Ken's Blog

Today, walking around Greenwich Village and seeing all the happy people wearing costumes, I finally made the connection between Halloween and Election Day. The connection was always there — I just hadn’t seen it before.

Halloween is the last official thing that happens before everybody here goes to vote. Voting, unlike Halloween, is a very serious and grown-up thing.

In Brazil they have just made the collective choice not to vote into office a corrupt authoritarian who seems to want nothing more than to squeeze the joy out of human existence. I am happy for the Brazilians.

This coming Tuesday we Americans have a similar opportunity. We can vote to take power out of the hands of corrupt authoritarians who seem to want nothing more than to squeeze the joy out of human existence. If we manage to do that, I will be happy for us.

Halloween is great fun. But it is useful to remember that there are, after all, real monsters.

Speaking of prototypes, I have an idea for future XR glasses. By “future”, I mean maybe about 10 years from now, when Moore’s Law predicts that computation will be about 100 times faster than it is now.

So we won’t really need to worry as much about things like compute power or the bandwidth and latency of wireless connections. But we will still need to worry about the laws of physics and optics.

I have some ideas about how we can pack all of the optical components into a thin form factor that looks like a lightweight pair of eyeglasses. The only constraint in my approach is that every part of each eyeglass lens needs to be the same distance from the corresponding eye (in other words, it needs to form a section of a sphere).

So I just needed to decide how large to make the eyeglasses. Given that the human eyeball can rotate about 90 degrees horizontally and about 70 degrees vertically, the design pretty much suggested itself. You can see the result HERE.

Sometimes I think something is impossible. Then I build a prototype.

Usually the prototype ends up cheating in some way. It may not do everything I want, and it might cut corners. But having a working prototype helps me to better understand what I really want.

It’s not that the impossible has suddenly become possible. Rather, it often turns out that the “impossible” problem can be successfully tackled if you reframe it in some subtle way.

Essentially, there is often a meaningful connection between the unattainable original goal and a similar solution that is effectively the same, and yet is indeed attainable.

Without building a prototype, I would never discover that attainable solution.

If I flee to Miami and declare myself to be Brazil’s legitimate leader, is that enough for me to become Brazil’s legitimate leader? I am curious about how that works.

To be clear, I haven’t won an election in Brazil, so I couldn’t really use that as evidence to backup my claim.

But apparently, from what I hear, that sort of detail is not necessary these days. Although it seems that fleeing to Miami is a key part of the process.

Also, I hear the weather is nice in Miami this time of year. So that would be a plus.

Now let’s transpose the time, but not the place. In a future New York City, when the AI connection to our eyeglasses or our contact lenses goes out, we will understand that a lot of things will temporarily become unavailable.

We will temporarily not know how to cook a meal, do our morning exercises, find items in stores, pick out furniture, write an email, start a conversation at a party, or do many other things that we normally take for granted.

Judging by current human behavor, what we won’t do is blame ourselves for being dependent upon AI. We will take it for granted that AI is something that should simply always be available, and we will blame the internet for the inconvenience.

Does that sound about right to you (no fair asking Claude or Gemini or ChatGPT)?

In New York City, when the power goes out we understand that a lot of things will temporarily become unavailable. We won’t be able to take the subway, charge our phones and computers, buy a pizza, run our air conditioner, or do many other things that we normally take for granted.

What we don’t do is blame ourselves for being dependent upon electricity. We take it for granted that electricity is something that should simply always be available, and we blame the power company for the inconvenience.

In some ways this makes me think about the future of AI. More tomorrow.

To me one of the ultimate puzzles in life is the motivation for working on puzzles. Case in point: Every day I work on solving a word puzzle that I see in the newspaper.

I spend time working on that puzzle even though I receive no reward other than solving the puzzle. But it gets even stranger than that.

I’ve realized that I work on the daily puzzle precisely because it is a perfectly recreational activity. If someone were to offer to pay me to engage in exactly that same activity every day, I would probably not do it.

If you are confused about the difference between chatbots and actual humans, you can ask yourself a simple question: What is their quality of life?

We generally have a pretty good idea of the quality of life of a person, if we observe them carefully enough. We can even do a reasonable job of ascertaining the quality of life of a dog, or a horse, or a mouse.

But what does it mean for a chatbot, whether Gemini or Claude or ChatGPT or one of their rivals, to have quality of life? The question is actually meaningless.

In order to have any sort of quality of life, you need to have a life. There needs to be an actual self.

Humans instinctively understand the vast difference between “alive” and “not alive”. When it comes right down to it — in cases where it really matters — we are not going to be confused by sophisticated illusions.

A few weeks ago I posted a simulation of eyewear that uses a half-silvered mirror to let you simultaneously see what is in front of you and to one side. The problem with that approach is that you can’t really get more than 60 degrees of horizontal view angle.

So I developed a variant that uses a compact half-silvered magnifying mirror pointing into a lens. To verify that it all works properly, I built an on-line simulation to trace the rays of light through the system.

My simulation first shows just an eyeball, then as you click it adds a parabolic curve that will contain the magnifying mirror, then clips that curve to just the curved mirror, then adds a lens to complete the optical path, then replaces the lens with a fresnel lens, which saves weight.

You can see it for yourself HERE.

They say you should never meet your heroes. But this evening I had dinner with one of my heroes, and the entire experience was quite wonderful.

The people who say that might just have the wrong heroes. Maybe people should get better heroes.