I wrote this to Daring Fireball after John Gruber posted a note about David Pogue’s blind testing of HomePod. I realized this morning I could be clearer or more complete in a relatively few words I’m typing out now. First, the email:
‘I respect David – musician and all – but he doesn’t grasp blind testing. Blind testing has the effect of making individual elements stand out without context. It adds a form of randomness, something statisticians know through ideas like power and the n of iterations. Run a blind test over and over and you may get something useful, like this mortality improvement actually occurs, but you also get a lot of randomness because each iteration approaches randomness for that context. This doesn’t mean the speakers sound the same or that any particular blind ranking of preference is better than another. It means that in a blind test these are typical results when – this is important – the preferences you develop when listening specifically over time is removed. Listening over time is the same as adding power and n iterations statistically: a picture develops that has meaning. Blind testing can be worse than A-B testing because it strips out the meaning that comes with power and iteration. With subjective measures of ‘like’, each path through the choices tends to be the same as any other path within certain limitations. At this point, I can invoke ESP: take away all possible context so there is choice of say 5 colors or shapes and you get results that fluctuate. The difference to ESP is you can’t learn to identify the next shape, meaning the iterations are entirely separate, while there is enough contextual detail in a continuing sound blind test that some people may learn. The interesting thing is this doesn’t mean they learn the best answer, just that they learn an answer. Think of all the studied products that flop and it’s only when they flop that you realize the answer generated by all that market research was wrong, that it was the pathway which for some reason was the one taken. But that gets into the philosophy of statistics in a deep way.
My impression of HomePod is that they’ve done the following: a) they recognized that nearly all people now have learned to listen to music through earphones or headphones, and b) they’ve recreated the earphone/headphone experience for a room. I like to say they’ve moved the ears from earphone to the ear room. The key, I think, is they’ve mastered a big chunk of the algorithms for adjusting relative pitch as loudness changes – since if you make some pitches louder, they sound louder than another pitch made the same amount louder unless you curve the response. They appear to have implemented this largely through a volume slider. I may be wrong – this is intuition – but I think one reason they bought Beats is Beats had actually been attacking this problem well; their headphones were made to match the experience of a listener taking in music at various volumes in various settings. You can hear it: if you turn up Beats, they give you bass but they keep the bass in the right relative arrangement when you turn the volume down. I surmise Dre and Iovine et al would be able to hear that way because that is what they do so well! Beats really was built by music producers instead of audiophiles and engineers.
My take on HomePod is that it now competes with my iPad for best Apple device I’ve ever bought. And the iPad changed my life. (It has become my computer and much more.)
I expect they developed HomePod to be played through your phone because your phone has a perfect volume slider and you can adjust each track and within each track to your liking, as you do with your earphones now. They needed voice, but the ideal listening method is to fiddle with the volume slider. Why? Because they also realized people can listen not only to just about anything but that they listen in playlists of songs and pierces by various artists, each recorded differently, sometimes with wildly different soundstages and dynamics. If I’m listening to St. Matthew’s Passion – Erbarme Dich – then I change the slider slightly when Nirvana comes on.
To connect this to the statistics stuff, the small adjustments you make with the volume slider has the contextual effect of making you believe the music sounds better. This is a well known effect. Think food served in a nice place makes you think the food tastes better. So they built in a way to relate to the device that really is Apple: the more you play with the it, the more connected you feel to it. That is so much of the essence of what Apple does. That is why my X is now competing for best device: they added in necessary elegance which connects to deeper enjoyment levels that reinforce as I become intuitively efficient with the elegant gestures. Again: Apple’s method of closing the gap between you and your device at work. If you were to blind test volume slider adjustments, you’d get garbage results. And if you were to play a track and pretend to adjust, you could test versions of placebo effect, meaning a real difference in effect or measurement that comes out of no difference in action. The power of context!’
Now the addition: it occurred to me that I can perhaps explain a bit more, given that people have great difficulty with the concept of randomness. At the bottom level of existence, there is a process by which completely random existence maps. The base level is a lattice I label CMs for the ‘C Mechanism structure’, where C stands for counting and choice, constraint and context. This lattice represents existence states across which processes count and choose within the constraint of context. Counting becomes SBE and choice becomes S-PS-2PS, while constraint and context become fCM. The graphical representation is fCMd.
Take a series of SBE lines in a 1-0 relationship so there is a common End if you lay out all the Starts around that End. Now tangle some. How does that happen? If you take an ideal square and pin it along a hypotenuse, all the other lines are longer than the hypotenuse. Think of these as slack and the hypotenuse is in ideal tension. By ideal tension, I mean a state of tension which maintains the hypotenuse across all dimensions. Example is you can’t pull it tighter and tighter because that compresses the square and all the longer lines start to get mashed together. As they mash together, the differences between their paths – which are connected point locations in CMs – become less clear and they can cross over each other. As the square compresses, SBE can run backwards because directionality is connected point locations in CMs and that can become unclear too (as life on this earth daily reminds). That makes a knot. This is the same process as shifting the origin of an ideal circle where the points are either at origin and circumference or at diameter ends.
As an aside, this is why circles have their mystery: they always appear both as an object with a diameter and as an object with an origin and thus a radius. This separates them from the sticks of SBE that connect the dots of CMs. Compared to the sticks, that is who we are: the patterns that appear across layers of stick-connected CMs lattice. I represent as best I can the directional pattern across CMs. This can be described as the side which believes in growth, specifically in nurturing growth, which means application of constraint in the short term to provide the greatest future growth. The other side approaches the issue differently: they reduce the pattern toward the level of CMs, meaning toward points connected by sticks. These two Endpoints form the Mudi of being, meaning the existence of being and the process of being. These align as stacks of x-yR along the zK – i.e., as stacks of planes conceived of as ideal squares or circles along an axis. The zK counts toward more CMs being filled with fCM, meaning with patterns overlaid on the sticks of SBE, and toward just sticks and points of CMs. Complex existence in one direction, and isolated existence in the other. The choice in the other direction is fCM entropic, meaning it reduces all the energy of connection in the pattern layers of fCM toward the lowest possible state visible.
I take this to a level above where the Mudi of being is counted. The directional choice of their side reduces Things to their essence until they disappear and become points at the end of sticks to be used in some other pattern. The directional choice of our side enlarges things so that thing exists in the next pattern. It becomes a choice of forever: do you join the group that wants to reduce your existence to potential, in which you are progressively erased as it shifts into the next pattern, or do you join the group that maximize your existence, in which you connect to pattern that is growing as it shifts to the next pattern. You are and will be a point in CMs, a multi-dimensional Thing, and you have a choice. That choice is always Between the steps of your existence as it iterates in CMs as a line of points. Do you pick the direction that grows or the one that shrinks? Throughout human history, people have joined the one that appears to grow: the movement, the idea, the belief without understanding that relative choice across multi-dimensional identities confuses directionality so what you think is ‘good’ direction is actually ‘bad’. The simplest way I can say this is: when you hurt people, you aren’t nurturing. When you harm, you aren’t nurturing. When you aren’t teaching, you aren’t nurturing. When you aren’t learning, you aren’t nurturing. When you aren’t opening your mind to the other perspective, you aren’t nurturing.
I walked across a field of evenly sized stones to get here. That’s a literal statement.
I got in my car and experienced intense vertigo. I decided that I had to go by the gym to make it home so I’d give working it out a shot. I did that: I realized during an extremely intense partially inverted stretch that I was intimately absorbed in observing all the things I was experiencing, and then I lifted my head and kept trying to count the new view but the depth of field was different so I moved my focus rapidly around without being able to specifically focus on things at the right ‘speed’ for the new field. Got really dizzy really fast. So I was able to phrase the vertigo as a visual field counting mis-match. Thus the Elizabeth Browning quote! I was able to control the vertigo!
jomi: playing with perception 