A Little Ludwig Goes a Long Way

A smattering of opinions on technology, books, business, and culture. Now in its 4th technology iteration.

Edge Software, Developers, Consumer goods, Auto Chips, AI Imagery -- Things I Learned This Week

06 March 2021

OK this is maybe 2 weeks of things, I am a little behind.

Edge hardware and software

I installed EEROs this week after a horrible install experience with Netgear ORBIs. The EEROs just work. EEROs had a great experience before acquisition by Amazon and it still seems to be great — and it is all about the software. Amazon is a software-first org and it shows, Netgear is not.

Another very intriguing network product — https://blues.io. A super simple iot network solution, from Ray Ozzie. It solves a real problem and has a great simple developer experience — I love this. I ordered my first device and received yesterday — looking forward to tinkering with.

Developer Experience

The Blues Wireless team clearly has focused on the developer experience — their documentation is great. I bet they have a great developer experience internally as well — which is critical to unlocking innovation as Nick Tune details. I wish I could have been as articulate about this in my past jobs as Nick is. A great developer experience makes everyone more productive, helps with retention, and unlocks innovation. And really it is not just about developers, everyone in the company should be treated the same way — we want everyone to be more productive and more innovative, and that means making their jobs easier and more satisfying.

Refactoring of restaurants, other consumer goods

This is an interesting article about ghost franchises — the restaurant delivery industry continues to evolve and mutate, driven in part by COVID.

This creates a consumer problem — an explosion of choices, ratings that are frequently gamed, no retail sites to sample at, and marketplaces that are complicit in the creation and gaming of the system. How do you reliably choose from the 14 places that say they will deliver the best burger to you?

This is not limited to restaurants. I’ve been looking for a toaster, here are some of the brands that Amazon and the web pitch to me as having a great compact toaster — IKICH, Dear Morning, BonsenKitchen, iSiLER, Cuisinart, iFedio, Hamilton Beach, BLACK+DECKER, Amazon Basics, Barsetto, Oster, LOFTer, Twinzee, Dash, KEEMO, SACVON, Keenstone, whall, KitchenBro, HadinEEon, KitchenAid, REDMOND, Elite Gourmet, Breville, Zwilling, Smeg, Krups, Revolution Coooking, Schloß. All of these have great ratings, many claim to be award winning (tho I suspect some of the rewards were created just for this brand), all have professional looking brand landing pages. I can’t really trust ratings, Amazon placement, Google placement.

There is a missing layer of authentic customer recommendations from trusted sources. Which is a difficult layer to create and maintain — many many startups have tried, and most end up getting sucked into the morass. I have no answer, but I have renewed my membership to Consumer Reports which is a great resource, but very limited in its category coverage.

Auto chip shortage

Automotive grade chips different than the high volume mobile phone and consumer electronics chips. Automotive use cases require automotive grade electronics, which can operate over a much greater temperature range than consumer electronics. My limited understanding of the technical issues: all chips suffer from thermal degradation, and I have read it is due largely to metal migration at the connectors. Automotive grade chips have some special designs at connection points to accommodate this. Another difference is that automotive electronics are expected to last a long time in the field — 10 years or so. No one has this expectation of phones.

These are reasonable requirements for the automobiles, but they keep the auto industry off in a sequestered submarket for chips, which are generally several generations behind consumer electronics chips, and are subject to unique supply constraints. This situation is not going to get better, as the shift to EVs/autonomy drives more chip content into the car. Chip dependence is going to grow for automakers, but their relative volumes and buying power versus the mobile industry or other consumer electronics players is much less and not going to improve.

What is an automaker to do? They are going to have to manage their chip commitments more directly. They may have to step in and more directly invest in chip or fab capacity and chip supply chain management. They may have to overcommit early to chip volumes. These seem to be the strategies that have worked so far for Hyundai and Toyota. These all mean money and time. And managing the chip supply chain is not really a strength of automakers — they are great at supply management and manufacturing in their historical domains, but they are sub scale in the chip industry.

I wonder if there is an alternative. I will hand wave about design issues here, and with the caveat that the last time I designed and populated a board was 1983 while in grad school. So take it for what it is worth.

Cloud Services have learned to deal with unreliable electronics — not by fighting failure, but by engineering for it. Data centers tolerate constant failure of underlying electronics, using replication and hot swap designs. While there is cost to have replicated hardware, it is dramatically cheaper than trying to harden individual hardware instances.

Could automotive ECUs be redesigned to support low cost replacement? Could some degree of replication be built in, allowing ECUs to be built on much cheaper (and more available) consumer electronics grade chips? Could ECUs be designed to have easy replaceability and a much shorter expected life, permitting low cost refreshes on a regular basis? Do ECUs even need “auto grade” chips — there are many stories about some new automakers simply using non-auto grade chips with some success. Can EVs use some of their battery power to maintain temps in the car to allow use of non-auto-grade chips?

There are certainly some challenging design issues here. But consider the cost and problems to secure chip capacity for future cars — it is worth exploring the use of commodity chips, and then innovating in the packaging and environmental support for them.

Random

The market for Icelandic horses is booming.

We may all be wearing wood some day — greener than cotton.

Vlad introduced me to beautiful AI generated dynamic patterns and beautiful AI generated imagery. I could look at these all day long.