Indeed. I worked there for about 4.5 years, initially on the clinical programming software (the software that makes it as easy as possible for a clinician to setup various channel levels for the hearing part of the device and how they map to electrodes). It was a fair bit of fun doing it.
I ended up wanting a change and went and worked on the TIKI (Totally Implantable Cochlear (K?) Implant). During that time, Cochlear ran into some profit problems (circa 2000 ish?) and they did the redundancy thing for nearly everyone on the TIKI project. That was my first redundancy..
A TIKI device was eventually implanted in a volunteer subject. It was found to hear too much body/blood flow noise and was deemed unsuccessful. The device was made to operate as a normal implant as well, so they simply got an external speech processor and they were able to hear.
It would be interesting to hear how and how well Med-El managed to filter out the body noise from the internal device.
Often, engineering in Australia is hard to come by. We're a small nation in terms of population. If you've got a great idea and you have good people, you can still go far. Cochlear is one example. Resmed is the other Australian medical device success story. There are failure stories too like the Telectronics pacemaker whose wire insulation perished over time...
I'm really hoping QSAT (https://qsat.au/) will be another Australian success story. QSAT have a very neat satellite ground based comms system that can track and communicate with multiple satellites simultaneously. It's derived from the CSIRO phased array project. Unfortunately CSIRO looks like it's fading away. It is also harder for QSAT (probably most Oz startups?) to find startup capital in Oz. In the US you can pick up 0.5Mill of the side of the road....
I really do wish QSAT well. They have an innovative product and excellent people working on it.
The strength of the correlation might be the only useful single variable we can have in discussions like this.
I'm starting to think that "Correlation is not causation" is a kind of slippery slope toward a world where nobody can ever prove the cause of anything. At the same time, just about everything is at least a mild carcinogen including food, the sun, air particulates, etc. But knowing that everything causes cancer isn't useful information. Knowing the strength of the correlations of each of these things is actually useful.
Problem is, doing Y only "causes" cancer after doing it nonstop for like 40 years.
What can you do today that would cause cancer tomorrow or next week? Even next year? Nothing. Yet, just about every environmental thing causes cancer over long periods of time.
Which leads to the absurd: being alive causes cancer.
Establishing a correlation is only a starting point. After that there is an entire field of statistics called causal inference that is devoted to establishing causality using a variety of tools.
The strength of correlation is NOT the only useful metric and should not be mistaken for more rigorous statistical analysis regarding causality.
It indeed is very worth being very clear about the limited data that makes the images.
It is also very cool.
..Nothing to add, simply echoing your sentiment :-)
If I create a view of a table containing JSON data (or any other kind of data), I can create columns from calculations on existing fields.
I'm guessing the advantage of generated columns is that they are real columns, not computed columns like in a view. This means that if an insert doesn't work with the new generated column an error will be generated? Is that a correct understanding ?
Is this perhaps the main advantage of this feature ?
The data creation and storage options, always computed or stored on write of dependant columns seems like a possible advantage too.
Main practical difference is that you can put index on generated column, and on view you cannot, SQLite has no materialized views. Constraint part you understood right, NOT NULL on generated column fails at insert, and note that VIRTUAL costs nothing on disk but can still be indexed, so STORED is mostly for when expression itself is expensive.
He has some great videos. I watched this one and was impressed at the calculations, the transparency and the celebration at the end when he reached supersonic speeds.
reply