bayeslives

I've been using both Windows (work) and Linux (home) on the desktop for many years. I far prefer Linux (I tried many distros, settled on Arch and Mint now). Package management and window management is way better in Linux. To name just a few.

To me, Windows10 is not better than Windows7. The package management is still a mess. Forced updates, no rollbacks, reboots, you name it. Permissions are also strange: why are admin rights needed to install some fonts? We still have this "Registry" with its voodoo. We still miss tools like dmesg. When thing go wrong, Windows gives some sort of simple log file, of course not in txt-form but in some proprietary format that needs it's own reader (to read some flat text, for crying out loud). The file manager ("Explorer") is still confusing. Microsoft seems hellbent in obfuscating where files are stored. Going to the terminal, the difference is even clearer. Linux offers first rate terminals; Windows look like an afterthought. Is it even possible to use a terminal full screen in Windows, nowadays? Tab completion, a decent history, pipes?

The funny thing is that Ranger, my favorite file manager, is able to give previews of files in MS formats (such as .docx) where Windows Explorer is unable to do so.... ????

Two big steps back are privacy and font rendering. It's bad enough that apps spy on me, I don't want my OS to"phone home". The font rendering is optimized for the few Windows on mobile devices users, at the expense of the bulk of the users (like me) who use it on the desktop and are now confronted with blurry fonts.

My OS must help me with my personal computer tasks. Personal means a) I want to customize the hell out of it, and b) my data is _mine_. Also, I'd like a nice readable font on my screen. Sadly, Windows10 does not meet those requirements.

Correct. I use Linux at home. Ubuntu and Mint have nice rendering OOTB, Arch got that great with the Infinality package.

OSX also has decent rendering (albeit too fat for my taste), Windows' subpixelrendering + matching fonts used to give sharp letters, yes too skinny, but legible.

With their new engine, texts are less legible. The letters are still skinny, but now they're also blurry.

All this to cater to the very small fraction that uses Windows on a mobile device; throwing existing desktop users under the bus.

Good points. The NHST thing was invented by Neyman & Pearson as a tool for decision making, not for finding the truth.95% confidence means your intervals will be not to far off, in 95% of all samples.

Perhaps this is nice fur Quality Asurance in factories. Where I do repeated measurements and where I want a simple YES or NO.

But science usually asks: "what can I learn from this specific data? I don't do 100 samples and I'm not interested in being "not to far off moist of the time". I want a best estimation based on this specific sample".

NHST does not give that answer. Bayes does.

Bayes did not "solve induction". If we define induction as telling which specific model generated this data: that is not possible. Countless models could in theory generate our data. What we need is some restriction. Like a prior. And when was the last time you started a research project without any idea what to expect? And if you did, wouldn't it be wiser to do some literature study, experts interviews etc. before starting experiments? Modeling the state of the art, pre-experiment, seems like a clever move anyway.

To name just a few of the NHST/p-value flaws:

1- I'm interested in P(H1 | data) but I get P(data | Ho). Contrary to popular belief P(data|Ho) != P(Ho|data). Let alone that conclusions about P(H1|data) can be drawn.

2- it is vulnerable to wrong interpretations. * No, a 95% confidence interval (a,b) does NOT mean there is a 95% chance that Mu is in (a,b). * No, p=0.04 does not mean they is a 96% chance that H1 is true.

3- the p-value depends on the intentions of the scientist. If you end your experiment after 80 observations, as planned, your p-value is different from that of an experiment that ended unplanned after 80 observations. So the same data have different evidential power, influenced by results you did not see in experiments you did not do. This is very unsatisfactory.

4- the idea of "an effect that exists or does not exist", based on some arbitrary threshold. The reality is, in many cases, uncertainty and variation. In group A I see effects of medicine A, with lots of variation between persons. In group B I see varying effects of medicine B. Then I introduce uncertainty by drawing random samples from A and B. Let's day I used those samples to make an inference: is, on average, medicine A better than medicine B ? Matras like "there is an effect, or there isn't" are not very helpful. Statistics should be about quantifying uncertainty rather than give false yes/no statements.

5. Basing decisions and knowledge on the data only makes t vulnerable to outliers, unlucky samples and so on. And why should you NOT use information, when it's there ?