Queen’s Gambit and more

Kenneth Regan is my partner here at GLL and a dear friend.

He is a longtime faculty member in computer science at the University of Buffalo, a longtime fan of the Buffalo NFL football team, and a longtime international master of chess. He is also one of the world experts on detecting cheating at chess. Players cheat by making their moves based on computer programs instead of relying on their brains. The key issue is: computer programs play chess better than any human.

Today I thought we might send Ken a tip-of-the-hat for being in the Wall Street Journal.

That is he is featured in a WSJ article called Queen’s Gambit about chess cheating.

This is not his first mention in the WSJ—see here. See this for additional comments by Cory Nealon, who says:

The game of chess is no stranger to cheating. But since the pandemic hit and tournaments have moved online, the services of chess detectives have never been more in demand.

Ken is perhaps the world’s most feared chess detective.

## Beyond Chess

Chess is important, but it is a game. Yet chess cheating may be an early warning of things to come. Might robotic methods in other areas soon start to dominate humans? Might chess be one of first areas to be lost to robots?

A 1970 movie that comes to mind is The Forbin Project. The plot is what happens when a computer system, Colossus, takes over the world. Its creator is Dr. Forbin, has the following exchange with it at the end of the movie:

Colossus later tells Forbin that the world, now freed from war, will create a new human millennium that will raise mankind to new heights, but only under its absolute rule. Colossus informs Forbin that “freedom is an illusion” and that “in time you will come to regard me not only with respect and awe, but with love”. Forbin responds, “Never!”

I think there is a reason to worry about computers like Colossus wiping us out. Perhaps not taking absolute control, but nevertheless making us second class citizens. I wonder if the following areas could be potential ones where we lose out to computers:

${\bullet }$ Protein folding? They already are doing quite well.

${\bullet }$ Play calling in the NFL? I made this one up, but I do wonder if using all the statistics available programs could do a better job than players and coaches. They probably could already do better than the Jets defensive coach, who was recently fired.

${\bullet }$ Solving Open Math Problems? We recently talked about Lean the proof checking system. But what if Lean could start to conjecture, to suggest directions, and give a proof outline? This would be, I believe, much more disruptive than the ability to check proofs.

## Open Problems

What are some areas that you think could be next? That could be taken over by robots?
Moreover, as more areas are dominated by robots, we will need more Ken’s to stop cheaters.

Thanks again Ken.

December 12, 2020 12:56 pm

Before they take over in areas like research math we’ll first see them decimate more routine white collar jobs. Things like mundane engineering design (the hard stuff like bridges will take much longer but routine stuff like engineering medium sized apartment buildings or consumer devices), secretarial services, research aides etc..

The danger, imo, isn’t that there won’t be jobs for people. The mere fact that human labor is scarce will make it statusful to purchase things made by people even if they are objectively less good (consider all the people buying handmade Swiss watches that are easily outperformed with cheap digital circuitry).

My worry is that we end up with a world where everyone is running to stay in place even though it’s in some sense not necessary. I mean think about fancy mechanical watches with complex features like phases of the moon. No one reallt cares directly about that, what we care about is sending the signal that we bought something fancy so we can impress others. If, instead, Swiss watchmakers had been half as productive no one would actually be less happy it’s just that the 10k watch would have a few less largely pointless features but it would be just as exclusive.

This is why UBI is so critical. Not because robots will take our jobs but because there is a risk they won’t and we’ll all continue to work just as hard producing positional goods. I don’t want that future. I want one in which work is a choice and people mostly do what they love (and the few who choose not to do so for outsized wages).

December 13, 2020 4:18 am

If computers take away our jobs, we’ll all have a 4 hour workweek and we can spend most of our time reading and writing blog posts (on a computer bought with universal income, of course).

3. December 13, 2020 1:28 pm

I like the telescope analogy: computers will help us see farther and deeper into mathematical structures, but we will choose what regions of the sky to point them at. Computers already act as assistants to work out tedious searches for counterexamples in finite cases, run through examples, keep track of basic logic to make sure we’re not making mistakes, simplify algebraic expressions, look for obvious patterns in numerical results, and so on (I use Mathematica and occasionally my own code for many of these things in practice).

Having this kind of assistant changes one’s style of research — for instance, some people give up rather quickly if a proof seems to involve a lengthy calculation, while others don’t think enough in advance about what quantity to calculate. Making calculations less costly makes it easier to explore this axis and set the right tone.

As far as computer-generated proofs go, proofs are valuable to the extent that we understand _why_ they are true, and gain a deeper sense of the objects and relationships they are about. One could try to get computers to look for understandable, “insightful” proofs, or provide explanations for their steps: this would be a little like asking AlphaGo to explain its moves. Is anyone working on this?

I’m also curious about how humans can glean insight from superhuman Chess and Go players. I’ve heard of particular take-home messages like “you don’t need to hug the sides of the board as closely as you thought” (Go) or “this endgame is better/worse than you thought it was” (Chess).

4. December 13, 2020 3:34 pm

“nevertheless making us second class citizens”

‘Due loss of social status’

Its the emotional core that made wild men wild (loss of status, often the result of defeat in battle) and ‘man like apes’ a considerable cause of anxiety.

“But what if Lean could start to conjecture, to suggest directions, and give a proof outline?”

Reminded of an old Joke between David Hume and Rousseau, Rousseau cites a passage from a French novel in which a bishop confronts an orangutan (chimp in modern taxa) and grandly utters “speak and I will baptist you”

Hume responds, ‘perish the thought, its no wonder these creatures prefer to stay mute!’ (myth that the monkey and ape, refused to speak to avoid hard work).

5. December 16, 2020 7:48 am

Re: to conjecture, to suggest directions, and give a proof outline?

Here’s a bit of material relevant to abductive reasoning and hypothesis formation …

6. December 18, 2020 8:16 am

Dear Dick & Ken,

Here’s a just-updated Survey of blog and wiki posts on three elementary forms of inference, as recognized by a logical tradition extending from Aristotle through Charles S. Peirce.  Particular attention is paid to the way these inferential rudiments combine to form the more complex patterns of analogy and inquiry.