Well put. This notion of "ignorabimus"[1] which he opposes would seem to have gathered a great deal of evidence in the meantime. What I wish was more widely understood today was the cost of attempting to use models that are effective in the natural sciences in domains where they seem to be consistently ineffective. The replication crisis [2] , where in some cases less than 40% of key studies in psychology and the social sciences could be replicated, is often chalked up to improper statistical analysis or lack of integrity on the part of the authors. But it seems more likely to be the result of relentlessly applying linear mathematics and statistical methods to a domain where things simply do not work the way they works in physics or the natural sciences. To be sure - there is some baby in that bathwater, but the social "sciences" have thus far failed to articulate the domains that are baby and the domains that are bathwater. The amount of human effort and time that is wasted because "We must know" and "We will know" - trying to fit a square peg in a round hole - seems immense.
I think your argument in quotes is a good one. :) Following a set of pre-defined instructions is not a creative act. It's not bad to build a puzzle, but I would argue that it's not nearly as meaningful of an experience as painting - or any other creative activity for that matter.
I worked on Scratch for 6 years, in charge of the online community. We often encountered adult programmers who were surprised by what was possible in terms of complex projects with Scratch. There was usually a great deal of concern about how the transition to "real" (text based) programming languages would go. It always seemed strange to me, this attachment to text based programming as the only "true" "real" etc. programming. At any rate, I saw quite a few kids make that transition without any problems. I've still never seen any systematic evidence that moving from tiles to text presents significant difficulty, and yet so much energy is devoted to "solving" this problem.
Here's an excellent podcast that goes in depth on the potential of enhanced geothermal of the same sort described above. Essentially they drill two holes and then frack the rock between them. This makes it possible to setup geothermal in many more places. https://api.substack.com/feed/podcast/74977669/bb712d64d007e...
Third enthusiastic Bangle JS supporter. The hardware is fine, but the ecosystem that comes with it is the best part. The excellent documentation makes it easy to make your own watchface or application.
When cooking, the rate of heat needed to maintain the same temperature is exactly the rate at which the food is cooling. So heating up a pot and insulating it is roughly equivalent to heating up a pot and continually heating it to maintain the same temperature. If we changed the culture of cooking to recognize this, we could save a lot of energy.
Shouldn't we make a plugin for SearX that learns from the results you click, so that the customization and machine learning is on the client side? That way search becomes a commodity, but the final selection algorithm's behavior is owned by the user.
Why don't we make browser plugins that automagically click through the popups and reject as many cookies as possible? It seems like this could be crowdsourced such that when the plugin doesn't already know the site, it asks the user to manually go through the process of rejecting as many cookies as possible, and then submits those actions to a database such that future visitors don't need to go through all the clicks.
Just got my Bangle.js 2 and like it very much. Also purchased a Watchy for the epaper display a few months back. Compared to Watchy, the espruino IDE and ecosystem for Bangle.js 2 is much more developed and the community much more active. There is support for widgets, custom settings, and apps other than watch faces, and the development / sharing path is much more thought through. Check out the app loader, and the 300 or so apps (all source available on github) here: https://banglejs.com/apps/
Privacy is important, but I also care about the terrible quality of search results I get from nearly all the major providers these days. Couldn't an aggregator like SearX host a machine learning layer that learns what results are more likely to be valuable to me, and ranks them higher in the results? Keeping the customization layer on my own server and improving search results would seem to be a big advantage both privacy and performance wise.
I purchased a Tuxedo laptop last summmer in order to support Linux only hardware vendors, but sadly it's been one of the worst computers - with the worst customer service - I've ever used. Sending it back for repairs doesn't help - they ignore the issues and fail to resolve them. At one point I didn't get a response for weeks. I'm not sure how their trustpilot score recently improved from fair / bad to great, but I'm suspicious that they could make such a turnaround so quickly. Read the bad reviews and they're quite consistent. https://au.trustpilot.com/review/tuxedocomputers.com
