Weekend Update

Okay, it's not the weekend, and this isn't Saturday Night Live. But it is an update. With only a handful of exceptions, I write one post per day of around a thousand words. That gives me a problem as to how to update the story when new information comes along. If it is suitable for a whole new post on its own, no problem. I just write the post.

A good example was the story of the Spectre and Meltdown vulnerabilities. When they became public, I wrote about them. When more information became available, I wrote about that. When there were informative discussions about what to do about fixing the problem, I wrote about that. See my posts:

• What is Meltdown? How Can It Affect Both Intel and Arm? (the original post the day the story became public)
• Spectre and Meltdown: An Update
• After Meltdown and Spectre
• Spectre/Meltdown and What It Means for Future Design (3 posts)
• Spectre with a Red Hat (2 posts)
• Paul Kocher: Differential Power Analysis and Spectre

That raises the challenge of what to do about updates that are important enough that I want to write about them but not worthy of an entire post. One of the aims of Breakfast Bytes is that if you read it every day, you won't miss anything important in the EDA and semiconductor ecosystems. So I've decided to put a lot of small updates into a post from time to time. This post today is the first.

Conferences

Many conferences are being postponed or going virtual. Since people don't want to plan for a non-virtual conference and then have to change at the last minute, some of these decisions are being made far in advance.

• CadenceLIVE: See my post Another Year of CadenceLIVE—with Updated Schedule for details. For now, at least, this is still accurate.
• SEMICON China has been postponed and will be in Shanghai on June 27 - 29.
• Embedded Vision Summit has been moved to September 10 - 20 and will be virtual.
• Arm TechCon will be virtual with the same dates, October 6 to 8.
• TSMC Technology Symposium has been pushed out to August 24, with the OIP Symposium the following day, August 25.
• DATE 2020 was postponed at the last moment and is being moved to virtual. DATE 2021 has been moved from Madrid to Grenoble and from March to February 1 - 5 (and I am the publicity chair so you will hear more about this from me during the year!)

John Horton Conway

John Horton Conway passed away on April 11.

I wrote about John Conway, who was one of my lecturers in undergraduate mathematics as well as being famous for his work on the classification of finite groups (there is a set of finite groups called Conway Groups even). He is most famous in the world of non-mathematicians for the Game of Life (a two-dimensional cellular automaton).

And there's always an XKCD for everything, even John's passing.

I wrote about Conway in:

• Short Papers
• Doomsday in 1900 Was a Wednesday

Wave Computing

I wrote about Wave Computing a couple of times, one of the best-funded and ambitious deep learning training chip companies. At one point they owned MIPS before selling it off again. They recently went out of business.

• Wave Computing: a Dataflow Processor for Deep Learning
• The Linley Halloween Processor Conference

Online Advertising

I covered various aspects of online advertising and tech regulation in a number of posts:

• Gordon Moore Killed the Oakland Tribune
• "GDPR Is an Enormous Regulatory Own Goal"
• Benedict Evans 2020: Regulating the Giants
• Razor Blades, Banking, and Antitrust

Here's the latest foolishness for foreign governments worried that the business model of their newspapers doesn't work anymore. I can't do better than crib Benedict Evans' summary from his weekly newsletter just last weekend. Before you read this, let me remind you that Google makes a lot of money advertising, but exactly zero of that comes from Google News since they don't run ads there. As I pointed out in the GDPR post above, if newspapers don't want Google to index their site, they can simply edit the robots.txt file that controls which parts of a site are indexed by Google's spiders (and other search engines). Note that "I" in this quote is Benedict Evans, not me:

A link tax in Australia: Australia has another proposal to tax Google and Facebook to subsidise newspapers, calling this a 'code of conduct' based on 'negotiation'. Given that the newspapers have nothing to offer in a negotiation and G&FB would have no choice, that sounds like a tax and a subsidy to me. These proposals have floated around all over the world for a decade or more, and tended to fail on general nervousness at the idea of making news media dependant on political patronage, as well as the question of why it should be internet companies in particular that pay (as opposed to general taxation, if you think this is a general public good).

And another link tax. Meanwhile, Google is in a slightly different fight in France. The French said that if you use snippets from other sites when you link to them then you have to pay, so Google removed the snippets from Google News. Now the competition authority says that since Google has market dominance this is anti-competitive, and wants to force Google to put the snippets back and to pay for them. I don't like anti-competitive behaviour, but I also don't think it's sensible to invent a business model that you wish existed and then force someone to follow it. Too many people in the news business still don't want to accept that Google does't actually make any money from Google News or news search results - no-one wants to buy ads on those pages. Again - if you want Google (or Facebook) to subsidise newspapers, better to be honest and make an actual tax, and accept all of the questions you need to resolve about how it works, rather than inventing fake revenue lines.

Zombie Satellites

Remember that guy in my post Zombies who discovered a satellite called IMAGE that NASA had left for dead was still operating. Well, he decided to try again with a US military satellite launched in 1967 and long ago abandoned. To try and access it, he had to build a new antenna, new electronics, and more. It was going slowly since he has a full-time business and a family. But then came lockdown, so he couldn't leave home. He finished the setup and succeeded in finding it in a geostationary graveyard orbit. It is the oldest functioning satellite still in geostationary orbit.

NPR tried to find out more, but it did not go well:

The MIT lab that built LES-5 still does a lot of work on classified projects for the military. NPR contacted its news office to ask if someone could say more about LES-5 and whether it really could still receive commands. But after repeated requests, Lincoln Laboratory finally answered with a "no comment."

NPR has the full story.

Starlink

Still keeping with the satellite theme. On 19th February, I wrote about Space X's Starlink satellite-based communication network in What If It's Not 5G, But Satellites? Since I wrote that there have been a couple more launches of 60 satellites, so there are over 400 satellites deplooyed. One interesting thing about the latest launch (which you can watch here) is that it was the 4th time the booster (stage 1) had been used, and the second time the fairing around the satellites had been used. The booster landed successfully on the barge, so it will presumably used for a fifth time at some point.

Statistical Testing

There is a lot of discussion online about tests by people who seem to have no idea what the words "false positive" and "false negative" mean, the critical parameters for any test.. But I covered that for you in 2016 in a post Happy Thanksgiving. Do You Have Toenailitis? As it points out in that post, it's not just pundits who are illiterate about Baye's Theorem, so are most doctors. Just like journalists generally don't have a lot of training in math, neither do physicians. I criticized journalists for their innumeracy in my post How Many Journalists per Square Acre?

I also wrote about statistical testing, p-values, and random controlled trials (RCTs) in Statistical Power...or Why You Shouldn't Be Allowed to Turn Right on Red. I recommend the paper in the BMJ (fka British Medical Journal) Parachute use to prevent death and major trauma related to gravitational challenge: systematic review of randomised controlled trials. Although this is an amusing spoof, there is a serious point to the piece, namely that randomized control trials are not the only way towards evidence-based medicine.

Neither of these is an update, they are just more topical perhaps than when I wrote them. Also, the toenailitis post went live on Thanksgiving Day 2016, so I doubt many people read it. I've since moved my holiday off-topic posts from the holiday itself to the day before, so that people might actually see them.

Sign up for Sunday Brunch, the weekly Breakfast Bytes email.