Web Excursions 2021-04-13
🌟 [Post of The Day] 15 Technologies I Thought My Son Would Never Use
Nine years ago this week, my son was born and, to mark the occasion, I wrote an article, predicting 15 current technologies he would never use, because they would be obsolete by the time he would have a reason to interact with them. Now, as Isaac turns 9, it’s a good time to take a look back at my predictions and assess their accuracy.
No More Wired Home Internet: TBD.
It looks like my son will remember using wired home Internet as a child, like I remember using rotary phones.
No Dedicated Cameras and Camcorders: Right
He Won’t Live with Landline Phones: Right
His Computers Will All Boot Super Fast: Wrong
He’ll Never Use a Windowed Operating System: Wrong
Though I hated Windows 8, I thought it was a harbinger of a world where every desktop OS, even macOS and Linux would be windowless.
Thankfully, users rejected the Windows 8 UI in droves and everyone in the industry realized that what works on a mobile device isn’t ideal for a PC.
He Won’t Use a Computer with a Mechanical Hard Drive: Right
He Won’t Go to the Movies: TBD, but likely to happen in a few more years. Ticket sales have been declining steadily since 2012, though in 2019, people still took 1.2 billion trips to the movies worldwide. The pandemic threw sales off a cliff in 2021, with just 223 million tickets sold, but even as life starts returning to normal, the box office will not.
A number of major studios dumped the 90-day theatrical window in 2020 and 2021 and consumers saw how much more convenient it is to just stay home.
He Won’t Use a Mouse: Laughably wrong
According to Baker, sales in 2020 were 13 percent higher than in 2013.
He Won’t Use 3D Glasses, Because 3D Will Be Glasses-Free: Wrong It seemed like home 3D entertainment was the next big thing. A number of promising glasses-free 3D technologies Factors: competing standards, expensive glasses, poor image quality and no evolution to 4K
He Won’t Use a Remote Control: Wrong
There Won’t Be Desktop PCs: Wrong For those of us who truly care about performance or want the pride that comes from building your own computer, there’s still nothing like a desktop. The number of people building their own desktops is on the rise. According to Baker, people bought 1.7 million empty PC cases in 2020, up from around 1 million in recent years.
He Won’t Dial Phone Numbers: Mostly Right
He Won’t Watch TV in Primetime: Right
For many of us, the real prime time is whenever Disney+ posts this week’s episode of Falcon and Winter Soldier or Amazon drops the next episode of the Expanse (in both cases, usually 3 am ET).
He’ll Never Use a Fax Machine: Right
Optical Discs Will Be Gone: Wrong There’s good reason to believe that optical discs and physical media in general will survive for quite a bit longer and that good reason is “when you own the disc, you actually own the content.”
Bottom Line
The “higher tech thing” isn’t always the best thing for usability, even if companies want to push consumers toward it.
It takes a really long time for well-established technologies to completely “die,”
Never underestimate the speed at which a child learns
一个匿名论坛的死亡 – 扯氮集
一个用户刚过万(精确的数字是11394个注册用户)的网络匿名论坛,生于2020年12月4日,逝于2021年4月12日。
按照创始团队的说法,这个项目本来自于他们的一个课堂大作业,然后就演变成一个iOS和安卓双系统均可运行的论坛型app。
交大学生做的,用户也主要是交大学生
注册邮箱必须使用@sjtu.edu.cn的邮箱,并验证这个邮箱。
按照团队在告别公告里的说法,他们“在技术上做不到查询用户真实身份”。
要求他们做到后台信息实名制注册
偏向于相信,他们只是用一个sjtu邮箱来核准你是否是真的交大人,但他们的确不记录id和邮箱之间的关系
用户可以在这个帖子里叫Alice,在那个帖子里叫Bob。
心理学上有一种对自我的分类:补充自我、补偿自我。
补充自我是指自我其实做不到的事,通过互联网展示,来佯装自我做到了。这常见于实名社区,也就是所谓的树人设
补偿自我是现实社会中无法展示而通过互联网展示,属于对心理的一种补偿。这常见于匿名社区
补充自我重社会资本的积累,补偿自我重心理宣泄。
作为一块宣泄地,这其实使得这类项目的运营难度之大,恐超他们的想象。
我并没有去过这个交大树洞的前辈清华树洞或北大树洞,所以我只能用豆瓣这个匿名社区来做个对比。
豆瓣和无可奉告的不同点在于:
豆瓣诞生于2005年3月,无可奉告则诞生于2020年。时代不同了。
豆瓣一开始并不是以一个匿名小组的主体形象出现的。它早期的形象(其实到今天都是)特别文青,无非就是书籍、电影、音乐的使用记录和点评。
摆脱了一个匿名社区容易踏入宣泄场的麻烦境地。你去了解了解书影音的信息并贡献点自己的看法,并不是一种宣泄。
2005年,当时的用户没有那么多元价值
这不能算一种值得骄傲的现象。但它带来一个结果,就是社区管理,远远没有那么麻烦。
权利意识处于觉醒中的状态时,人会变得很敏感。
无可奉告的用户量刚刚过万——但其实,交大本硕博一届学生就可以达到上万。
美国的文化地理学家泽林斯基提出过一个“第一有效定居理论”,用来诠释一个地区的主体气质是如何构成的:虽然最初定居者的后代只占现在人口的一半,但其影响力却远远超出一半。
线上社区有异曲同工之妙。 第一批用户,以及给予这第一批用户的环境,对于这个社区的气质养成——或者我们说成是这个社区的文化养成——是决定性的。
尤其在今天,一个社区的想要获得一种有效的自治文化,起手的管理引导的重要度,远超过后期的日常管理。
开发团队的理想主义气质。
他们在发布后三天进行了一次大更新,试图用产品功能去设定环境——这很好,但这很不够。
因为这是反人性的。
论黑话 – 扯氮集
张一鸣的一段内部讲话
黑话的来源,就是一种隐匿性的表达。它的出发点,就是让圈外人听不懂,以达到某种保密的目的。
黑话当下有了一种“大词”的特点。
黑话最盛的地方,是学术圈。
我并不觉得学术黑话的使用目的是为了保密,它的目的是这两项:
其一、表明表达者是这个圈子里的人:找到我们。
其二、表明表达者的实力:找到“我们”这个群体中的我的位置。
大词(黑话)的使用,能满足“我是优秀的我们之一”。
黑话不是一个、几个或者一堆大词那么简单。
黑话其实是一个系统——这里的系统,可能是一整篇文章,也可能是一个段落,甚至就是一句话也可以被视为一个系统。
作为一种系统,什么时候带有大词的,会让人感觉还好。什么时候带有大词的,会让人感觉是黑话表达呢?
系统论:还是原来那艘忒修斯之船,还是原来那面阿能诃鼓。
一个系统,由三部分组成:元素、连接、功能。
而元素这个部分,是最不重要的。
重要的是元素之间的连接,以及连接之后所能完成的功能。
如果连接和功能都相同,即便元素都不同了,系统依然是原来那个系统。
回到黑话这个系统来。
之所以会有一种强烈的黑话感,实质在于:连接的方式不对劲,而不是元素本身。
我个人并不太过于反对大词本身,但我不大看得上大词的连接方式:堆砌。
张一鸣举的那些例子。问题在大词吗?不是的,是大词的连接方式。
所谓云里雾里,是连接,不是元素。连接出了问题,功能也就出了问题:不知所云。
同样是 1080p,为什么「高码率」还得多花钱?
对于画质一般使用 PSNR(Peak signal-to-noise ratio)「峰值信噪比」 来表述,压缩后越接近 50dB 则越接近原视频画质越好,保持 35dB 以上压缩失真就不会很明显。
PSNR 是个很好的参考标准但不能完全体现视频真实的观看质量,Netflix 和南加州大学联合开发的 VMAF (Video Multimethod Assessment Fusion)视频质量指标表现要更好一点,
也是 Netflix 目前主要使用的参考标准。
随着码率的提升画质的提升是逐渐递减的
动画片有很多大的色块,运动也相对简单,所以对码率的需求其实很低,
如 Netflix 上动画类型的内容在 4k 规格下播放,码率也常常会在 1-2 Mbps 之间;
对于动作电影这些有大量运动镜头或者是画面有很多胶片颗粒、水汽、灰尘的内容 4k 规格下常常就得需要 16Mbps 甚至更高才能保证画质。
Netflix 目前会根据具体每一部片子的内容在维持高画质的情况下来压缩到合适的码率
针对每一个自然镜头进行优化
理论上我们的分辨率应该是针对不同码率不断的改变,但是现实不允许我们把同一个视频压缩为成千上万种的分辨,这对视频压制、存储、到适配都不现实。
最终我们是 「帕累托最优」 的选择,像是 480p、720p、1080p、4k 这些,一个阶梯图形码率和分辨率对应的结果。
很多影视剧集内容虽然多个平台都有,但往往只有持有版权的平台才有最高规格的视频。
比如 Disney+ 上漫威的电影基本都是 4K HDR,而国内的平台基本都是普通的 1080p。
Why We’re Freaking Out About Substack
Substack has captivated an anxious industry because it embodies larger forces and contradictions.
For one, the new media economy promises both to make some writers rich and to turn others into the content-creation equivalent of Uber drivers, even as journalists turn increasingly to labor unions to level out pay scales.
This new ability of individuals to make a living directly from their audiences isn’t just transforming journalism.
It’s also been the case for adult performers on OnlyFans, musicians on Patreon, B-list celebrities on Cameo.
In Hollywood, too, power has migrated toward talent, whether it’s marquee showrunners or actors.
This new direct-to-consumer media also means that battles over the boundaries of acceptable views and the ensuing arguments about “cancel culture” are no longer the kind of devastating career blows they once were.
(Only Twitter retains that power.)
Newsrooms can offer all sorts of support that solo writers don’t get.
Jessica Lessin, the founder and editor in chief of The Information, a newsletter-centric Silicon Valley subscription publication, said part of its edge was “sophisticated marketing around acquiring and retaining subscribers.”
Compared Substack with the unregulated, decentralized internet of a decade ago.
“In the golden age of blogging, writers hated each other but they went back and forth over each other’s ideas.
Now, people call the manager all the time,”
Ghost isn’t the only alternative, of course.
Twitter recently bought the newsletter platform Revue,
and Facebook is developing ambitious plans for a rival that will provide a platform for local journalists, among other writers.
The left-wing commentary site Discourse Blog moved to a rival platform called Lede.
Others, like the tech analyst Ben Thompson, cobble together email, blogging and payment services to be what he calls “sovereign writers.”
Substack and its backers are alert to the risk that the service could be replaced by someone charging a few dollars a month.
But they note that many writers simply don’t want to be bothered with anything other than writing, and happily pay the premium for that.
NVIDIA Unveils Grace: A High-Performance Arm Server CPU For Use In Big AI Systems
More broadly speaking, Grace is designed to fill the CPU-sized hole in NVIDIA’s AI server offerings.
The company’s GPUs are incredibly well-suited for certain classes of deep learning workloads, but not all workloads are purely GPU-bound, if only because a CPU is needed to keep the GPUs fed
NVIDIA is going their own way by building an Arm server CPU with the necessary NVLink functionality.
By 2023 NVIDIA will be up to NVLink 4, which will offer at least 900GB/sec of bandwidth between the CPU and GPU, and over 600GB/sec between Grace CPUs.
Critically, this is greater than the memory bandwidth of the CPU, which means that
NVIDIA’s GPUs will have a cache coherent link to the CPU that can access the system memory at full bandwidth, and
also allowing the entire system to have a single shared memory address space.
As for CPU performance, this is actually the part where NVIDIA has said the least.
The company also isn’t saying much about how the CPUs are configured or what optimizations are being added specifically for neural network processing.
NVIDIA’s Project Denver, which was originally announced just over a decade ago, never really panned out as NVIDIA expected.
Grace, in contrast, is a much safer project for NVIDIA;
they’re merely licensing Arm cores rather than building their own, and
those cores will be in use by numerous other parties, as well.
Both the Swiss National Supercomputing Centre (CSCS) and Los Alamos National Laboratory are announcing today that they’ll be ordering supercomputers based on Grace.
MiSTer-devel/Main_MiSTer
MiSTer is an open project that aims to recreate various classic computers, game consoles and arcade machines, using modern hardware. It allows software and game images to run as they would on original hardware, using peripherals such as mice, keyboards, joysticks and other game controllers.
MiSTer-devel/Main_MiSTer
The heart and engine of the whole platform is the Terasic DE10-Nano development board, made in Taiwan.
Terasic - All FPGA Boards - Cyclone V
The DE10-Nano Development Kit presents a robust hardware design platform built around the Intel System-on-Chip (SoC) FPGA, which combines the latest dual-core Cortex-A9 embedded cores with industry-leading programmable logic for ultimate design flexibility.
Users can now leverage the power of tremendous re-configurability paired with a high-performance, low-power processor system.
Altera’s SoC integrates an ARM-based hard processor system (HPS) consisting of processor, peripherals and memory interfaces tied seamlessly with the FPGA fabric using a high-bandwidth interconnect backbone.
The DE10-Nano development board is equipped with high-speed DDR3 memory, analog to digital capabilities, Ethernet networking, and much more that promise many exciting applications.
HN
Jorge1o1: I don’t really know much about game emulation so I was curious about what differentiates this FPGA game project vs traditional CPU emulation.
From their github page
Traditional emulators on CPUs execute code sequentially.
This is a tricky method of emulation because real hardware has many chips and all of them work in parallel...
This requires a lot of CPU power to emulate even an old and slow retro computer.
Sometimes even a modern CPU working at 100 times the speed of the retro computer is not enough,
so the emulator has to use approximation, skip emulation of some less important parts, or assume some standard work of the emulated system without extraordinary usage.
FPGA doesn't need high frequencies to emulate retro computers;
it works at much lower frequencies than traditional emulators require.
Since everything in FPGA works in parallel, it is no problem to handle any possible usage of the emulated system.
rtkwe: There was a good article from Arstechnica a decade ago that pointed out why you need so much more power to get perfect emulation.
To get exact emulation takes a lot of power because there are a few games which use odd tricks that are hard to document and precisely reimplement in software.
FPGA emulation gets around that by more directly emulating the hardware.
TillE: byuu wrote a good article about this, unfortunately it's no longer available, but basically it should be self-evident that there's nothing inherently more accurate about hardware emulation.