Saturday, June 13, 2026

IEEE TQE: Getting to Review (or, Some Friendly Submission Advice)

Hi, quantum researchers and authors! Some friendly, informal, totally non-binding advice from your Editor in Chief at IEEE Transactions on Quantum Engineering (TQE). (Written in November 2025, posted here with small edits June 2026; my term as EiC will expire in March 2028. My policies may change before I step down, and the next EiC will certainly have their own policies, so the content here is only provisional; however, most of it will be helpful at any point in time and for any journal, I believe.)

The number of submissions to TQE has been climbing rapidly this year. (Hoo boy -- I wrote that in late 2025, and it's doubly true in mid-2026!). Thanks to every who reads, submits to, reviews for, edits, and produces TQE!  The professional IEEE staff are amazing, and the editors tireless at a thankless job, so let me thank them publicly here.

With the increase in submissions, there are also naturally more papers with relatively easy to fix problems that slow down my share of the processing, and in turn slow down the whole journal.

So, when submitting a paper, ask yourself three questions:

  1. Is my paper going to generate a "clean" report in the plagiarism checker?
  2. Is my paper going to be easy for reviewers to work with?
  3. Is my paper tuned for this journal?

If every submission was clean on these fronts, my workload would go down at least 20-30%, and stalls in the pipeline would decrease even more.

1. So what is a "clean" report in the plagiarism checker?

Well, the checker returns a number, expressed as a percentage, that is very roughly the fraction of the paper for which it found matches with other materials in its stash (which includes almost everything in English on the web, not just other research papers in journals).

No paper ever written returns zero, so it's always a matter of making a judgment call.  No one is going to ding you for boilerplate or generic text like "A graph G={V,E} comprises a set of vertices V and a set of edges E," but that's a human call, not something the system can automatically handle. If a paper has a moderately high score, I often set it aside for a deeper look later, slowing the movement of your paper through the system.  So, if your paper will have a high score, please explain why in a cover letter.

The checker will pick up arXiv and other preprints as well as shorter IEEE conference versions of the same paper. Those are entirely legit, but I have to check; calling them out in a letter smooths the process.

If your paper draws on your own previously published work, YOU MUST SAY SO in a cover letter, and if this paper is an extended version of a conference paper, also in a footnote in the paper. i.e, "Portions of this paper previously appeared in x," or "This paper is an extended version of y."

In IEEE, there is no such thing as self-plagiarism. You are allowed to reuse your own work, with credit, of course. But a couple of issues come up: if the work is substantially the same, why bother republishing it? Explain what is new in this version.

And, as noted, there is the issue of copyright. PLEASE EXPLAIN THE COPYRIGHT STATUS OF THE PRIOR WORK. Without permission, IEEE cannot republish work where the copyright is held by others. If you still hold the copyright, or IEEE holds the copyright, we're good, but please tell me!

If you transferred the copyright to another company or organization, you also need to let me know, but we are probably going to ask you to rewrite it. If you really want to reuse the same text (or figure), it is your responsibility to get permission and make sure it is credited properly.

All of the above could have been written any time in the last two decades. The new twist is, of course, AI.

I'm not going to discuss a fully general AI policy here; that's for another time.

I get a lot of papers these days that have one sentence copied (generally but not always with one adjective or adverb switched for a synonym) from paper A, a sentence copied from paper B, a sentence from paper C...I don't think a human is deliberately copying from ten or twenty or even thirty separate sources. I'm pretty convinced that it's an LLM "helping" with the text, perhaps when translating from another language.

Of course more than half of the authors (and readers) of TQE aren't native English speakers, and using tools that help with English is legit. But you're still responsible for the final product.

(No, unfortunately, I don't have a pointer to a free plagiarism checker; building and running them is expensive, so they are paid services. If you're using an LLM, try asking it! If you are at a university, most of them have access to a checker, but it may or may not be available to you as a researcher rather than for use in checking classroom work.)

To recap, anything that makes the report complicated slows down the process.

2. Look at your paper. 

I mean, LOOK at your paper, then ask yourself if someone who hasn't seen it before will find it appealing and easy to read, and therefore be favorably disposed toward the material.

Here I mean formatting and presentation: figures, text, and equations. (I don't mean the style. To submit to TQE, you don't have to put it in TQE style; we will review papers submitted in RevTeX, basic IEEE journal or conference format, ACM conference format, etc. The paper will have to be reformatted to TQE style later, though.)

A pet peeve of mine is figures with fonts that are too small to read. If I have to zoom in to 200-300% before the fonts become legible, there is a good chance I will desk reject the paper and ask you to redraw and resubmit. That delays your paper by weeks, and makes more work for everybody. Generally, the font in a figure should be the same size as the font in the caption. This problem is exacerbated if the figures are bitmap rather than vector graphics, which suddenly seems to be a problem. If I zoom in and I can't tell A from R, the paper is definitely going back to you. Not only is bitmap worse for print and scalability, it's also worse for searchability and perhaps for accessibility for visually impaired readers. Vector graphic text is usually searchable with ctrl-F, which can be handy. (Photos of equipment, of course, are okay in bitmap format.) Unless it affects readability, I generally won't send a paper back just because some plots are in bitmap format, but we will usually ask you to provide vector format for final typesetting.

This applies to equations, too; I sometimes get papers where the equations are bitmaps inserted from some other program. This is a pretty big no-no. Microsoft Word's equations aren't as pretty as LaTeX's, IMO, but they are fine when done carefully. Learn to use your tools.

Equations and figures should be numbered. Figures should have a caption and be floating, not inline. It's also friendly to compare your figures against recommended practices for different kinds of vision impairments such as the different types of color blindness.

I sometimes get papers that are just plain ugly: lots of changes of fonts and even font sizes, different line spacing sometimes even within the same page, etc. Except when it's REALLY bad, I won't desk reject a paper for that, but you can be sure it starts reviewers with a negative impression.

Check your references; particularly using BibTeX, different style files will leave out different fields, and I get a lot of papers with incomplete references. Clickable DOIs aren't required but are helpful, and if you confirm that they work then you save yourself the embarrassment of sending me an LLM-hallucinated reference.

Think about the process of writing a referee report. Help the reviewer (and later, the reader) as much as you can. Referees who get papers that require effort to read will put off the task, delaying a decision on your paper. (Such referees may also be less inclined to agree to review other papers we send them, on the assumption that handling our papers is a hassle. So you help everyone else, too, by submitting as sharp a paper as you can.)

3. Know your audience.

In particular, for TQE, papers that begin by explaining X, Y, Z and CNOT gates waste space and opportunity -- what a collaborator of mine calls "momentum" -- and you will lose readers. For the TQE audience, it's not necessary, and I don't want it 100+ times per year in our journal. This is also strongly correlated with newcomers to the field, whose work is often too basic to represent a publishable advance, so once again you start off on the wrong foot with reviewers. If there is too much of it, I will desk reject the paper and ask you to fix and resubmit.

Conversely, it is fair to assume that TQE readers are unfamiliar with the basics of e.g. machine learning or your problem domain and to include some background material. But if you are spending a lot of space on purely classical ML, have you submitted to the right journal?

The scope of TQE is, broadly, quantum computing, communications and sensing. I desk reject papers on quantum-inspired (read: classical) algorithms and on post-quantum cryptography (PQC), which deserves to be reviewed, read and used by classical cryptographers and systems engineers. I also desk reject papers that are pure physics, without a clear connection to the use of superposition, entanglement and interference (discrete or continuous) to surpass classical systems. True supporting tech such as classical cryogenic processors or cooling systems for quantum computers is very welcome, though!

In short, to repeat, know your audience -- the readers, reviewers and editors.

Let me reiterate, this is informal, friendly, non-binding advice to help writers create papers that are ready for the TQE review process.

Good luck to all authors, I love reading your work. You are working in the most exciting field right now, IMO, and we are all on the same side -- building, deploying and using the best systems we can. Stop me and say hi when you see me at a conference!

Posted by at No comments:

Friday, June 12, 2026

Spelunking CACM, Vol. 26 (1983): Japan's Fifth Generation


1983 marked the first quarter century of CACM itself. The outgoing Editor in Chief, Robert L. Ashenhurst, marked the occasion with a January special issue reprinting some twenty-one of the best articles from the archives. The paper copy of this issue must be a collector's item. Every article in that is worth reading, even today, and I'm pleased to see that my own spelunking found many but not all of those -- some fun things to go back and look at later!

In February, Peter Denning took over as EiC, and instituted many changes, including going to a full-color cover (hinted at on the January cover).

I didn't realize the Computer History Museum went back this far, but there is an article by Gordon Bell on a visit to a famous computer, organized by the museum (which he founded). "The Computer Museum's first considered priority is to save history, the second is to display it, and the third is to interpret its historic role," we are told. (Oh, wow, Bell was visionary enough to begin organizing the museum in 1968, and incorporated it in 1982! Thank you!)

Programming Pearls came in this year, under the stewardship of Jon Bentley. The first one, appropriately enough, is about one of the canonical CS problems -- sorting data stored on disk.

Doug Comer wrote a history of CSNET -- while the project was still ongoing! It mentions our Dave Farber, but only as a participant/performer.

Cook's Turing Lecture on computational complexity (a religious document, if ever there was one) led me to an article on Ultracomputers, a switched (indirect) multicomputer architecture that Schwartz claims scales to thousands of processors. Now that's vision! I like that Cook dedicated a section to space-time products. I'm curious what the modern thinking on this is.

But for me personally, given my own location and interests, the most eye-catching item this year was the special issue with several articles on Japan's Fifth Generation (which refers to the programming language generation for non-procedural, goal-oriented computation), the astoundingly ambitious, MITI-directed, top-down AI inference machine project. (This is the source of the image at the top -- is there no higher-resolution version available?) Articles included an introduction (Pamela McCorduck, who co-authored a book with Ed Feigenbaum on the topic), a somewhat separate article on Japan's software management processes (Paul S. Licker), a trip report (Ehud Y. Shapiro) on the Oct. 1981 conference sponsored by the government, and an interview on the US response (Rosalie Steier interviewing B. R. Inman). The trip report in particular is a gold mine of anecdotal history and contemporary views of the project, and features the memorable quote, "We may argue whether it is better to think and program in Lisp, or think and program in Prolog; but both are certainly superior to thinking in Lisp and programming in Prolog."

Today, the Fifth Generation is largely considered to be a failure as a project, but my opinion is much more nuanced, and at any rate it's complicated. That's all for another day.

Posted by at No comments:

Friday, May 29, 2026

Some Top Quantum Researchers in Japan

 A while ago, I created a blog post on quantum architecture researchers in Japan, mostly young ones and all very good. But I'm often asked more generally about quantum research and researchers in Japan, so here is a list to refer to. Note that it's ordered by Google Scholar h-index (in parentheses; taken when I wrote this blog post -- we're not being scientific here), which is almost inarguably a terrible way to do it, but I don't have a better method at hand. h-index and influence in Japan both correlate with longevity and probably with other factors such as gender.

A lot of other factors influence publications and citations, including ability to both publish and speak in English, but conversely non-native Japanese people are at a disadvantage in the gossip and influence network inside Japan, including hearing about and ultimately influencing the creation of funding opportunities, so there are plusses and minuses.

A few important people don't have Google Scholar profiles (at least not public ones), so I have dropped them into this list entirely qualitatively roughly where I think they belong. Note also that a few of these people don't fully curate their Scholar profiles and so may have either extraneous or missing publications.

(Todai = University of Tokyo)

(Yeah, when you get to the bottom of the list, I've included our own younger people, whom I know better than the younger people in other places. There are a lot more researchers with h-indices in the 10-25 range. I picked some I like. So sue me.)
(I think the top 20 or so on this list is a relatively complete list of the senior people in the field here; if I've forgotten anybody, let me know!)

Posted by at 1 comment:

Tuesday, March 24, 2026

Quantum Networking: Drafty Internet Drafts

Speaking as an individual contributor, not as chair of the Quantum Internet Research Group:

We have two new very drafty Internet Drafts out:

Timing Regimes in Quantum Networks and their Physical Underpinnings
Get the HTML version, which renders the figures, names and equations right.This one is intended to bridge the physics and the network engineering; there should be no design decisions in this document, but it should tell us how to use the physics to decide. It should be useful across all possible quantum network designs. In particular, I originated the document to help us understand layering in quantum network architectures.

Comments submittable as email (either direct to both authors or on the QIRG mailing list), or as Issues or Pull Requests on GitHub.

A Quantum Network Architecture
This one is the start of the description of our own quantum network architecture. Still very far to go. 

Comments submittable as email (either direct to all three authors or on the QIRG mailing list), or as Issues or Pull Requests on GitHub.

Comments very welcome on both documents!

These two are the start of what we expect to eventually be a full technical description of our network, spanning as many as two dozen documents.

Posted by at No comments:

Voluntary-Participation Surveillance Society and, oh, Little Things Like Aircraft Carriers

 Hilarious, inevitable, tactically troubling, and evidence of our voluntary-participation surveillance society. https://www.lemonde.fr/en/international/article/2026/03/20/stravaleaks-france-s-aircraft-carrier-located-in-real-time-by-le-monde-through-fitness-app_6751640_4.html

Strava is the app I use to record my bike rides; "everybody" is on it. It can obscure your start and end points if you don't want the exact location of your house identified, but if you're on a public app like this, you've already volunteered to share a great deal about yourself.

Warfighters are heavily warned against sharing opsec-sensitive info, and I've been told that modern briefings instruct them not to use Strava while on deployment. But it was inevitable that someone would forget. I'm sure the poor guy has already been disciplined and his shipmates won't let him live it down, for sure.
Posted by at No comments:

Friday, March 06, 2026

Dave Farber: Obituaries and Archival Materials



As you know by know, our beloved Professor Dave Farber passed away last month. The impact of his loss will resonate for a long time. Here is my list of articles about his passing, plus a few other things. Let me know if you see other articles that should be included. (Picture above: we cooked Dave turkey on Christmas Day, 2025.)

Perhaps my favorite quote from this set of sources is from Dave Crocker, in the Wall Street Journal:

His mind was, from what I could see, largely undisciplined, which is how he could make these connections so unexpectedly and usefully, but not always usefully. He would say things that put things in juxtaposition because his mind just wandered in various ways.

To me, that's high praise.

Notices from Keio University & Japanese Organizations

Other Employers
Major Newspapers
Trade Organizations
Historical Materials Archived at Keio
Other Historical Materials
Posted by at No comments:

Monday, February 16, 2026

Eulogy for Dave Farber

[Photo is at Miyajima, Hiroshima, on Dave's 90th birthday trip: Keiko Okawa, Yukie Shibuya, Yasuo "tsucchy" Tsuchimoto, Kaori Suzuki, Rod Van Meter, Dave Farber, Jun Murai, Catharina Maracke]


 My comments from Dave Farber's funeral on February 13, 2026:

David Jack Farber.
Born April 17, 1934.
Passed away February 7, 2026.
Occupation: professor.

Those are the basic facts you will see on a census form.

Worked on telephone switching systems, built distributed systems, helped kick off important precursors to today’s Internet, served as chief technologist at the FCC, taught at five universities – those are the things you’ll read in the obituaries in the newspapers, and we heard about them yesterday from Manny, Jun, and Jiro. We heard from Manny about family, and Dan about Dave’s love of food and of Japan. (For those of you who didn’t join us last night, we hope to make the video available later.) And in a few minutes we will hear from Vint Cerf, whose career Dave helped shape.

Dave literally changed the life direction of several Keio students that I know of, and doubtless affected many more in ways I’ll never get to hear about. As a Keio faculty member, I’m deeply grateful for that. After Vint, we will hear from Taro, who is one of those whose life was changed.

So what more is there to say?

If I had to describe Dave, I would pick the 3 Cs: connection, clarity, and curiosity. Yesterday we talked about all three of these.  Dave's ability to connect people might be the true core of his career – it’s the thing that brought us all here today.  Dave was exceptionally clear in his explanation of things – and also saw clearly. Dave’s curiosity was endless; personally, it’s the thing I admired most about him, and he and I had long conversations about wide ranging topics (yes, including quantum computing), and it might be the core of his longevity.

A few days ago, while helping Manny clean up the apartment, we found a number of articles that Dave had printed out for later reading. I found a couple of things I had written on Dave’s desk – my blog entry on web3, and my class and research group policy on the use of AI. That's certainly an example of his curiosity.

Dave was not a particularly religious person, though he was proud of his Jewish heritage. G.G. was raised Greek Orthodox. I was raised American Baptist. One place where all of these things overlap is in the Book of Ecclesiastes (which, by the way, takes its English Christian name from the Greek). Ecclesiastes is known as Kohelet (which I’m sure I’m not pronouncing right) in Hebrew.

So, with your permission, I’d like to read a little from the King James Version of Ecclesiastes. It will be a familiar passage, and it’s one that’s often used in Christian funerals, at least in America, but it should resonate with everyone regardless of religion, I hope.

3 To every thing there is a season, and a time to every purpose under the heaven:
2 A time to be born, and a time to die; a time to plant, and a time to pluck up that which is planted;
3 A time to kill, and a time to heal; a time to break down, and a time to build up;
4 A time to weep, and a time to laugh; a time to mourn, and a time to dance;
5 A time to cast away stones, and a time to gather stones together; a time to embrace, and a time to refrain from embracing;
6 A time to get, and a time to lose; a time to keep, and a time to cast away;
7 A time to rend, and a time to sew; a time to keep silence, and a time to speak;
8 A time to love, and a time to hate; a time of war, and a time of peace.

Dave knew all of these things. Dave did most of them rather frequently. We have done about half of these things just in this last week, talking about Dave and preparing for today.

Dave was much more than a colleague and mentor. Dave was my friend, and I will miss him.

My condolences again to Manny, Mei, Carol, Nate and Sam. Thank you for sharing Dave with the rest of us. May his memory be a blessing.

Go in peace, Dave.  G.G. and Joe are waiting for you.


Posted by at No comments:

Friday, January 16, 2026

TDC toolkit new release!

 If anybody out there is using our Multiharp toolkit for counting photons, a new release is out:

https://crates.io/crates/tdc_toolkit
https://docs.rs/tdc_toolkit/latest/tdc_toolkit/multiharp/device/struct.MH160DeviceConfig.html
https://github.com/moonshot-nagayama-pj/tdc_toolkit

Thanks to Andrew and the rest of the team who do the hard work!
Posted by at No comments:

Tuesday, December 30, 2025

Breakthroughs of the Years

 Since 1996, Science magazine has named a single "Breakthrough of the Year" as well as several runners-up. This year's is the growth of renewable energy, especially solar. But what about the past thirty years, can we see any themes or a bigger picture? Here is a quick look at the categories, as roughly laid out by me:

  • Life sciences: 17
    • Human health: 8
      • HIV (understanding, treatment as prevention, lenacapavir): 3
      • Cancer (immunotherapy): 1
      • Other (stem-cell therapy, genetic variation, COVID-19 vaccine, GLP-1): 4
    • Evolution (evo in action, Ardipithecus ramidus): 2
    • Other bio/life sciences (cloning Dolly, whole-genome sequencing, RNA interference, reprogramming, CRISPR, single-cell sequencing, AlphaFold): 7
  • The physical universe: 9
    • Cosmology and "deep" astronomy (accelerating universe, dark energy, gravitational waves, neutron star merger, black hole VLBI, JWST): 6
    • Exploring the solar system (Spirit, Rosetta): 2
    • Particle physics (Higgs): 1
  • Technology: 3
    • Nano/quantum tech (nanocircuits, quantum machine): 2
    • Energy tech: 1
  • Mathematics (Poincaré proof): 1
That's thirty years of astounding science, for the benefit of humankind as well as sheer curiosity.

It would be worthwhile to compare to the annual list at Physics World, and also to look at the runners-up and the "busts" of the year, which Science also reports, but that would be more work than I care to put in this morning.

It's humbling to be reminded that quantum information is at best a small corner of the global science and technology effort. Even the broader field of computing, with all it has done in the last thirty years, makes the list above only once, and at that for its contribution in understanding proteins rather than the technology itself. Of course, technology is very often about incremental accumulation of small advances, rather than "breakthroughs". Still, time invested in thinking about where we have been, where we are, and where we are going would be well spent.

Posted by at No comments:

Monday, December 08, 2025

Quantum Curricula

 (Just a quick note for now, more later.)


Posted by at No comments:

Saturday, November 15, 2025

Spelunking CACM, Vol. 25 (1982): Worms and Grapevines



 Wow, right away in January there are things I find relevant to our work today. One such is an article on a test to see if documentation of control flow or data structures helps more in understanding a piece of code, coming down firmly on the side of the data structure. Of course, that was a very limited test done with a short program and without the benefit of today's tools and practices, but it's nice to see where we were at the time.

I lived through the Morris Worm in 1988, and I knew the name came from a Brunner SF novel, but I either didn't know or had forgotten there were not just technical analyses but implementations and tests of worms before 1988! (The image at the top is taken from this article.) Fascinating study there on a multi-node, multi-segment worm that is extremely hard to kill. Prescient, and frightening. It has been a long time since we had something quite like it, despite the many forms and instances of malware; let's all hope we have built a resilient system, robust and well-defended against it happening again. The article also includes a nice summary of early ARPANET distributed programs, including pointing out that routing is itself a distributed computation -- an under-appreciated insight even today.

One article that I teach, even today, is on Grapevine, which is one of the seminal distributed systems. Everyone in computer systems should know about that work at Xerox PARC.

The ACM Classification System originated in 1964, and was redone here in 1982. With the obvious exception of quantum computing, I find that classification surprisingly solid, more than forty years later. It has since been revised, in 1991 and 1998, but not in over a quarter of a century, so people either don't use it or our forebears Simply Got It Right.

State of the art in AI in 1982? A 20-page effort to build an English grammar diagram useful for parsing sentences.

I haven't stopped to look more closely, but the February issue has several articles on queueing systems analysis.

A couple of personal interest: one on debugging via program slicing, a technique one of my Ph.D. graduates used for quantum programs, and one by Bill Swartout and Bob Balzer on the intertwining of specification and implementation. Just a few years later I would be the juniorest worker at USC/ISI, where they worked.

For what it's worth, fall of 1982 is when I entered college. Next year is my fortieth reunion, looking forward to seeing people I haven't seen in a long time.







Posted by at No comments:

AI Policy: Whew

 I am working on a policy for appropriate use of AI (especially but not only LLMs) in my classes and research group. It's an important task, and I'm trying to not just lay down a set of written rules to be blindly followed but a bit of history and a set of principles. I do expect the policy to be revised regularly as the technology evolves, but I want to clarify my own thinking and encourage students to both think for themselves and read as widely as possible.

This is a nontrivial task...

Posted by at No comments:

Tuesday, October 21, 2025

Takaichi Sanae: Japan's First Woman Prime Minister

237 votes for Takaichi Sanae (高市 早苗) on the first ballot in the Lower House. It's official, she will become Japan's first woman prime minister.

Interestingly, she worked for Democratic (and liberal) Congresswoman Pat Schroeder of Colorado as a congressional fellow in the late 1980s, when Takaichi was in her mid-twenties. Either she was still forming her political views or she didn't know enough about American politics to realize who she was working for, but for most of her political career she has been conservative to very right wing.

Koizumi Junior will become Defense Minister. I haven't seen a full cabinet lineup yet. Always interested in who the ministers with large research portfolios are, though I rarely know enough about them to have an opinion.

In the photo (mobile phone photo of a television) at the top, Takaichi sits in front of three former prime ministers at the back of the Diet Lower House. I know so little about Japanese parliamentary procedures that I have no idea how the seating chart works. (Actually, there are many more important things I also don't know.)

Posted by at No comments:

Thursday, September 11, 2025

MOSIS in the 1980s

 

I found some of the key history of the MOSIS Project I was looking for!  The table above is from a 1991 DARPA report, DARPA Technical Accomplishments. Volume 2. An Historical Review of Selected DARPA Projects, Accession Number: ADA241725. During the 1980s, MOSIS fabbed 12,201 projects (computer chips designed by researchers from universities, government labs and companies) in technologies ranging from 5-micron NMOS to 1.2-micron CMOS.  What an astounding total!  I wish I also knew how many wafers, chips, designers and organizations are represented by those numbers.

I also found two key annual reports summarizing all (most?) of USC/ISI for DARPA, 1980 and 1982.  Unfortunately the absolutely crucial year of 1981, when MOSIS went live, I have been unable to track down so far.  The 1980 report states:

In August 1980 MOSIS accepted designs for the first fabrication run using the software developed for automatic interaction with users. This run had 65 projects submitted by designers from 8 organizations: ISI, UCLA, Caltech, Jet Propulsion Lab, Stanford University, National Bureau of Standards, Carnegie-Mellon University and Washington University at St. Louis. These projects were packed into 18 dies on the wafer.

And then elsewhere:

After a period of debugging, the MOSIS system became operational in January 1981.  The MOSIS system accepts VLSI designs expressed in CIF [1] and handles all the fabrication steps needed in order to provide the designers with actual chips.

The original VLSI research staff were: Danny Cohen, Yehuda Afek, Ron Ayres, Joel Goldberg, Gideon Hess, Dave Johannsen,  Lee Richardson, and Vance Tyree.  Support staff: Victor Brown and Rolanda Shelby.

By the 1983 annual report, we learn that, "Over 40 universities and hundreds of designers now submit VLSI designs in electronic form via any network to MOSIS. MOSIS delivers chips and will soon deliver user-specified printed circuit boards to designers 30 to 35 days after receipt of a design."

The 1984 annual report lists 35 total MOSIS "runs": 17 4-micron nMOS runs; 5 3-micron nMOS runs; 11 3-micron CMOS/Bulk runs; and 2 3-micron CMOS/SOS runs.  By the time I began working at ISI in June 1986, runs were "closed" weekly on Thursdays.

It really was an incredible period in computing technology, and MOSIS was an essential contribution.  I was young (20, when I started working at ISI) and stupid (okay, I've still got that part), and so I had only the tiniest glimmer of how important everything going on around me was.

Now that I have found most of the hard data I wanted, I can update the Wikipedia entry on MOSIS!

Posted by at No comments:

MOSIS History: Help a Guy Out?


 If you're reading this, and you know anything about the early days of the MOSIS Project at USC/ISI, mosey on over to https://en.wikipedia.org/wiki/Talk:MOSIS and drop a note there.  I'm trying to enhance the article on MOSIS, but of course it needs to be documented history, not just, "I remember..." or "I think..."

If you don't know what MOSIS is, it's the Metal Oxide Semiconductor Implementation System.  It's a multi-project wafer fabrication service dating back to the early 1980s, and it transformed VLSI research in the United States.  (I've never understood why it was emulated aggressively worldwide, though a few other MPW systems have existed.)

I worked for MOSIS for a couple of years, though my job was on the compiler for the ICL programming language that most of the processing tools where implemented in.  I had basically nothing to do with the VLSI side or the production run side.

Image at the top from the March 1984 user manual, available at https://apps.dtic.mil/sti/citations/ADA139744.

Posted by at No comments:

Saturday, August 09, 2025

August 9th (Nagasaki)

 

Yesterday (August 8) we were in the city of Nagasaki. It's a beautiful little place, with a charming streetcar and centuries of important history, especially in Japan-Dutch relations during the Edo Period. And yet, of course, its most famous role in history is now, and very likely will remain, as the target of humanity's second atomic bombing of other humans.

We visited the Nagasaki Atomic Bomb Museum.  I was last here at New Year's 1994, and the current museum building opened in 1996, so this was effectively the first time for me. The museum is, if possible, more powerful, moving and terrifying that Hiroshima.  The exhibits pull zero punches.

The museum opens with some of the most disturbing images and artifacts anywhere on Earth. From there, it goes into more personal stories and mementos.  It also covers in surprising depth the physics and medical impact of the explosion.  It concludes with a history China-Japan conflicts starting in the late 19th century, a separate timeline for the Pacific War, followed by postwar nuclear developments and politics, with a moving call for global disarmament.

Many of the exhibits have at least short accompanying explanations in English, but many also do not; they are probably covered in the audio guide that I didn't borrow.  The war histories state dates for key events, such as the Manchurian Incident and the attack on Pearl Harbor, but without any description, so interpretation is left up to the viewer and their pre-existing knowledge of history.

We were there the afternoon before the 11:02 a.m. 80th anniversary.  There was a scattering of foreigners in somber suits, most with guides/interpreters.  I presume they were there as guests for today's ceremony.

I took a few photos (which is allowed through most of the museum), but most of them feel too raw, too personal.  The photo at the top of this posting is from the memorial, which is attached to the museum.  The glass tower in the middle back normally holds books listing the names of the many victims, but a small sign said that the books had been removed in preparation for use in today's ceremony.


Posted by at No comments:

Thursday, July 31, 2025

History of the Asian Internet

 If you're a history buff, a collection of writings on the history of the Internet in Asia:

https://sites.google.com/site/internethistoryasia/home?authuser=0

Posted by at No comments:

Monday, July 28, 2025

Bollocks

 Professor Peter Gutmann, of the University of Auckland, a well-known computer security researcher, has made something of a name for himself as a quantum skeptic, at least with respect to quantum cryptanalysis and post-quantum cryptography (PQC).  His argument is roughly two-pronged:

  1. Quantum computers aren't making progress on factoring at all; and
  2. even if they did, computer security people and cryptographers have much larger problems to worry about.
I agree with him pretty strongly on point #2.  I've said various times that the mathematical vulnerability of Diffie-Hellman key exchange or RSA authentication is not very high on the list of corporate CSOs.  On point #1, I'd say he's technically right about the current state, but rather dramatically wrong about concluding that he can project onward for the next couple of decades and declare the world safe from quantum computers.

Sometime in the last year or two, Peter gave a talk somewhere titled, "Why Quantum Cryptanalysis is Bollocks". The slides themselves have no date or talk venue, but evidence from within them suggests mid-to-late 2024. Some blog postings also assert that date, as does a very recent article in The Register, which is how Peter and this talk came to my attention.  The slides make the talk look like a lot of fun, so I wish I could hear it in person or even via recording.  So let's take a look at the logic in the talk, then I'll tell you why I think he is misunderestimating the quantum folks:

  1. Germany had a massive weapon systems boondoggle during WWII.
  2. OWASP lists a lot (tens of thousands!) of threats to the security of computer systems, and the highest-ranked attack on the mathematics of encryption was #17,245 (not a typo).  Roughly, the argument is:
    1. Success with mathematical attacks are high-effort and low success probability;
    2. even if you succeed, you recover a few bits of the contents of one message; and
    3. of the top ten high-priority problems, when you succeed you win big -- you get "the plaintext of all of the messages."
    4. (And holy cow, not directly in Peter's talk, but there are always examples of how human stupidity is the number one threat!)
  3. NSA can already factor 1,024-bit RSA keys, if they're willing to commit leading-edge supercomputer time in allocation chunks of a year at a time.
  4. Quantum computer-based factoring has grown from 4 bits to 5 bits over the last 20+ years.
  5. Quantum computers are physics experiments, not computers.
  6. (Brings up poorly-regarded D-Wave factoring.)
  7. PQC is very hard to implement well, and costs a lot in engineering time and network/execution time.
  8. (Various disparaging of academics and standards organizations as becoming self-justifying.)
  9. "Quantum cryptanalysis is the string theory of security" and my dog can factor just as well.
Let's see...let me respond one by one:
  1. Yes, and while cautionary tales are good, cautionary tales are not the same thing as prediction or even valid analogy.
  2. Okay, but I think the implied message here is off: if you can crack one Diffie-Hellman key exchange, you gain the ability to read everything in that conversation (n.b.: it's harder than just factoring or discrete log, there are other mechanisms involved), but the bigger catch would be the RSA private key of an important individual, which would allow you to impersonate them across a range of systems; certainly there are organizations that would pay a lot of money for that.  Of course, I'd argue that it's pretty easy for truly high-value targets connecting to high-value systems are likely secured via shared private key, so hacking RSA is lower value. Peter is definitely more knowledgeable than I am in this area.
  3. Okay, but is that relevant to the anti-quantum argument? Is the argument just that people won't really commit big resources to factoring? I'd like to hear the oral argument that accompanies this point.
  4. This is the big one: he's saying progress in development of quantum computers is so poor that we can effectively discount them as any sort of threat. Ooh...okay. It's a fair point that reported successes in the literature on the task of cryptanalysis are advancing at a glacial pace.  (We have worked on this topic.) But projecting from past progress to future progress is dangerous in this field. We have known since the beginning of this field that error correction would be necessary.  Until we hit the threshold that allows quantum error correction to be executed effectively, progress on hardware did not translate into equivalent algorithmic successes.
    Well, guess what? The relentless improvement in hardware means we have passed that basic threshold on at least two very different hardware platforms in the last two years.  At least two other companies have released roadmaps that take us to large-scale, fault-tolerant systems within five years. At that level, that means they think they know how to solve all of the problems standing in their way.  Even if they are off by a factor of two, that still means we're there within a decade, I'd bet sooner.
    So my opinion is that pooh-poohing the likelihood of the advent of cryptographically relevant quantum computers (CRQCs) seems unwise.  I think it's bordering on irresponsible to assume the technology won't happen; the argument instead needs to be about how much to prioritize countermeasures.
  5. In today's environment, strongly agreed.  Dave Farber said to me several years ago (perhaps as far back as 2019, though I think it was a little more recently than that), when I showed him some Qiskit code, "This isn't an application, it's an experiment."  I think we as a community need to think very hard about how to deliver hardware, runtime systems and tools, and applications to customers.
  6. (Pass.)
  7. Cost of PQC is high -- oh, yes, definitely.  I attend IETF meetings and listen to people moan about how hard it is.  I'm not an expert here, though.
  8. (Pass.)
  9. Funny!  (I need a dog.  I love dogs.  But I'm allergic to dogs, work too much and travel too much, and I think dogs in Japan don't have good lives, but all that's for a different posting, some other day...)
Verdict: pundits, and occasionally quantum people, oversell how soon and how big the impact will be -- I'd agree with that.  But the machines are coming.  Make no mistake about that.  So, it's up to you, as a business person, engineer, researcher, bureaucrat, CSO.  How will you respond?

(Also, if you care about broader skepticism of quantum computing, you may want to go look at a blog posting I wrote about four years ago. Geez, time flies.)

Posted by at No comments:

Tuesday, July 22, 2025

Spelunking CACM, Vol. 24 (1981): Hierarchical Storage Management, OS Support for Databases


Alan Jay Smith was a consultant for a company I worked for in the late 1990s. Very smart guy, already seemed "old" to me then, but he was quite a bit younger then than I am now.  Geez.  He worked on processor architecture, but his most influential work has been on storage systems, and here in 1981 he was already deep into what became known around this time as hierarchical storage management (HSM) systems.  Even well before this article, though, mainframe computers used a combination of hard disk and tape to hold user files (as in the figure above, taken from this article) that were considered active; if your file had suffered the unfortunate fate of being evicted from disk, when you asked for the file, it had to be fetched from tape -- a task that could take a couple of minutes up to an hour, depending on whether tapes had to be fetched and mounted by a human.  Alan worked on understanding inter-reference patterns, hopefully reducing the incidence of that unfortunate fate.  He found that not only was it a long-tailed distribution, with many files never accessed beyond two days after their creation, but among those whose active lifetimes were long, some showed periodic use that could be predicted and taken advantage of.  Modern server and storage farms operate at a scale and level of parallelism inconceivable in the 1970s and early 1980s, but I wonder how many of the designers of today's systems are familiar with the foundational work done by people like Alan Back in the Day.

Preparata and Vuillemin described CCC, Cube-Connected Cycles, as a compromise on the physical infeasibility of a large-scale hypercube interconnect.  They limit the overall structure two an ordinary three-dimensional cube, with each vertex replaced by a ring of nodes sufficient to make up the extra dimensions; i.e., if what you want is a 5-cube, the overall structure gives you 3 dimensions and each ring emulates 2 dimensions, so each vertex is replaced with a 4-node ring, as in the figure below.  Later analysis by Dally would propose k-ary n-cubes with small n instead of this somewhat more irregular structure.



Lamport proposed a password authentication system in which the terminal actively participates; the user identifies themselves to the computer, the computer issues a challenge (from a predetermined challenge sequence), the terminal calculates the challenge using the user-supplied password. With this scheme, the computer (or server, in a modern client-server system) never holds the password or enough information to allow an eavesdropper to create and crack the next iteration of the challenge.  Of course, this requires that the terminal (client) itself be secure. I'm not enough of a cryptographer to know how influential this particular paper is, but it's worth noting that it comes five years after Diffie-Hellman (one of only two references in this "Technical Note" short paper).

David Patterson implemented a high-level language, called STRUM, for microcoding a Burroughs D-machine. I have done very close to zero microcode myself (one student project during my master's degree, and reading some VAX microcode around the same time), though I suppose things like SCSI processors and network packet processors are about halfway in between ordinary instruction sets and microcode. Microcode programmers are Titans of the old gods; it's a very early idea, attributed to Maurice Wilkes himself. I don't think Patterson's approach ever really took off; as far as I am aware, writing microcode still involves very low-level management of latches and data routing and the like, not really amenable to high-level languages, in my opinion. Interesting project, though.  (Interesting side note: this article took over two years from initial submission to acceptance.)

One article I read as a grad student is Stonebraker's Operating System Support for Database Management. Today, it reads like a fairly basic, almost plodding analysis, but it was groundbreaking at the time, and in my own field this kind of clear-eyed analysis of how good a job quantum hardware and software do for particular tasks should become standard practice.

Overall, I have to say, not CACM's most exciting year.

Posted by at No comments:

Monday, July 07, 2025

I Believe in Web3...Just Not That Web3

(Note: This was originally written in summer and fall 2022, and for various reasons I decided not to publish it then, despite the obviously enormous amount of work I put into it. One reason was that I wasn't completely satisfied with it, so I still consider it to be a work in progress. Given the enormous volumes of writing out there dedicated to the broad topic of web3, it would also be rather bold of me to think that I have much new to offer. This is, instead, a way to organize my thoughts, which I am willing to share with you. If you are kind, I will be happy to carry on a dialog that might improve my understanding, as best my time permits.) 

I believe in the re-decentralization of the web. I want creators to be paid and I believe in the idea of micro-payments, and it would be nice if that meant something other than advertising. I love Larry Lessig's idea of managing the incentives that web2 companies have to create addictive, outrage-driven products. Although the Russian aggression in Ukraine brings the whole liberal project into question, I still think increased flow of goods, people, knowledge and principles across borders leads us toward a better world. And Joi Ito's description of web3 as being about community brings it into focus, gives it a direction.

What I don't believe in is most of the technologies being touted as transformative and necessary to web3.

What I don't believe in is blockchain as a currency, a store of value, or a speculative investment. I do believe it potentially has value as a public record of certain communications. It is also a brilliant technical innovation, still searching for the right way to be applied.

I very definitely don't believe in NFTs. I can't see what value they add at all.

And the metaverse...ah, the metaverse.

Backing up for just a second, blockchain, NFTs, and virtual reality/metaverse are the rather disparate technologies that are getting welded together and touted as the cure for everything that ails today's World Wide Web (and there is a lot that ails the web). Collectively called web3, the purported win is that they enable DeFi (decentralized finance, in contrast to CeFi or the "fiat economy", which is never used as a compliment), DAOs (distributed autonomous organizations), and more.

Supposedly.

And supposedly, in at least some tellings, not only do these technologies solve what ails the web, they solve some fraction of all the world's problems.

Which means, I suppose, we need to begin with a quick look at what some of those problems are and how we got here before we look more closely at the technologies on offer. 

The Vision

Among quite a number of things I read, the longest and most coherent was Joi Ito's book in Japanese, 「テクノロジーが予測する未来」(tekunorojii ga yosoku suru mirai, or The Future that Technology Predicts, more or less), which I read most of. It partly inspired this posting, so I will be referring to it quite a bit, but I don't want this to be just an analysis of Joi's arguments. The book is very focused on the notion of community and the ability to quickly create and scale up new communities. Joi is well known for being quick to "try on" new ideas, always looking for something to remake the world, so his thoughts are interesting even if not always as grounded in technological or human feasibility as we might wish.

Joi describes web1.0 as read, web 2.0 as write, and web3 as join (or maybe "participate"). This takes us from 1.0's monolithic web servers whose installation and maintenance and publication required significant technical expertise and capital, through blogs (and the early days of online shopping), to 2.0's Facebook-dominated world of SNSes where anyone can write or share photos far more easily and a handful of hypergiant e-commerce corporations control what we search for, buy, read and use. Today, the buzzword web3 is supposed to help us build large-scale, global, autonomous communities, with as little regard for existing prejudices, practices, rules and laws as we can get away with.

Joi describes project-based organizations that come and go like movie productions. Given the chaos, stress, grift and uneven distribution of rewards in moviemaking, I'm not so sure that's an attractive description. It also sounds like a macro-scale gig economy, where many people have to hustle for every dollar, and I am certain that is not the right model for everyone (though it may be for some).

The word efficiency comes up repeatedly. To Joi, this seems to be the heart of what these new technologies bring, and it is a seductive Siren. After all, the worldwide web itself is "only" a more efficient way of publishing and sharing information. If a new efficiency really takes hold, it can transform the world. 

Ownership of not only the things you create but the things you buy is frustratingly difficult to even understand, let alone manage, today. Amazon can delete things from your Kindle, and prevents you from selling them on to others. HP can remotely disable the printer you bought and paid for, if your credit card that must be filed with them expires. John Deere has...a complicated relationship with the right to repair something you own. (Okay, now we're getting a little far from web3.) John Deere also lays claim to the data that their Internet-connected farm equipment collects. One of the principles of web3 is to return ownership of data to those who generate it.

The Internet is sometimes touted for its "permissionless innovation" (which, of course, is how we got the web itself -- no one had to give Tim permission to deploy the first web server). I would called DAOs "deferred legality". Spin up a quick bulletin board or server or Github project for the community to meet, invite others in, establish a handful of rules on how decisions are made (By humans? By an algorithmic "smart contract"? How do you allocate weight in voting?), how work is distributed and how people are paid (in tokens, presumably, whose utility for buying real-world groceries may vary), and you're in business -- maybe literally selling something, maybe just collectively creating something fun.

Joi has a chapter on how web3 will advance education. His first two points are about the gamification of education, and proof of credentialing. I'm all in favor of learning being fun, and if setting it up as a quest makes it easier for today's kids to follow and finish a plan, that's fine. For credentialing, inflated resumes or outright forged diplomas are a real-world problem, so I think it makes sense to have some sort of digital certificate with cryptographic non-tampering authentication that can easily be checked by prospective employers, ideally single-click from a submitted resume.

web3 is touted as a way to disintermediate a lot of creative businesses and see that creators get paid more directly for their work. Some people even talk about it as a means of improving digital identity, reducing forgery and impersonation and outright theft of digital identities. It is even described as supporting better democracy.

I'm really not sure I've done the vision justice, but at best it's amorphous and so far this is the best that I can do.

The Technology

Is there really a need for me to describe the tech here? Let me toss in a short description anyway.
Blockchain is a distributed ledger, with recorded entries that are proposed by an individual node (is "client" the right word here?), then committed using a large-scale distributed computation, involving many nodes (see below on "Getting Squeezed"). After being committed globally, a record can be more quickly confirmed using a local computation. To paraphrase one of my students, it's a single, world-writable, world-readable database. Anyone can record essentially anything on the blockchain; as it happens, the first proposed use was as an ersatz form of currency that people are expected to exchange for real-world goods.
NFTs are...well, what are they??? An NFT is just a token on a blockchain that represents...represents...that you, um, bought something digital, like an artwork, from someone, who may or may not have retained the copyright and may or may not make more of them and in any case even if you are the one and only legal owner of the digital artwork you can't stop someone else from copying it so if you want to really enforce your legal rights to it, go ahead but I don't think anyone has actually tested this legally yet.  (Whew. Ugh. I know. This really belongs below in the discussion of what's wrong with the technology, or maybe just below here in the section on legal matters, but I can't find any positive way to describe what an NFT is. Want a better explanation? Google it up yourself. There are a zillion people out there eager to explain to you why you should spend money on NFTs, but far fewer who can really tell you what one is and what you are actually buying.)
All right, I give up on the technical explanation unless you want me to talk about hashing rates and cryptographic security and the like. If you find anyone who can give you the Heilmeier catechism on either blockchain or NFT, let me know.

The Legal Matters

We actually just covered some of the legal issues of NFTs, sort of, so let's look at DAOs in the real world.

Joi touts the innovative, token-based governance of DAOs, but it's really not clear how contract disputes, labor disputes, legal liability, taxation, and adherence to international norms should be enforced. Environmental and workplace regulations (not to mention rent, equipment and insurance of all forms) are pushed to the individual participants. Given that the vast majority of DAOs will remain too small to bother with, it can be argued that the deferral of resolving legal status is the right approach, but it's worth noting that this even includes the matter of legal jurisdiction.

Of course, I am not the first to think of any of this. In fact, Wyoming already has a law on the books on how to incorporate an LLC for a DAO. Tennessee also has such a law, signed into law in April 2022, and aims to be the "Delaware of DAOs". Services to help you set up a Wyoming DAO abound, and they encourage you to put in as much real money and to clarify these issues as completely as you can. Of course, the more completely you specify these things up front, the more it looks like a conventional small company with employee ownership, but there do seem to be differences. Consider, for example, a smart contract used to make decisions with impact outside the immediate group, such as buying or selling something. What if the transaction happens to be illegal in one or more jurisdictions? Who can be held liable, and how can the DAO be modified to make sure it doesn't happen again?

The Fit: Does the Tech Do the Job for Web3?

In a word, no.

Transactions

In meatspace, transactions differ in complexity and longevity. Buying a house requires much more paperwork than buying a donut. Not only is there more at stake financially, the world at large cares much more and for much longer (centuries, even) that I bought a house. It is natural that real estate transaction processes are more complex, and very little of that complexity has to do with the difficulty of physically or electronically signing, attesting and publishing documents or even moving the bits that represent large bank balances.
In the middle of the complexity and longevity scale are our everyday purchases. Lattes, gas, train and movie tickets, lunch. Many of us, worldwide, make several purchases a day, and a system that aims to handle all of this therefore needs to handle, oh, say, a hundred billion transactions a day.
At the far end of the spectrum is the ephemeral, ubiquitous, continuous cascade that is billions of people surfing the web. To build a micropayment system that pays content providers as you surf their contributions, we need to target something that scales to literally around a trillion transactions a day. 
How do we do this today? Well, via true distribution of the work involved. The vast majority of transactions worldwide are small, two-party transactions that don't need a true, worldwide, globally readable record. Maybe the coffee shop needs to summarize its sales at the end of the day, and the money in your possession needs to be transferred to them at a mutually agreed-upon exchange rate of coffee to currency. (In fact, that's part of what currency actually is: a medium of exchange, a store of value, and a unit of accounting.  It's such a common trope that there are Cliff Notes for it.) That money "in your possession" can be anonymous cash in your pocket, e-money stored in a card such as a Japanese Suica, in your bank account and accessed through a network such as Visa, or credit offered to you by a company such as Visa. It does need to prevent you somehow from double-spending your money, which is more of a challenge when it is held as bits than as some sort of physical token such as a bill or coin.
Bitcoin runs at around ten million transactions a month. By design. Its performance is some seven decimal orders of magnitude lower than what we need. If we ran Bitcoin for the age of the Earth, we could cover about one day's worth of web browsing. And that's just the computation for the payments. Imagine how much storage is needed!
And Bitcoin is, by design, global. That transaction rate wouldn't be a problem if it represented only a single coffee shop -- nobody sells ten million cups of coffee a month. So systems that are working to improve scalability of Bitcoin or other blockchains are really trying to solve a problem that Bitcoin introduced. The approaches that I have seen, including ideas such as "lightning networks", seem to regain an order of magnitude or at most two. This still leaves us with a system a billion times slower than what we need. The only way to get there is to create true distributed systems. Maybe not so unlike the one we already have, despite all its flaws...

Education

I don't see much overlap between web3 goals and education.

Radical transparency, as in David Brin's The Transparent Society, is an interesting notion, staking out one extreme of the design space. But I don't believe that students should have their every move out there for everyone in the world to see. Students need room to make mistakes, to find themselves, learn and change and grow without worrying what a potential employer will think. Yes, I know they are oversharing on SNSes 24x7 and have a different view of the whole process than my generation does, but that doesn't mean every homework set, poorly thought out essay, dispute with a professor and bad grade needs to be recorded on a ledger for all the world for all posterity. FWIW, seeing their github activity seems like a positive thing when looking for skills and accomplishments, but even that needs to be put in the context of the whole human and their experience.
Joi goes on to talk about completely revamping education, especially in Japan, around self-driven, purpose-based learning and what amounts to micro-credentialing. I'm in favor of empowering learners, and there is a chance to rethink how we structure learning, but I don't see how this connects to web3 at all. (Moreover, micro-credentialing takes away from the curriculum structure, and I believe in the value of a structured body or knowledge both for engineers and as citizens. Much more on this topic some other time.)

DAOs and Human Organizations

If you needed any persuasion that blockchain doesn't really help the already disadvantaged, this might do it: although cryptocurrencies are touted as being a way to share wealth and empower impoverished people, today the distribution of wealth is more uneven in cryptocurrencies than in real-world, fiat currencies and stores of value such as real estate. 

Instability, Scams and Ponzi Schemes, Burning up the Earth, and Other Such Minor Issues

Surely there is nothing more to be said by this point; a lot of people have pointed out problems with both blockchain and NFTs. Most importantly (in line with #1 in the list in the next section), I think it is absolutely unconscionable the amount of energy expended every day for such a small number of transactions. Proponents have talked for years about shifting from proof of work to proof of stake, but a) it doesn't seem to be happening, and b) proof of stake appears to exacerbate some governance and consensus problems.

Arvind Narayanan had a nice thread on blockchain, quoting a blog posting by Bruce Schneier. Bruce and the others he links to cover the core arguments pretty well, so I am not going to reiterate them all here. Bruce is one of the signers of the letter to Congress urging regulation of crypto finance. Quoting just a couple of Arvind's tweets,

[B]lockchain has so far proven useless. Worse, it's proven a costly distraction to people and communities who are trying to solve real problems...I can't tell you how many times I've talked to energetic students with great ideas about what's wrong with our institutions who, in a sane world, would be working on fixing our institutions but instead have been seduced by the idea that you can replace them with smart contracts.

Let me address just a few of the issues that I think haven't gotten as much attention as they deserve, at least in the set of things I have been reading.

Getting Squeezed

Miners join a network voluntarily, with the idea that they provide a service that others will pay for. The single biggest problem in this libertarian paradise, from what I can tell, is that very aspect. With only a handful of miners worldwide, miners could make a comfortable living and the environmental impact would be low. But as long as there is profit to be made, new miners will join, driving up the collective mining rate and increasing competition, such that the probability of an individual miner receiving the payout for successful mining goes down. Thinner margins will mean that only those who can efficiently run large-scale operations can afford to stay in business. As a result, there is a massive explosion in worldwide mining capability that today damages the environment and distorts the market for semiconductors, but that can't hold indefinitely. I figure that ultimately there will be a crash or consolidation of miners such that we end up with the Walmart of miners, with everyone else driven out of business. (We may be seeing this already.) Or, perhaps the better analogy is Subway franchisees.

Instability

"Pump and dump" and other scams abound, but an even bigger issue, if possible, is whether the system itself actually works and can be trusted to always work when we need it. As I write this in summer 2022, the last several months has seen a lot of instability in the cryptocurrency markets, even in the so-called "stable coins" that are supposed to be pegged to a currency such as the dollar, but in reality are still vulnerable to the fundamental issues of liquidity (one of the key concerns expressed by Paul Blustein) and whether or not someone actual wants to buy what you are holding at a time when you have little choice but to sell.

Lately I have been seeing ads for automated crypto trading accounts. Even in the highly regulated world of stock trading, algorithmic trading is potentially the most destabilizing technology introduced since the stock ticker itself. Lots of agencies oversee the large operators, and economists in every major bank, government and trading house must be scrutinizing the situation and looking for positive feedback loops that can cause markets to gyrate out of control. They have also instituted "circuit breakers" in case major problems develop. The financial crisis of 2008 may have shown that small investors can and will still lose their entire investment, but the existence of the big houses and the regulators reduces the number of scams and provides some (emphasis on "some") recourse when troubles occur. The cryptocurrency market has none of this.

Scrip and Taxes

A question: are DAO tokens just company scrip? I don't think so, but there are enough similarities to be disquieting. In the early 20th century U.S., with human mobility rising rapidly but not yet easy and the megalopolises not yet a majority of the population, many remote, small towns were essentially one-industry, even one-employer, towns. Employers often paid employees in scrip instead of U.S. dollars. Scrip could buy goods, at inflated prices, at the company store, or be exchanged for dollars at disadvantageous rates.

DAO tokens feel a little like scrip, a little like getting paid in stock. Tokens can be exchanged for goods, but only within a limited community. Because you aren't physically limited, you can shop anywhere that will take your tokens/scrip, but to participate in the broader economy you have to find someone who will change your tokens for money that works in your local economy.

Of course I understand that there are different kinds of tokens (some say as many as six), some of which are closer to currency and some of which are closer to voting stock. But if part of the design is to maximize liquidity and the velocity of the economy, won't even the stock-like tokens be traded rapidly and likely wind up concentrated in a few hands? I'm pretty unclear on how the dynamics of all of this is supposed to work out, how it is likely to work out, and what the failure modes are. But I'm pretty doubtful that DAO tokens and cryptocurrencies with little value behind them are ultimately stable.

Speaking of trading and the economy, if you are paid for work in some DAO's tokens, which you can trade for goods, when does it become taxable income? When it gets exchanged for local fiat currency? What if that is never?

Vision, Revisited

Okay, with all that under our collective belt, let's revisit the vision. I actually like the idea that our organizing principle should be community. Here's what that means to me, in terms of technical challenges:
  1. Climate change and sustainable development. Without solving these things there is no community, no human security, no meeting of basic human needs.
  2. Data and information systems security. As IT people, this has to be Job One. Add in general systems stability (CIA = confidentiality, integrity and availability), and this is far more important than some random, clever new feature.
  3. Establishing personal autonomy, privacy and empowerment. Note that, to liberal me, this does not imply leaving people out there on their own, with no support.
  4. DEI. (To the extent to which it's different from the above.)
  5. The next billion. Kilnam Chon has a talk with the title Future Internet for the Other Billions. Ever since its inception, the Internet has always faced the challenge that adding the next group of connected people has meant reaching groups that are less technologically literate and perhaps poorer, and with greater environmental and infrastructure challenges of all sorts.
  6. Technologically, the end of Moore's Law. Estimates of data center energy consumption range from 70 TWh/year to three times that, or around 1% of the global total electric power generation. Although our efficiency has increased dramatically, we, as an industry, still have work to do.
If you are working on one of those issues, kudos to you. If you aren't, but you do other work that you find rewarding, that's great, too. But if you're working on something that actively leads away from solutions to the above, or if you're fooling yourself that somehow the current proposed web3 technologies do any of the above, I would encourage you to rethink, carefully considering the big picture.

Final Thoughts and Notes

..wait, I've gotten all the way to the end here without mentioning "the metaverse", or virtual reality worlds. Maybe that means...it's really a separate thing?  FWIW, I was intrigued by VRML all the way back in the mid-1990s. It seemed like a good idea at the time. Whether the tech could keep up was another matter. I suppose, eventually, we will have Snow Crash-style virtual reality, but not yet. People -- including many who are unhappy with their real-world circumstances -- will probably like it. But other than a place to hang your virtual art you spent a lot of tokens acquiring, it's not clear to me that it's either necessary or sufficient for web3. (There have been lots of novels and short stories set in virtual worlds in the last four decades or so. One recent one I read is A Beautifully Foolish Endeavor, with a villian I swear is modeled on Jordan Peterson.)

Since I have invested two decades in working on a technical area or two whose real-world value has yet to be proven, I'm sensitive to criticisms of blockchain that it's a hammer in search of a thumb to hit. I have worked in large companies, in startups, and in academia in both Japan and the U.S., but people like Joi and many, many serial entrepreneurs have far more experience than I do, so you should probably trust them more than me on what works well and what doesn't. 

One friend of mine said, "It's so complex it's hard to see how stupid it is." This friend has spent far more time studying all of this than I have. Until recently, I had not invested much effort in studying web3, and I am sure it shows. I will endeavor to keep learning, and may update this posting. If so, I will add a change log at the bottom.

A final note: I recently had a conversation with one of our students, Shaimay Shah, who is due to graduate momentarily. He said (quoted with permission),

I think the web3 tech is my generation's way to maybe make a difference...I want to look back 20-30 years down the line and tell my kids that I made a difference to society.

That's the goal, and it's a good one. The question is, which of society's problems can be solved via web3, and what tools will take us there?

References

A few of the things that I read that are at least moderately helpful, some from true believers and some from skeptics, some linked to inline above but most not. Apologies for the kind of ragged formatting and lack of inline citations above. Even deeper apologies for not having them organized in any useful fashion, this list grew organically.
  1. Nick Weaver, "The Web3 Fraud", my favorite technical criticism on the technical costs of running a web3 site.
  2. Dave Farber and Dan Gillmor, Cryptocurrencies Remain a Gamble Best Avoided.
  3. https://www.msn.com/en-us/news/technology/bored-ape-yacht-club-the-nft-collection-that-s-becoming-a-real-offline-brand/ar-AAQT8yv
  4. https://www.vice.com/amp/en/article/y3v3ny/all-my-apes-gone-nft-theft-victims-beg-for-centralized-saviors
  5. https://moxie.org/2022/01/07/web3-first-impressions.html
  6. NFT Mona Lisa
  7. https://internetcomputer.org/
  8. https://dfinity.org/
  9. https://www.technologyreview.com/2020/07/01/1004725/redesign-internet-apps-no-one-controls-data-privacy-innovation-cloud/
  10. https://www.stephendiehl.com/blog/against-crypto.html
  11. NFTs being stolen:
  12. https://twitter.com/arvalis/status/1468814628276695041
  13. https://docseuss.medium.com/look-what-you-made-me-do-a-lot-of-people-have-asked-me-to-make-nft-games-and-i-wont-because-i-m-29c7cfdbbb79
  14. Inequality in crypto:
  15. Ashish Rajendra Sai, Jim Buckley, and Andrew Le Gear, "Characterizing Wealth Inequality in Cryptocurrencies".
  16. Khristopher Brooks, "Bitcoin has its own 1% who control outsized share of wealth".
  17.  David DSHR Rosenthal, "Can We Mitigate Cryptocurrencies' Externalities?"
  18. https://twitter.com/doctorow/status/1493288001107021826
  19. Irving’s blog posts: 
  20. https://blog.irvingwb.com/blog/2022/04/what-is-web3.html
  21. https://blog.irvingwb.com/blog/2021/12/the-metaverse-the-next-major-phase-of-the-internet.html
  22. https://www.mollywhite.net/annotations/latecomers-guide-to-crypto
  23. Links from Mike: (1) Useful background reading about the future of the metaverse: https://www.eurasiagroup.net/live-post/the-geopolitics-of-the-metaverse   (2) especially the first and last sections. https://moxie.org/2022/01/07/web3-first-impressions.html
  24. https://conversationalist.org/2020/03/05/the-prodigal-techbro/
  25. https://decrypt.co/100687/ethereum-creator-vitalik-buterin-contradictions-web3-values
  26. https://twitter.com/VitalikButerin/status/1526378787855736832
  27. Joi Ito 伊藤穰一 on web3 town Yamakoshimura 旧山古志村
  28. Another article on web3 town Shiwa-cho, a town of 33,000 people in Iwate-ken.
  29. https://ssir.org/articles/entry/the_good_web#
  30. https://ethereum.org/en/developers/docs/web2-vs-web3/ 
  31. https://inrupt.com/solid/p frustratingly short on details, but Tim Berners-Lee's vision sounds pretty good.
  32. Ethereum consortium, Web2 v. Web3. From Ethereum's point of view, of course, but pretty level headed. What's missing is whether Ethereum is either necessary or  sufficient, or even a way, to achieve the goals laid out.
  33. https://www.zenbusiness.com/how-to-start-a-dao/
  34. Cointelegraph: Deconstructing sidechains — The future of Web3 scalability.
  35. https://cointelegraph.com/news/deconstructing-sidechains-the-future-of-web3-scalability
  36. https://asia.nikkei.com/Opinion/Bitcoin-will-be-remembered-as-a-historically-insignificant-fallacy
  37. https://doctorow.medium.com/moneylike-d20f8279a72e
  38. https://blog.makerdao.com/the-different-types-of-cryptocurrency-tokens-explained/
  39. https://www.fsa.go.jp/en/policy/bgin/ResearchPaper_qunie_en.pdf
  40. https://www.nasdaq.com/articles/what-is-ethereum-name-service-and-how-do-you-get-a-.eth-web-3.0-domain
  41. https://www.researchgate.net/publication/260438995_Trends_in_worldwide_ICT_electricity_consumption_from_2007_to_2012
  42. https://www.nature.com/articles/d41586-018-06610-y
  43. Morgan Ames, "Laptops Alone Can't Bridge the Divide," a must-read on the failure of technology alone to solve problems in education.
  44. Folding Ideas, "Line Goes Up -- The Problem with NFTs", a two-hour, fast-paced dissection of NFTs and blockchain both. Mostly clear and mostly calm, but occasionally heavy on the jargon and slips into occasional fits of outrage. Surprisingly watchable, despite the length and topic.
  45. https://nymag.com/intelligencer/article/three-arrows-capital-kyle-davies-su-zhu-crash.html
  46. https://web3isgoinggreat.com/
  47. https://www.nature.com/articles/s41598-022-18686-8

Change History


Posted by at No comments:
Subscribe to: Posts (Atom)