Hello Dr. Knuth;

I saw your talk at uwaterloo about family algebra an cant stop thinking about familes of families of sets and observed with rapt attention ans you took to the blavkboard and began scribbling in Greek.

But I undertood it. When you suggested uniwat was the ideal place to work on the polynoial - asymtptotic time that really grabbed my attention.

Allow me please to back up a little, I have to admit the ecitement of actually writing to you leaves me a little breathless.

Like any person in both sets of "very curious" and "likes math and computers more than eve aquariums" I read your books when I was quite young I'm 66 now) hacked around an IBM 1130 from 1970 on, took off in grade elevem to work on an 1130 at the Hamilton Sectator Newspaper, then gdXII, then I worked at a place called Teklogix in Mississausa where Dave Xonrow was writing gcc(nee DECUS C, nee RSX11M C, nee COnrog C) and began using it for work in 1975. Dave is one of those special people like Brian Reid, Jon Postel, Dennis, Pike & Ken who'm I hold very dear and know then pretty well, or knew them in Jon and dmr's case. Ok I have never talked to Ken but my first wife was Ken's siste' bridge partner - Mary Ann Coyle.

Unix opened up a new world for me. It was raational and extensable, so when I went to uniwat in 1977 and suffered the first (mandatory) welsome to terminal bsed bach job processing in Fortran (witfiv-s actually) and to his great credit, Eric Manning, upon being told I spent the last two years programming in c and want to play with unix, told me to di just that and og please btw write a 6502 emulator for my kid's SIM-1. Which I started and had I think a nice opcode decode and dispatch program in c althiugh my partner (mandatory) wrote one as a big if statement and got his running first handed it in and it was done. So, bored, I answered a ad for a BCPL programmer to convert troff to C. Rob BEach of the Typesetting Task Force there, signed me up as it was he I had to ask for a Unix account that semptember.

I figured I was set for life, here I am doing unix bits in C. REal world stuff, not the Business Basic/DiBOL AP, AR, payroll apps and had to do to ddord Universty. That should have been a clue right there.

Your adminition to not be distracted should have at that point being going off like Doctor Who's cloister bell. But I digress (even further).

Figuring if I could do this in Canada, I could probbly do iit somewhere else, like Bell Labs in NJ but I wanted to do graphics so  drove to LA in my brothers car with my friends girlfriend and my mum's credit card, more or less on a whim, I was supposed to start a new job 90 km way doing very boring Fortran encding of agorithems that used a multidimnsional something to do infrared pattern recognition. It sounded geechy anyway.

6 days later I figured I'd settled in well after half an hour there and found a job writing a lanfuage an dinterpreter at a company called Cado Systems on Torrance California. It's an interesting business case and made all the higher ups very rich. The langauge I designed, yes, another business basic, but with hard disks and softeare tht would emulte a lot of things. THen we made a 2 processor version then a 4 processer version then an 8 processor version. I left just before we released a 64 user 8086 version (it had up to 8 each of (terminl processor broards, disk manager boards (8089) and cpu/ram boards) and one of these, I'm told still runs today, it does payroll for the Navy according to Wikipedia. 

After than I did an englisj arabic cp/m computer for some really weird company, then pc graphcs stuff for Bell+Howell, then surface mount pick and place robots for the first Apple Mac at Excellon, then workstation software for Danford Assoc. in San Pedro (Next door to Logicon where my wthen wife worked and who made the Intercomp 143 printer in my high school, a 1403 clone) and this was ice because we used Apollo workstationsm which was refreshing after a bunch of CP/M and PC work.

Around that time I'd been there ten years, so I bought an AMiga abdmoved back to Toronto where I was one of thee entities that
offered internet service to the public in 1996 here in Toronto.
Instad of taking the easy way out and offering dialup for
big bucks I only offered www/cgi and briadband, which at the time 
was 128K ISDN (two bonded pairs)., So, less that what I was making, but 
it was nice. And fun.

Since then my focus was on two areas. Domain names and a thing I call IPS that stared as a web editor and is now a sprawling beast of a web
toolset I used to make large sites of pictures and words with a special
attention to typefaces and display/print technologies. We have a mutual
friend ir turns out, Chuck Bigelow, whom I hold in great regard.

My domain name work was a disaster. I had had some success on usenet
(cf. how aquarium newsgrouos got started on usenet 1987-1990) trying to 
do the right thing and in the end everybody gt what they wanted, but the domain name thing was ugly. Despite having Postel's blesing
and the dns software people and half the isp's, cprporate interests
wone out in the and and I like Dan Bricklin's quote that "Icann is  rent extraction scheme that has 404'd half the net". But we have new
top level domains and I apologize the names are lame and the whole
thing has turned into a cheap cel-phone popup in mall, in essence. But
I was there, at the heart of it, in DC, Geneva, Singapore, Ottawa but mostly in Toronto, I lived jsut south of Algonquin Park for 15 years
then oved back to toronto 10 yers ago.

I'm retired not, well shit, I've been retired since I was 40 but I didn't tell anybody. And, funny enough, the only person who noticed
was Einar Stefferud, who I assume you know, but if you don't he was Brian Reid and Jon Postels thesis advisor on Scribe and network junk
repectiviely. 

The last few years I've been working on code, painting bad paintings,
and have a goood life here in Toronto with my wife of ten years who
was an art student who wanted to be a programmer whereas I was a aprogrammer who wanted to do art and it worked. She's vicous 
javascript programmer that can find bugs I can't. And goodnss knows
I've had more practice doing that.

I recenly upgraded the 3 laptps here with 4K screens and all that
high speed thunderbolt and SSD stuff and it's taken me 6 months to 
figure out what is going on with cpu's thse days and I think I have
a good handle on it at least I can't find any more cpus to read about.

You Tube's service was now something I felt I had to py for and I
must say on a good sized (but NOT large) 4K tv, the oppertunity
of seeing you talk at Uniwat made youtube useful in a way few things
have.

So, I don't know if you managed to incent any of those folks there
but I am interested in the selection and optimization of the algoritms
you are working on. By "selection" I mean choosin the right
algoritjem dynamically. Look I don't know a lot about this stuff,
I did two years at Math in Uniwat, but I after differntial calculus
which I loved, came integral calculus which I just have no aptitude for
at all so I left, went the facuolty of Integrated Studdies where
I desifned a program of Biology and computers (taxonomy of West african
Cyprinodont "killifish and multirocessor OS and app research) which they 
approved and then I moved to LA, so that sort of stopped. I have kept
up with the Killifish work (see i.killi.es) and the plan9 gang seem to
have taken care of that.

But I'm really not a maath guy. I'k good at some of it and enjoy it
and do understnad what you taoked about rather wall I th9ougt, but I'm nowhere near the level of the math profs at uniwat. They're razor sharp
and I know few better. Plus they're all decent folk.

But when I hear "we do this and the cokmputer jst goes away and never
comes back" well, (cocks western hat) shoot son, we've been fixin' that
sorta stuff around these parts fer ages". While I may not be much help
I'd at least like to look at it and maybe in a while I might have some ideas. I'm pretty sure I can understand anyting if I go slowly.

So, there you have it. That's what  I'd like to look out and the only
help I can offer a prori about polynomial time is, what is the constaint? Ie, if you had say four $15,000 100-core Ryzan 9 threadrippers and it was ok in a few minutes does that help in lower
bound edge cases? Yes, I realie $60K of cpu is not much of an answer'but if you wanted to see what it works like on chea fast hw
that's what I'd do.

My analysis of computing perfomance with what's ouut there leads me to
conclude, and I say this as somebody who's used IBM/Lenovo laptops
all his life) that the mac mini kate 2918 sate grey machines are the best. The i5's are craY fase and they have three 40 Gigbit thunderbolt ports which can drive 4K video or all three in unison can be used as
a cpu interconcect at 120gb/s. They're about $500 here. Cheaper non
thunderbolt mc mini i5's are about $75-$100 here now. Even three of them is a lot of computing and decent i/o for the money. Of course
this only happens because as Brian Reid says "Macs are expensive high end fashioanle disposable coputers" and every time the new stuff comes
out, bang then dump the older ones. I had no idea Apple made SO MANY
computers now in thousands of individul condifurations just of that
"Mac Mini" machine, but wow.

So, I'm probably going to get three of some sort of mini as budget allows and use those for my new Plan9/Golang sandbox here. Nominally I have a 12 year old lenovo latop running GhostBSD on a dual core 2gig
cpu that's actually rather nice and old xp running on another the same
just to use one paint program. I'll learn vm and run this on a mini I think.


Sorry for the dump and I hope you can relate. I had a lot of fun doing all this stuff and I have a few cycles to spare I think and would be
y\tickled pink if I could be of any halp whatsoever. Even if you want to jsut know why your fishtank turned green.

Best regards, Don;
Richard Sexton

sexton@mejac.palo-alto.ca.us (server in Calif.)

This email is coming from this GhostBSD box in my home. I had no idea email worked without me having to mess with anythng but I can't see how it could know where to return it, so you'll probably need to reply to
the address above or to rjs79g@gmail.com

P.S. Due diligence suggests you contact Brian Reid who has mentored me unselfishly since 1986, Mejac is his server. He'll probably say "clever but no business sense" and you should talk to him because he's so fun
full of life and twinkle in the eye smart. Plus he's funny. And then you can tell him "He said you'd say that" when he does.


/rs