Category: Uncategorized

My website host of over fifteen years went out of business, leaving my account ‘suspended’ without notice. While they said they sent two notices, I have spent over a decade complaining that I do not receive notifications from them. Essentially, all my work from 2002 was taken down. Anyway. Their support team brought my hosting up for for 48 hours to move to a new host. 

Except that during this period I was on a holiday with only my iPad. It held up splendidly for all sorts of maintenance actions, proving it’s a computer in every sense of the term. Migrating to a new host involved a terminal, FTP, downloading and uploading files, zipping and unzipping them locally and remotely. Finally, a web-based chat session to fix some final issues. The iPad handled all of this without any issues. In many ways, especially the endless logging in and out using FaceID and Keychain, and with battery life, it was faster than a Mac.

PS the only drawback is the complicated multi-tasking behaviour. There needs to be a faster way to view and switch between arbitrary open app windows, and neither Cmd-Tab nor the slow-drag-up gesture to reveal open windows is up to the task.

The wheel was developed awfully recently in human civilisation, given how critical it has proven to be since. Modern homo sapiens is about 130,000 years old, yet humans only invented the wheel in the Bronze Age, 5,500 years ago. That means 96% of human existence has been without the wheel.

It seems the wheel was so late in being invented because it was only useful in the context of a wheel and axle system, which while conceptually simple, is hard to get right:

To make a fixed axle with revolving wheels, Anthony explained, the ends of the axle had to be nearly perfectly smooth and round, as did the holes in the center of the wheels; otherwise, there would be too much friction for the wheels to turn. Furthermore, the axles had to fit snugly inside the wheels’ holes, but not too snugly — they had to be free to rotate. 

And

The success of the whole structure was extremely sensitive to the size of the axle. While a narrow one would reduce the amount of friction, it would also be too weak to support a load. Meanwhile, a thick axle would hugely increase the amount of friction. 

Interesting opinion piece by the VC firm Andreessen-Horowitz about how AI-heavy businesses resemble both SaaS and more traditional services businesses, and how founders of such companies should plan for potentially lower margins, more custom work for customers, and poorer defensibility.

In a nutshell, the article argues that

– Running AI software requires a lot more compute power and human oversight than traditional software, reducing margins

– Customer problems are diverse and data sets are very different even for similar businesses, so there are a lot of ‘edge cases’, sometimes as much as 40-50% of total cases

– AI models, however expensive they are to run, are becoming commoditized, bringing down barriers. There are also fewer – or no – network effects.

Regardless of how challenging it is, AI is making it possible to solve many new types of problems including those involving analysing and generating media – images, audio across all sorts of domains including health and medicine. I’m looking forward to this over the next decade.

This article on companies that are ‘feature factories’. I experienced a sense of discomfort reading the ‘signs’ the writer lists, having been guilty of some of them myself when I ran product management earlier, and then at other companies since. Examples:

No measurement. Teams do not measure the impact of their work. Or, if measurement happens, it is done in isolation by the product management team and selectively shared. You have no idea if your work worked…

No removed features. Primary measure of success is delivered features, not delivered outcomes. Work is rarely discarded in light of data and learning.

Prioritization rigor is designed exclusively to temper internal agendas so that people “feel confident”.

Roadmaps show a list of features, not areas of focus and/or outcomes…

Once work is shipped, team has little contact with support, customer success, and sales.

100% worth a read if you’re a business owner or a founder.

This NYT opinion piece contends that Michael Bloomberg’s US presidential campaign is taking a leaf out of Donald Trump’s playbook:

Take Mr. Bloomberg’s brazen spending, which has prompted claims that he’s an oligarch trying to bypass democracy by buying the presidency. Plenty of candidates would get defensive at such speculation. Mr. Bloomberg is unfazed. Who cares?! At least he’s in the conversation. More than that, the conversation is now centered around the idea that he could very well win.

The whole thing sounds Trumpian because it is. The Trump campaign was unabashed in 2016 and beyond about its plan to “flood the zone” with garbage or ragebait. The strategy worked in part because it engaged and energized his base. And, as Sean Illing detailed recently at Vox, it exploited a media ecosystem that is built to give attention to lies (in order to debunk them) and outlandishness (because it’s entertaining or newsworthy).

This reminded me of reading Matt Taibbi’s excellent book Insane Clown President. It’s essentially a mea culpa, with columns he wrote from the 2016 campaign trail followed by a commentary in hindsight, written soon after the elections. It demonstrates in real time how the candidate broke norm after norm, each supposed transgression further disqualifying him in the press as a serious candidate but in reality cementing his appeal with the population that voted for him. That process, it seems that Bloomberg recognises, has created new norms with fewer constraints, rendering propriety superfluous, an outright liability.

Report this week on how India’s infrastructure projects are increasingly being financed by foreign capital as Indian banks are shying away from project finance:

This is a sign that the domestic funding pipe, which broke after the debacle at Infrastructure Leasing and Financial Services Ltd, is yet to be completely fixed. Surely, when the country’s largest lender, State Bank of India, reported no growth in corporate loan offtake for the December quarter, it points to one thing. Lenders are still preoccupied with fixing legacy bad loans. Private sector lenders that have fared well, too have kept away from big sanctions.

Financing long-term capital expenditure in India has always been a problem. This RBI working paper from 2019 [PDF] on the under-developed state of the Indian corporate bond market describes how Development Finance Institutions in South East Asia were set up explicitly for financing projects that would begin paying off ten years or more after financing. However here:

In India, development banks were gradually converted into universal banks, based on the recommendations of the Report of the Working Group on the Development Financial Institutions (DFIs) (RBI 2004). 

Which means short-term deposits from individuals and companies are being used to finance long-term projects. This imbalance in liquidity creates several problems, not least that a default or delay by a project puts individuals’ bank deposits in jeopardy. The paper goes on:

In India, the proportion of firms using banks as the primary source of working capital is higher than most developing countries. Further, the proportion of loans requiring collaterals as well as the value of collateral (as proportion of loan) are among the highest in India. This indicates the prevalence of asset-backed lending in India, which is essentially a feature of a relatively less developed financial system with limited expertise to gauge the credit risk of unsecured lending.

As a result of this, the effective borrowing cost increases and therefore

The large corporates can raise debt from the overseas markets, the cost of which, even after adjusting for hedging cost, tends to be lower than the cost of borrowing through the domestic market-based sources.

Which brings us full circle. An underdeveloped corporate bond market leads to an over-reliance on banks and short-term savings account deposits to fund much longer-term projects, therefore requiring collateral, raising the costs of borrowing, making it easier to borrow from overseas.

Very interesting results from a study on macaques:

Essentially, they zapped different areas of the brain and observed how the monkeys responded. When the central lateral thalamus was stimulated, the monkeys woke up and their brain function resumed — even though they were still under anesthesia.

“The animal went from being deeply anesthetized to opening his eyes, looking around the room, and even reaching out for objects within only a few seconds of the stimulation turning on,” Redinbaugh describes. “Shortly after the stimulation ended, he went back into unconsciousness like nothing happened.”

These stunning effects suggest that there is a reciprocal relationship between two areas of the brain — the central lateral thalamus and deep cortical layers — and this relationship operates like an “engine of consciousness,”

William Gibson: 

For the moment, we live in both the physical world and in cyberspace, but the barriers between the two are blurring so as to become almost indistinguishable. “The online/offline distinction is going to be fully generational soon. Only old people will think of being on or off,” he says.

Soon, “being present in the moment” will mean having a choice of moments that we can be present in. One offline & many online. Although that is the distinction we make today. Tomorrow they will be equivalent.

This is ostensibly about AirPods, but it’s really about the possibilities of software:

… a simple tap of an AirPod or a quick voice command can take us to a different location via sound. Utilizing HomePods as sound receivers, an AirPods wearer would be able to “move” from the kitchen to family room. A quick tap of one AirPods, or Siri voice command could bring the wearer from the family room to kitchen to answer a family member’s question or simply to be “in” the room.

Admittedly, Apple is the only entity that’s even close to being able to deliver this sort of tech. It’s a combination of consistent hardware, fine-grained geo-fencing, identity & permissions and only then the actual audio-switching software.

This bit in the article took me back to an idea I had had years ago:

Another example involves utilizing AirPods to deliver different sound experiences to different people despite being in the same location and looking at the same thing. As an example, a single presentation shown in a school or office setting can end up delivering a dozen different experiences to those in attendance.

As an engineering undergrad student, I wondered simplistically if frequency division multiplexing could mean that different groups in a single movie theatre could enjoy different movies. The video and audio feeds would be FDMed; each group’s glasses and earphones would extract different video and audio signals. You’d get the experience of enjoying a movie together in a theatre even if you were split between what movie to watch.

The traditional path a startup takes is building a product (MVP or not), then outreach to find the right audience. This writer discusses a counterintuitive process of building an audience first, and then building the product (“audience-first products”):

Having an audience will give you a unique perspective on the world. You’ll receive instant, high-quality feedback on the ideas you share. People are taught to hunt down interesting people, ideas, and opportunities. But masters of the Internet attract them. Sharing ideas attracts like-minded people, who double as a feedback loop to make you smarter and more interesting.

It also has this advantage:

If you have a built-in audience, you don’t need to raise money. If you write well about a common and specific problem, Google will send you lots of free traffic.

The writer makes it clear that this is more appropriate for building bootstrapped companies with focused audiences, than those that increasingly use capital as a moat. Of course, once you have a product and an audience that finds in it real value, you can always scale.