Maybe he told the police that and maybe he told the public that -- he DID NOT testify this at his contempt hearing.
4,724 karma · joined January 19, 2012
I am @crdrost on Twitter and Facebook. I almost never check my Facebook.
I have an M.Sc. in Applied Physics from the TU Delft in the Netherlands. I am from Ithaca, NY originally.
You should listen to more Troubled Hubble.
Don't you forget the, the image in your retina
To foresee the future, and find a use for
Them telling me, everything,
that would happen to me anyway.
If it isn't science, it doesn't exist;
We make our own problems and blame them on
Birds and bees, but we ain't seen nothin' yet:
and don't you ever forget it.
Every night I wake up,
Thinking there's something missing in my heart:
a canvas for the art.
Every day I hear the sound of whirring fans,
But nothing more:
another scene yet to be scored.
You're the empress, I'm the king:
together we will rule this world
You'll paint the scene, I'll write the melody:
we will rule this world.
And of course you can't become,
If you only say what you would have done:
So I missed a million miles of fun.
(Of course if he was lying about the failure of his memory, that assertion might have gotten him into more hot water -- it would potentially be a perjury charge. But we should not say definitely "he was lying when he said that to police" but just "he didn't make that case for himself at court," if I'm reading these court documents right.)
Just to give the relevant excerpts, the supplemental order that the judge issued to explain his reason for the contempt order states that he's using a well-established legal framework for figuring out memory-failure type contempt cases, quoting one of those cases directly as:
> A civil contempt proceeding on a witness' asserted memory loss requires a three-step analysis that shifts the burden of production to the witness, but always leaves the burden of proof with the government. First, the government must make a prima facie[1] showing of contempt; i.e., that it made an authorized request for information, that the information was relevant to the proceedings, that the information was not already in the possession of the government, and that the witness did not comply. Second, once the government has presented its prima facie case, the witness must provide some explanation on the record for his failure to comply. If the witness fails to meet this "burden of producing evidence," the government's prima facie case is sufficient to meet its burden of proof for a finding of contempt. The witness may meet his burden, however, where, as here, he testifies that he does not remember the events in question. Finally, if the witness meets his burden of production by claiming a loss of memory, the government must carry its burden of proof for a finding of contempt by demonstrating that the witness in fact did remember the events in question, thereby establishing a willful failure to comply.
Note that testifying "I don't remember the password" is enough to activate part two. The supplemental order then goes on to apply this in the only obvious way,
> [first, ] the Government's prima facie case of contempt was largely, if not entirely, uncontested ... [second, ] at the September 30th Contempt Hearing, Mr. Rawls did not testify or call any witnesses, he did not offer any documentary or physical evidence into the record, and the only evidence he elicited was in the form of cross-examination of the Government's witnesses. Crucially, Mr. Rawls offered no on-the-record explanation for his present failure to comply with the August 27th Order... [therefore, ] in light of the Government satisfying its burden of proof, the Court adjudged Mr. Rawls to be in civil contempt of court.
[1] prima facie is a phrase which one might not be familiar with unless one has a philosophy or legal background; it means "at first glance" -- i.e. the government's showing might at a deeper stage be questioned but it has to "look like" contempt in order to start the process.
Vacuum is only one of the best insulators around if you're near other warm things, I'm afraid. Out in space the dominant form of heat exchange is the Stefan-Boltzmann law, which says that if you have a half-acre of total surface area on your moon-base, and it's at 300 kelvin, then you need to supply about a megawatt to keep it warm.
It's not an insurmountable problem, of course -- Tesla just announced a battery for Australia at 129 MW-hrs, which would sustain the thing for 5 days, which is not enough but it's only a half-order of magnitude off. (I'm also not considering something like nuclear heating; it's not uncommon to have a 1000-megawatt nuclear reactor in the US and that's just the power output, not the heat -- so it's not a terribly bad way to go if you can shield the rest of the base from it. The only hazard with that is one recently discussed by John Oliver on his show, firing nuclear things into space is heavily, heavily complicated by the nowhere-near-close-enough-to-zero failure rate of launch rockets and the insanely-scary-cost of distributing a bunch of reactor-grade nuclear fuel into the atmosphere.)
In fact our best understanding of why Mercury has the 3:2 relationship appears to be (if I'm reading these papers right) "because of historical reasons." That is, now that it's in that state, that state is stable and unlikely to decay to the 2:1 or 1:1 state, but getting into that state is much less clear and potentially requires that in the past, the other planets of the Solar system had nudged Mercury into a more eccentric orbit than it has now. The claim is that the eccentricity of that orbit caused the 3:2 resonance, then in more recent history these perturbations from other planets averaged out more to make Mercury's orbit less eccentric, but it still holds on to the resonance from its past.
This mathematical effect is incredibly important, it means that sunlight, during the day, averages out to being half as strong as its maximum over the whole of the Earth's surface. The calculation isn't even particularly difficult: the surface area of a sphere is well-known to be 4πr², half of that or 2πr² is illuminated at any one time, but the actual irradiation that we receive is proportional to the cross-section area, which is just the area of the circle: πr². So if the Sun-directly-overhead light were to be illuminating that entire half of the world, we would get k·2πr² light for some k, but instead we only get k·πr² light for that same k, so it works out to be 1/2 when averaged over the whole surface of the Earth.
While I was at university a fellow student asked me to guess the coldest place in the Solar System. I guessed "the middle of the dark side of Mercury." I guessed this for a couple reasons: (1) I knew I needed a rock without an atmosphere since atmospheres sustain convection currents that transmit energy, and (2) I figured since Mercury is so close to the Sun it's probably tidally locked to the Sun and therefore this part probably has not seen a speck of sunlight in millions of years.
It turns out that general relativity makes point #2 wrong and my friend gleefully informed me that he was looking at an article (there were many, so let's take [1] as representative) suggesting that it might be in a crater on the South pole of the Moon. This has basically the same reasoning of (1) and (2) above, except substituting the shadow of a crater for the shadow of tidal locking. But Mercury is still in the running -- the only issue is that we might not be looking at the dark side of it, but rather, again, at its poles.
[1] https://www.space.com/7311-moon-craters-coldest-place-solar-...
If you're less clear on what this distinction is: HTML+CSS+JS transmitted over HTTP based on domains looked up via DNS -- this is the Web. Things like email and BitTorrent and peer-to-peer networks and older things like Usenet newsgroups have always existed on the Internet parallel to the Web, to say nothing of, say, SSH and ping and other developer-friendly utilities, or newcomers like BitCoin.
I've given a chunk of thought to how we might reboot the Web as a new service living on the Internet and I've come away with more questions than answers. What the Web has done extremely well is to convince a bunch of companies that they need to have a Web Server living on the thing and hosting their content. What it has done extremely poorly are these problems which maybe we don't even think about: link-rot, the fact that you can't throw a dime at a YouTube video you liked and trust that the content creator will get 8 cents and YouTube will get 2, the fact that nobody can directly advertise to you, "watch this video and we'll throw a dime at you!", the fact that you can't actually just fire up your browser and start publishing a feed of videos and articles that you like interspersed with a bunch of content you're writing -- you instead need to sign up with some social network or blogging platform and then play by their rules. The problem gets even harder when you think about systematically purging illegal content and discouraging spam and so forth -- indeed the key trait is to be able to say "I want to identify who is holding onto this content and go after them via non-technological enforcement measures" and such. One gets the impression that the Web is a very niche tradeoff in a vast configuration space and that there are many other possibilities -- but it's hard to see one that does something so fundamentally different to the Web that people will instantly be addicted, "THIS is what I want."
The issue is inconsistent standards. Committee members are held to a lower standard than occasional contributors. Some people don't like that inconsistency and are leaving that culture. That's fine, because Ayo and Node can still share patches. The issue is purely that people who work on Ayo won't have to deal with any of the committee members from Node who can't be bothered to abide by their own code of conduct.
Codes of conduct are only meaningful if they apply to core contributors with all the more force than occasional contributors. The chief accusation lobbied against a CoC is "oh you just use those to go after people you don't like," much like, say, the police sometimes use US law. The only security against this accusation is that the CoC is applied with more forgiveness at the lowest levels and more stringently to the people-at-the-top. It's not social justice; it's consistency.
Admittedly I think a bunch of HN would like the anti-SJW approach of "let's just burn the CoC and forget that it ever existed," and in that sense I suppose that it is "social-justice-ish", but these are orthogonal concerns. The one concern is that members of the Node.js CTC are beyond reproach for shitty things they do; the other concern is defining clearly what things are shitty.
Given that these people are excluded by the biased application of these rules, this forms a space where they can feel more welcome as those rules will be applied fairly within this space. This should encourage novel solutions and contributions to Ayo.js which of course right now Node.js is free to merge in -- but the people who work in Ayo.js won't have to deal with the perceived inconsistent standards of Node.js or the perceived-as-hostile personalities that can only make it in Node.js but not Ayo.js.
In general forks are not bad things and being 'forkophobic' is a sort of 'open-source smell'; see Bryan Cantrill's talks on forking illumos. I don't think anyone is ready to step up as a BDFL for Ayo.js just yet, but I don't think it's out-of-the-question either.
So let me give the physics backstory here. Both N-type and P-type doped semiconductors are conductors, and they're both electrically neutral -- but in one of them (N-type) we've added atoms which have "too many" electrons in their outer shell (compared to the semi) and those electrons form conduction states shared among all of the atoms; and in the other (P-type) we've added atoms which have "too few" electrons in their outer shell, and then these electron-holes form conduction states shared among all of the atoms.
We join these two at a PN-junction, and then something very "obvious" happens: the electrons fall into the holes. This is actually maybe not as straightforward as it sounds because we say that there is no "bias voltage" across these two but there is certainly an internal electric field, as the P-type side has more electrons than makes it neutral and the N-type side has fewer. What's actually happening is that what we call "voltage" is generally the "chemical potential" of the electrons which is normally dominated by electrical potential, but in this case we see a competing effect from diffusion, the excess electrons from the N-type dopants have diffused to the P-type dopants and thus both conduction bands evaporate in this "depletion region", where the diffusion tendency is balanced out by the electric field that points from the N-type side to the P-type side.
Now we put a "forward bias" on it, this means raising the P-type voltage relative to the N-type side, so we are putting more holes into the P-type side, more electrons into the N-type side. And they attract each other across this divide, but as they try to connect they get overwhelmed by the electric field between the two, and "roll back" to their side. This happens until the voltage bias is exactly large enough to cancel out the electric potential difference of the boundary: then electrons in the N-type side successfully can fly a little distance into the P-type side and holes from the P-type side can fly a little distance into the N-type side, allowing the local electrons to "fall into" the holes: and a net current can flow again! This voltage difference is actually a material property for the semiconductor in general; so for Silicon this "diode drop" is 0.7 volts; when forward-biased the diode appears to be something like an 0.7V battery wired against the current, when reverse-biased the diode appears to be a circuit break, and in the middle you might have some sort of interesting linear and nonlinear regions.
If you just connect two diodes together backwards, one or the other of them is going to be reverse biased unless you inject or extract current out of the region between them -- the only flows possible are "outside to center" and "center to outside" until you get to the breakdown regions of the diodes.
Okay now that we're clear on diodes let's talk about the NPN in a typical bipolar junction transistor. The key here is that the P region is actually very small. Usually we call one of the N regions (which is usually more heavily doped than the other) the "emitter" and the other the "collector". The usual recipe for the voltages in normal operation is,
Vb > Vc > Ve
Since Vb - Ve is a typical diode drop, we see that holes in the base have no problem with combining with electrons in the emitter and current flows freely between these two. On the other hand we see that the depletion region between the base and the collector has actually grown.
This is where the thinness of the base is really important. The depletion region isn't really a place where electrons can't move, it's just a place where electrons coming from the N side find themselves repelled by an electric field from excess electrons on the P side. But we just said, the electrons coming in from the emitter are no longer being repelled by the excess electrons in the base; they're penetrating into the P side. Well, once that happens, the electrons from the emitter might also get far enough that they feel these electrons "behind them" pushing them into the collector, where they are likewise attracted by the unbalanced positive nuclei on the other side. So it's kind of a rollercoaster ride where electrons with high enough energies can fly into the base and then find themselves dumped into the collector.
This causes an electron current from the emitter to the collector which tends to be proportional to the electron current from the emitter to the base -- and the proportionality constant is usually governed by how thin this interface is, since it requires the electron to diffuse far enough across the base for the residual forces to push it into the collector. For modern BJTs this proportionality constant can be in the hundreds or thousands: for each hole that makes its way into the emitter to recombine, hundreds or thousands of electrons could be coaxed past this point-of-no-return where they fall into the collector.
The mental model you want here is that there is this base-emitter diode which, if it's reverse-biased, you usually see nothing (unless you totally reverse the BJT and pretend the emitter is the collector and vice versa --- usually the aforementioned current gain is much smaller) -- and you have an ideal current source connected pointing from the collector to the emitter, and the magnitude of that current source is given by this proportionality factor with the comparatively weak current going from the base into the emitter.
Um, yes it is. A harassing response which does not address the substance of your points is clearly a derailing response. That you can't "clearly" see this says more about you than it does about her.
> And what does it even mean, getting others to speak for her?
Literally, it means asking a colleague to say the exact thing that she wishes to say, because she knows that if she says it he will try to ad hominem it. This is standard English.
I would also point out that some of her vocabulary, "Beyond ____, also ____", is also standard English and it forbids your egregiously malformed interpretation of her comments. That is, you seem to think that the "targeted harassment" is "his constant, unproductive criticism" when in fact the criticism is clearly, in context, the "what he has done to the health of the project" which she has just been talking about -- and she has saying that "beyond" that, he "also" has targeted her with harassing statements. That construction explicitly forbids your sophomoric armchair psychoanalysis which can't separate someone who happens to disagree with you on social policy issues as "she can't separate someone disagreeing with her on social policy issues with harassment." (I don't actually believe that you are this way -- I don't know you -- I just think that your comment reeks of it and I would observe in passing that paradoxically we all accuse ourselves of our own sins. I mean, I'm terribly guilty of the latter, at least.)
It does have an effect on the social culture that's touched by that CoC, and you might like or dislike that effect. You might also have been totally distanced from this so as someone who is an outsider who has met a bunch of these people, let me say that in my personal interactions with these folks, Node.js has been more friendly to LGBT folks, issues of accessibility for the hearing and sight impaired, people with mental health issues like depression, people of color, women, and nonbinary folks, than just about any other community I have ever met. Meanwhile they have never come across as any of those alt-right buzzwords -- as "social justice warriors" or "man-hating" or "white-shaming" or anything; I have never felt unwelcome personally as a white American man.
Someone on the TSC has been anti-(what-the-community-is-about), and is actively apologizing for CoC enforcement and saying mean things which is discouraging others from participating in Node.js, and the technical community had a vote to censor him. They voted not to, and now someone for whom those community values are important has quit the TSC because they didn't support this decision.
I feel like everybody's rushing to see everyone else's choices as invalid but these are all valid private decisions. That community is allowed to decide that it likes this culture that its CoCs create, Rod is allowed to decide that he does not like it, the TSC is allowed to vote that Rod's dislike is not enough to kick him off the committee, and Myles is allowed to decide that this TSC support of this guy who doesn't like this culture is enough for him to no longer want to be part of the TSC.
Now if the original post goes public, it makes sense that the details be blanked as "not fair to Rod", since the rest of the world's eyes are going to look on just these things and might not have full context.
You ask "Where did he say that?" and I'd be surprised if there's any one succinct place -- it's one of the two main topics of the whole memo, and the memo does not have a coherent topic sentence or even really a coherent argument, so I think it's likely absent.
But, like, Damore makes a case to the effect that "biological explanations can't be ruled out" and then reverses those weasel-words by suggesting that his biological explanations be used to guide policy by, say, encouraging pair programming which he supposes to be something that women are likely better at on average. This sort of move suggests that he thinks the biological effects that he's citing (see note [1]) are big enough to guide policy, which they're not. You don't need to take my word for it -- the main author of the article Damore is citing was asked to read Damore's memo and this is his take on it [2].
Of course the problem is even worse in that this article which Damore used to write his article is psychological; it is based on doing a personality test in a bunch of different nations. That makes it very hard to conclude anything biological about it, so every time that Damore mentions "biology" in his memo, that is an interpretation of his own devising. The original personality-study article also interprets its findings biologically but it is really tenuous [3]. In fact neuroscientists have also been studying the brain and they have not found a clear biological difference between male and female brains [4].
[1] He gives a summary of a Wikipedia summary of an article by Schmitt et al. (2008). The PDF is freely available by the university at http://www.bradley.edu/dotAsset/165918.pdf but the sample sizes were I believe later corrected as an erratum, so I am not sure which one this has.
[2] Schmitt, evaluating later research as well, summarizes by saying that sex differences are only "accounting for less than 10% of the variance" and that using this to guide policy is "like operating with an axe. Not precise enough to do much good, probably will cause a lot of harm," in an article at http://quillette.com/2017/08/07/google-memo-four-scientists-... .
[3] The argument in the article involves their surprise that the majority of the discovered effect apparently disappeared in Africa and East Asia. Their interpretation is literally that those cultures are so much less economically developed than we in the West are, that their women must feel so much less free to just be themselves, and therefore they act more like men as a sort of baseline survival tactic. Read the paper; it's a very 'you cannot possibly be saying what I think you're saying, can you?' type of experience.
[4] See the links in the article https://www.newscientist.com/article/dn28584-a-welcome-blow-... for a nice summary.
SICM is basically a mainline classical mechanics textbook of the form that I was able to do my Bachelor's in, but combined with some tantalizing snippets of Lisp code which make it seem like there's something deeply interesting there, too. There's a magical feeling where you're like "holy crap if I could nail down the mathematical notation a little more and get some of these routines as efficient library built-ins, this would be like a programming sandbox that could teach a student physics by direct interaction."
If you want that sort of story, one textbook that you might incredibly like would be Griffiths' Introduction to Elementary Particles, which has a very readable first section going into the history of how we came to have the Standard Model that we have today, some names of who discovered what, etc.
A. Zee's Quantum Field Theory in a Nutshell is also very nice for getting a sort of pleasant appreciation for quantum field theory if you've got some mathematical background.
It also depends on what you take for granted as a baseline. If you can find Feynman's New Zealand lectures, for example, you will notice that he deliberately avoids introducing explicit complex numbers or explicit integration, and still manages to convey what both of those mathematical formalisms allow the theory to do. (There are also some little gems. Like, if you pay attention to the part where he says something like "I wish I had brought an example of one of these surfaces where we've erased lines of the mirror so that I could show you" -- then you're in the right position to say, "holy crap, I understand the rainbows that I see in the bottom of CDs/DVDs now!" after a second.)
Sussman of Lisp fame went on to write The Structure and Interpretation of Classical Mechanics, https://mitpress.mit.edu/sites/default/files/titles/content/... .
I'd be remiss if I did not mention that the father of String Theory is on a quest to provide everyone with the education needed to appreciate the current theories of physics; see http://theoreticalminimum.com/ or just look up Susskind on YouTube; e.g. "Susskind Statistical Mechanics" turns up https://www.youtube.com/watch?v=D1RzvXDXyqA .
If you really want cutting-edge sometimes-somewhat-unbelievable stuff, the Perimeter Institute publishes their lectures-for-the-public online; you might for example really like Penrose's idea that maybe someday when all the black holes have evaporated, all of the particles become massless and they no longer experience time so we can just evolve the system to t=infinity after some finite time: so we discover that what we get as our boundary at infinity could be conformally rescaled to something resembling a t=0 Big Bang -- a "conformal cyclic cosmology". See http://pirsa.org/index.php?p=speaker&name=Roger_Penrose for more of that sort of stuff.
Getting it all running on Linux was doable but definitely not for the faint of heart. But once I did, Xournal was one of the most useful pieces of software for my physics Master's; if the lecturer provides PDFs of lecture slides in advance you can literally just take notes on the slides themselves; or just file them away in a neat folder structure.
Maybe the biggest difference from paper is you can quickly draw a rectangle around part of the preceding equation, and either drag it or copy-and-paste it.
Touch input was nice for browsing and certain games, but was never a huge part of my work.
Multiplication doesn't in general require an abacus if you're trained in roman numerals. Let's take your "more complicated numbers", 42 * 13.
Expand XLII * XIII = X * XLII + XLII + XLII + XLII, the first requires you to know a times table to see that it's CDXX, the rest you'd add mentally, LLL - XXX = CXX, so it's CDXX + CXXVI = DXLVI. Five hundred forty six.
Like, I'm not saying it's as clean as 42 * 13 = 420 + 126 = 546, it's not. But it's not educationally prohibitive either.
I got some of my CS courses from a man named Kevin Walsh who I think was a Ph.D. candidate at the time; I think he's now teaching at Holy Cross. If my memory serves his group solved (at least for the time being) the Sybil attack problem on the then-active distributed-filesharing-networks at the time, but it never really grew in adoption because the stakes were really low.
Their basic idea was (again if memory serves me right) to let people self-classify into clusters, so your esteem of someone else's ranking of websites would be based on your own ranking of those websites and how they match. In that way, you get a fully distributed web-of-trust system which nevertheless can't be Sybil-poisoned: to inject their own phishing sites, an attacker first needs to inject trustworthy reviews for a bunch of other unclassified sites, which violates the Sybil assumption that your reviewer identity is disposable or forgeable or whatever and demands that you contribute more to the network than you take away.
I'm not sure if this could be quasi-centralized for browser reviews of proposed short-names for domains, since it assumes a big peer-to-peer decentralized system with lots of people reviewing lots of things and lets them sort themselves into their own groups; it gets dodgy if you're retroactively like "okay so here's the main group now". There is a hint of what it would look like if you try to recentralize it and say "this is the consensus for that domain name's shortname" in the form of how blockchain networks look: there you have a similar story of "if you want to break our consensus that your site is a phishing site, you first need to join our network and outperform the rest of it." But yeah, I think that sorting out the data requires some sort of complicated thinking. I'm more dreaming that sorting out the interface and getting approximate correctness for 90% of traffic so that people get accustomed to that little green box being there, is not especially hard.
On the flip-side, this is already a problem in the form that browsers have to detect phishing somehow and there are Sybil attacks on that based on marking other sites as phishing to try to decrease peoples' confidence in these phishing indicators. So even though there is a technical problem here, I'm not sure I see it as novel to this particular approach.
It's well worth remembering that one-in-three is a really good batting average and one-in-four is MLB average. Actually that's why I like baseball as a sport in general; it's not usually as exciting as other sports but when you play it and you swing and the ump calls "STRIKE ONE," you have to swallow any disappointment and pick up that bat and get ready again: because that next pitch is coming, a new opportunity.
A more-SSH-like approach would be for browsers to identify PSL-level domains with simple site names, which can be done by-hand for the Alexa top 500 if nothing else; so "https://docs.google.com/whatever" maps to a green "Google docs> whatever" in your URL bar; click to reveal the full URL. The idea is that a local store for this mapping is distributed across all of the browsers out there, so by default you get a yellow arrow on `-> https://idontknowthisdomain.example/` and perhaps as HTTP goes south we get a red one on `-> http://thisdomainneither.example/`. But you can right click that yellow arrow and it asks "Do you trust the following domain?" with a "Name: ____" box to submit your own name. After that, this domain has been white-listed for you to have the given name, just as in HTTPS. Possibly you can even have the page itself suggest the name via a meta tag. The important point is that the user has to say that they trust this domain to have that common-name, before they see it pop up in green.
From there, the browsers can do what they already do best: phone home. If you get millions of different users suggesting the same official name for a web site, it's more reasonable to automatically add them so that more than just the instantaneous Alexa Top 500 is covered.
You still rely on the core certs to provide privacy, but people who visit a phishing site now see by default either a yellow or red arrow where they are accustomed to seeing a green one that they either set up themselves or someone else set up for them.
The most emotionally gripping anecdote for me, in short form: they got really cocky and promised the moon (a new operating system that would be a programmer's dream); then scopes crept, and deadlines slipped; as they tried to understand better why they were totally sucking, they realized that nobody had solved the actually-hard problem that they needed to and everyone just assumed it would be OK; even after significant optimization work they were two orders of magnitude slower than their competition. Clients had waited for years for this thing, bought hardware for it, and now they were being told "look, we're cancelling this project, we just can't do this well enough."
Hoare expresses his then-disbelief that he wasn't fired for the massive failure, especially given his then-belief that MIT was going to succeed doing its variant of the same thing (Multics -- insert notes of irony). Then,
> At last, there breezed into my office the most senior manager of all, a general manager of our parent company, Andrew St. Johnston. I was surprised that he had even heard of me. "You know what wen wrong?" he shouted -- he always shouted -- "You let your programmers do things which you yourself do not understand." I stared in astonishment. He was obviously out of touch with present day realities. How could one person ever understand the whole of a modern software product like the Elliott 503 Mark II software system?
> I realized later that he was absolutely right, he had diagnosed the cause of the problem, and he had planted the seed of its later solution.
After a grueling meeting of "what the @#$% just went wrong and how the @#$% do we fix it?" he describes the reorganization, and it's remarkably anticipatory of something like Agile. He establishes small teams who visit clients directly and figure out what they need; he discards clients' huge-project requests in favor of incremental improvements; those teams then have to convince him personally that this feature is important/useful and not too hard to implement; finally they work on the feature.
It's very much describing a management style which provides some sort of quality-control and pushback, but does not micromanage or mistrust. And it's got a sort of fractal structure: Hoare is only able to succeed because his own supervisors did not destroy him for his failure but trusted him to learn; his system is to act as a filter for the requests hitting his programmers, but he still trusts them to write awesome software if they have a strong sense of direction.