Links:

• Introducing Nebula, the open source global overlay network from Slack
• nebula: A scalable overlay networking tool with a focus on performance, simplicity and security
• Nebula VPN routes between hosts privately, flexibly, and efficiently
• How to set up your own Nebula mesh VPN, step by step
• LINUX Unplugged 329: Flat Network Truthers
• Cloudy with a chance of neurons: The tools that make neural networks work
• Welcome To Colaboratory
• ImageColorizer Notebook
• DeOldify: A Deep Learning based project for colorizing and restoring old images (and video!)

Plus an update on ZoL SIMD acceleration, your feedback, and an interesting new neuromorphic system from Intel.

Links:

• ZFS On Linux Has Figured Out A Way To Restore SIMD Support On Linux 5.0+ — Those running ZFS On Linux (ZoL) on post-5.0 (and pre-5.0 supported LTS releases) have seen big performance hits to the ZFS encryption performance in particular. That came due to upstream breaking an interface used by ZFS On Linux and admittedly not caring about ZoL due to it being an out-of-tree user. But now several kernel releases later, a workaround has been devised.
• ZFS On Linux Runs Into A Snag With Linux 5.0
• NixOS Takes Action After 1.2GB/s ZFS Encryption Speed Drops To 200MB/s With Linux 5.0+ — A NixOS developer reports that the functions no longer exported by Linux 5.0+ and previously used by ZoL for AVX/AES-NI support end up dropping the ZFS data-set encryption performance to 200MB/s where as pre-5.0 kernels ran around 1.2GB/s
• Linux 5.0 compat: SIMD compatibility · zfsonlinux/zfs@e5db313 — Restore the SIMD optimization for 4.19.38 LTS, 4.14.120 LTS, and 5.0 and newer kernels. This is accomplished by leveraging the fact that by definition dedicated kernel threads never need to concern themselves with saving and restoring the user FPU state. Therefore, they may use the FPU as long as we can guarantee user tasks always restore their FPU state before context switching back to user space.
• no SIMD acceleration · Issue #8793 · zfsonlinux/zfs — 4.14.x, 4.19.x, 5.x all have no SIMD acceleration, it is like a turtle. very slow.
• Chris's Wiki :: ZFS on Linux still has annoying issues with ARC size — One of the frustrating things about operating ZFS on Linux is that the ARC size is critical but ZFS's auto-tuning of it is opaque and apparently prone to malfunctions, where your ARC will mysteriously shrink drastically and then stick there.
• Software woven into wire, Core rope and the Apollo Guidance Computer — One of the first computers to use integrated circuits, the Apollo Guidance Computer was lightweight enough and small enough to fly in space. An unusual feature that contributed to its small size was core rope memory, a technique of physically weaving software into high-density storage.
• Virtual Apollo Guidance Computer (AGC) software — Since you are looking at this README file, you are in the "master" branch of the repository, which contains source-code transcriptions of the original Project Apollo software for the Apollo Guidance Computer (AGC) and Abort Guidance System (AGS), as well as our software for emulating the AGC, AGS, and some of their peripheral devices (such as the display-keyboard unit, or DSKY).
• The Underappreciated Power of the Apollo Computer - The Atlantic — Without the computers on board the Apollo spacecraft, there would have been no moon landing, no triumphant first step, no high-water mark for human space travel. A pilot could never have navigated the way to the moon, as if a spaceship were simply a more powerful airplane. The calculations required to make in-flight adjustments and the complexity of the thrust controls outstripped human capacities.
• Brains scale better than CPUs. So Intel is building brains | Ars Technica — Neuromorphic engineering—building machines that mimic the function of organic brains in hardware as well as software—is becoming more and more prominent. The field has progressed rapidly, from conceptual beginnings in the late 1980s to experimental field programmable neural arrays in 2006, early memristor-powered device proposals in 2012, IBM's TrueNorth NPU in 2014, and Intel's Loihi neuromorphic processor in 2017. Yesterday, Intel broke a little more new ground with the debut of a larger-scale neuromorphic system, Pohoiki Beach, which integrates 64 of its Loihi chips.
• Dancing Demon - YouTube — Written in 1979 by Leo Christopherson for the Radio Shack TRS-80 Model I computer. This is the best game ever for at that time.

Links:

• Microsoft’s neo-Nazi sexbot was a great lesson for makers of AI assistants — What started out as an entertaining social experiment—get regular people to talk to a chatbot so it could learn while they, hopefully, had fun—became a nightmare for Tay’s creators. Users soon figured out how to make Tay say awful things. Microsoft took the chatbot offline after less than a day.
• Microsoft's Zo chatbot is a politically correct version of her sister Tay—except she’s much, much worse — A few months after Tay’s disastrous debut, Microsoft quietly released Zo, a second English-language chatbot available on Messenger, Kik, Skype, Twitter, and Groupme.
• How to make a racist AI without really trying | ConceptNet blog — Some people expect that fighting algorithmic racism is going to come with some sort of trade-off. There’s no trade-off here. You can have data that’s better and less racist. You can have data that’s better because it’s less racist. There was never anything “accurate” about the overt racism that word2vec and GloVe learned.
• Microsoft warned investors that biased or flawed AI could hurt the company’s image — Notably, this addition comes after a research paper by MIT Media Lab graduate researcher Joy Buolamwini showed in February 2018 that Microsoft’s facial recognition algorithm’s was less accurate for women and people of color. In response, Microsoft updated its facial recognition models, and wrote a blog post about how it was addressing bias in its software.
• AI bias: It is the responsibility of humans to ensure fairness — Amazon recently pulled the plug on its experimental AI-powered recruitment engine when it was discovered that the machine learning technology behind it was exhibiting bias against female applicants.
• California Police Using AI Program That Tells Them Where to Patrol, Critics Say It May Just Reinforce Racial Bias — “The potential for bias to creep into the deployment of the tools is enormous. Simply put, the devil is in the data,” Vincent Southerland, executive director of the Center on Race, Inequality, and the Law at NYU School of Law, wrote for the American Civil Liberties Union last year.
• A.I. Could Worsen Health Disparities — A recent study found that some facial recognition programs incorrectly classify less than 1 percent of light-skinned men but more than one-third of dark-skinned women. What happens when we rely on such algorithms to diagnose melanoma on light versus dark skin?
• Responsible AI Practices — These questions are far from solved, and in fact are active areas of research and development. Google is committed to making progress in the responsible development of AI and to sharing knowledge, research, tools, datasets, and other resources with the larger community. Below we share some of our current work and recommended practices.
• The Ars Technica System Guide, Winter 2019: The one about the servers — The Winter 2019 Ars System Guide has returned to its roots: showing readers three real-world system builds we like at this precise moment in time. Instead of general performance desktops, this time around we're going to focus specifically on building some servers.
• Introduction to Python Development at Linux Academy — This course is designed to teach you how to program using Python. We'll cover the building blocks of the language, programming design fundamentals, how to use the standard library, third-party packages, and how to create Python projects. In the end, you should have a grasp of how to program.