254 karma · joined October 17, 2013
* Discovery is back to being passive - recommendations come from the company that you are subscribed to.
* Quality is paid for separately - the subscription has a fee, your data has a separate fee.
* Quality is not guaranteed - it depends on your current internet connection.
* There are dead zones.
* Most apps I've used do a terrible job showing you only the downloaded music you have.
* Your available music is subject to the deals between record labels and subscription services.
* Subscription services aren't incentivized to house more obscure / old / rare / hard to find music.
* Music is siloed in to different subscription services. ( We are moving to a Showtime vs Starz vs HBO vs Netflix model )
What really gets me is that at the end of the day - going the legal route of a subscription service is truly only marginally better for artists.
The idea of Django throwing errors when you try and run async unsafe code within an event loop gives me cause for concern. I can just imagine folks who don't understand adding 'yield', 'await', 'yield from' until the errors go away - while being confused with the resulting execution. I think there is always a place for synchronous code in Python - I think Django should lean in to the simplicity afforded by it.
Last note - I also am very weary to mix asynchronous code with other primitives like threading, multiprocessing. If I am going the route of async programming - I usually just intend to optimize a single thread - and then I leave it up to the OS to optimize across multiple processes.
- Paper ballots and Risk limiting audits do not actually constitute a perfect solution to this problem. They do not constitute what is considered an End to End verifiable voting system. You cannot actually definitively know that your vote was counted in the final tally, and cast as intended. The Risk limiting audits as currently structured can only verify that a computer counted a small sample size of votes as a human would. It cannot confirm all the votes were counted, or that your vote was not tampered with.
- Paper ballots also do not address the needs of folks voting over seas, or those affected with disabilities.
- When current election systems are often described as easy to hack, the systems referenced are often hardware that is 10-20 years old. If you look at the landscape of the Election's industry, there are only 3 vendors that account for over 90% of the public election market share in America. These system's were put through certification many years ago, and continue to operate through those certifications. The result of an inability to innovate in this space, is that we will continue to use those systems (that many perceive as vulnerable) for the foreseeable future.
- Smaller election vendors are encouraging the idea of recurring certification, as well as more stringent certification requirements.
Most people who use these forums are very capable of understanding concepts like Asymmetric encryption, checksums, and other tamper proof methodologies for high value data. If you can trust the idea of an Encrypted e-mail server ( e.g. Lavabit ), or encrypted chat servers (Signal), it should not be hard to imagine that there is a secure way to conduct elections electronically and through the internet.
Finally, given how the past few decades have progressed, it seems that you would be on the wrong side of history to say that electronic voting cannot be done safely in a digital form. The process needs to be secure, but it also needs to be convenient, safe, fast, and trustworthy. Observing a few votes being cast in your polling station is nothing compared to observing every vote being cast for a given election.
It doesn’t use any asyncio parts of python but is just meant to show what’s happening under the hood.
To me it is really a question of, how do you vet software before you use it ? Will you take the time to understand your stack before it's deployed? When you employ certain abstractions, how much visibility do you lose ? Is it worth it ?
Docker is fine as one of many alternatives, but to turn your nose up at well written, battle tested software which is part of most Linux distro's is a little crazy. I would definitely bet that people who chose to learn how cgroups work, how systemd works will see their skills age more gracefully.
It reminds me of Dev-ops candidates that I have interviewed that laugh at the idea of ssh-ing to deploy a new version of code, without understanding that Ansible is generally doing the same.
I think the type checking community also has quite a ways to go when it comes to asynchronous code. It becomes quite worthless to check that a future is being returned by a function, but it could be more helpful to know what that future will eventually yield.
https://github.com/ltavag/blocks
Always love to see the interest here, I'm excited for the technology to become detached from Crypto speculation!