enough how pleased I was with this operation, too often a clunk part of app installs on Mac involve putting something that was in my downloads folder and remembering to move it into my applications folder.

Nice touch

Nice touch

It was only two lines into writing my own math that I started to hit the limitations of the app. Numi either didn't under what "Square root of 900" was, or I wasn't saying it in a way it expected. In another statement, Numi was also confounded by the question "What's a 16% tip on $400?" but felt confident enough to answer "$464" to the words "16% tip on $400". The worst failure of the calculator was when it failed to put anything on screen for the phrase "eight plus one".

Do you expect your calculator to answer eight plus one?

Do you expect your calculator to answer eight plus one?

I'm not saying Numi is bad. I don't think it is, in fact I think there's a lot to like about it. Dmitry Nikolaev has a lot of good thinking that went into this product, and when it does work, it's really an excellent way of working with a calculator. It felt better to me, a light math user, than using even the default calculator on macOS or iOS. It also calls into the question the singular states of default calculators by showing multiple problems and solutions on a single page — one of the apps beast features.

What I am saying is that it's an application that highlights just how hard it is to make computers understand what people are trying to say is an easy to implement feature for devs, and how failure to parse any single piece of information with an invisible interface usually winds up with the user just giving up on it's reliability all together. I can't help but wonder how this thing would work if it was powered by one of the large AI assistants out there. It's simply to easy to make this thing fail, so I won't be adding it into my workflows unless it becomes more robust. Or at least, show error states and give me tools to correct my own syntax.



I made a new Giant Bombcast Theme Song

I'm a listener to a bunch of different podcasts, one of my favorite is the one from A weekly roundtable on the latest in video games. I noticed a history of a bunch of various themes over the years, but in recent years, the theme has remain unchained. I wanted to create something that felt like it fit into the history of Giant Bomb themes, without being too overly reliant on guitar.

My favorite mix (above) has synths from all sorts of places, but some may notice the lead pulled from Super Metroid. I also created an alternate mix with some Japanese Koto on it, to harken back to the home of modern console gaming.

Below is a YouTube video with various themes throughout the years, some of which I really like, so I wanted to create something that sounded in the vein of what I considered the "main theme" of the show, and remaster it with some clean synth production.

The current theme:

Older Themes:

Mike Monteiro on Education and Ethics

I know, I know. Another Medium article, but this one has some salient points. Especially how continuing to create service-based products further reinforces income inequality.

If we achieved income parity, the tech sector would collapse! You can’t build an economy on the need for a class structure and then act surprised when it results in a class structure. Is this the kind of system you want to be supporting with the short amount of time you have on earth?

The piece also dives into design ethics, a topic I'm pretty passionate about, even if it diverts from the initial point of the article a bit.

And as our work gets more complicated, the ethical concerns about what we’re doing get bigger and nastier. (Look no further than all the shit we went through with Uber.) And I’m out there talking to a lot of designers. I’m genuinely scared at the amount of them who don’t see ethics as a part of the craft. This needs to change.

Adam Silver on Inline Validation

Looks like Adam Silver and I feel similarly about Inline Validation. It just isn't good enough in most instances, he provides a ton of examples in this medium article. He even goes on to make a few points I missed, including this thought on false positives:

Some implementations use inline validation to show green ticks when the user fills out a field successfully. This may provide a sense of progression and stop people feeling like they need to check through their answers later on.