Moldbug did not simply "mention the existence of slavery."
He explicitly celebrated it. He talked about it being a natural order. He argued that "the innate character and intelligence of some is more suited to mastery than slavery."
This is pure, racist drivel. This is an evil way of thinking. There is a reason that racism, genocide, slavery are so abhorred in the modern world; they are poisonous ideas that have caused some of the largest atrocities history has known.
He is a fascist, arguing for a violent dictatorship, and arguing that slavery is a natural order. These are not mere rhetorical devices; this is the core of his argument. His entire argument is steeped in violence.
> Thus we may charge Carlyle not only with Hitler, but also with Stalin. Both dictators! You see what these dictators do.
> Of course, we are on equally safe ground in noting that both Hitler's party and Stalin's originated as political parties, and reading both genocidal maniacs as accidents of democracy.
I read Moldburg and I see an intelligent guy exposing the hypocrisies of politics left and right in an erudite, reasoned manner. I read yours and I see outright dumb slurs and lies.
> So I guess, there's still a lot of work to be done.
And I think this is the most interesting part.
One of the most depressing things about all of the "this image recognition algorithm performs better than humans on this task" is the idea that we've pretty much solved the problem, and it's just a matter of some more optimization and tweaking to handle a few edge cases.
This kind of problem, where the dominant solution simply gets it so wrong, and the problem cases are uncommon enough that any statistical solution is generally going to treat them as noise, reveals that in fact that there is likely plenty of room for entirely new, novel ways of approaching the problem to handle these kinds of cases better.
It's actually more exciting that there's so much more to be done, than to say "well, it's basically a solved problem, we just need to do some tweaking and optimization."
Definitely agree here, want to add a link to this paper which shows how far we still have to go (and questions whether the current models will ever replicate human vision): http://arxiv.org/abs/1412.1897
This reminds me of one of Richard Feynman's famous quotes: “We are trying to prove ourselves wrong as quickly as possible, because only in that way can we find progress.”
Indeed, discovering these "broken" edge cases is exactly what we need to converge upon a more correct solution.
IRC is used for quick casual conversation. The internals forum is used for discussion, "pre-RFC", and the like. The rfcs repo is use for actually discussing formal RFCs, as well as wishlist language and library bugs. The rust repo is for the actual implementation of rustc and the standard library itself.
Unicode is not a clusterfuck. Overall, it is a very comprehensive, well thought out standard, that has improved the situation for interoperable internationalization dramatically over the hundreds of separate encodings that preceded it.
The clusterfuck is mostly in the essential complexity of the problem; the world's languages and writing systems are quite complex and varied, and all of the writing systems and punctuation conventions were designed for reading, hand-writing and hand-typesetting by people who understood the language in question, not automatic typesetting and processing by a general purpose computer.
The complexity of the problem was increased by the necessity of providing migration paths from legacy encodings to Unicode and back again; without such a guarantee, bootstrapping the world into using Unicode would have been a much more difficult proposition, but that constraint also means that many oddities of legacy encodings have had to be preserved in Unicode in order to be able to preserve that round-trip mapping.
Unicode, and sister projects like the Common Locale Data Repository, are doing an admirable job of navigating and standardizing this complex problem.
There are definitely aspects of the Unicode process where they have gotten it wrong; UCS-2/UTF-16 is one of them, in hindsight it is apparent that UTF-8 is superior in pretty much every way. Having a variable width encoding but which most of the world's text fits in a fixed width, and which has endianness issues that necessitate inclusion of an non-textual byte order marker to disambiguate, has caused a number of problems and incompatibilities. There may be a few other points of legitimate criticism, like some aspects of Han unification.
But on the whole, outside of a few problems like those, the "clusterfuck" is caused not by Unicode, but simply by the essential complexity of the problem involved. Language and text are simply difficult things to model in a computer.
> The complexity of the problem was increased by the necessity of providing migration paths from legacy encodings to Unicode and back again; without such a guarantee, bootstrapping the world into using Unicode would have been a much more difficult proposition, but that constraint also means that many oddities of legacy encodings have had to be preserved in Unicode in order to be able to preserve that round-trip mapping.
While I agree that preserving round-trip integrity was essential for Unicodes success, I'm not sure if the approach taken to achieve that was the best one. I would have preferred that the complexity tradeoff would have been shifted to software converting between Unicode and legacy encodings by having more complex mapping tables and cleaner code point space.
I also think that Unicode Consortium should have been more aggressive in segregating (and discouraging the general use of) legacy compatibility features/codepoints and the stuff that is actually supposed to be used. My personal pet peeve is precomposed characters.
In a more general note, I sometimes wonder if it would have been beneficial to have separate layers in Unicode and have more focused on providing generic primitives. As a simple example it is mighty convenient that I can type 2³ = 8 in plain text, but arguably it would be even nicer if instead of special 'SUPERSCRIPT THREE' codepoint there would be generic superscript modifier codepoint that could be combined with any character.
Speaking of superscripts, they demonstrate well one aspect that I dislike in Unicode, the way they have absorbed legacy encodings verbatim. The numeric superscripts (e.g. ⁰ ¹ ² ³ ⁴ ⁵ ⁶) happen to have inconsistent look on my machine because the superscripts for 1, 2, and 3 are from Latin1 while the rest are in their own block.
I think there are a few things wrong with this argument. I'm unconvinced by the argument that this should actually be considered a modifier letter.
Consider any English word with an apostrophe, e.g.
“don’t”. The word “don’t” is a single word. It is not the
word “don” juxtaposed against the word “t”. The
apostrophe is part of the word, which, in Unicode-speak,
means it’s a modifier letter, not a punctuation mark,
regardless of what colloquial English calls it.
Modifier letters, in the sense used in the Unicode
Standard, are letters or symbols that are typically
written adjacent to other letters and which modify their
usage in some way. They are not formally combining marks
(gc=Mn or gc=Mc) and do not graphically combine with the
base letter that they modify. They are base characters in
their own right. The sense in which they modify other
letters is more a matter of their semantics in usage; they
often tend to function as if they were diacritics, in
dicating a change in pronunciation of a letter, or
otherwise distinguishing a letter’s use. Typically this
diacritic modification applies to the character preceding
the modifier letter, but modifier letters may sometimes
modify a following character. Occasionally a modifier
letter may simply stand alone representing its own sound.
Punctuation marks are standardized marks or signs used to
clarify the meaning and separate structural units of text.
Based on these definitions, the apostrophe seen in contractions and possessives is definitely punctuation, not a modifier letter. Modifier letters indicate some effect on sound or pronunciation, either modifying an adjacent letter or having a sound on their own. U+02BC (MODIFIER LETTER APOSTROPHE) is such an example, being used to indicate a glottal stop.
Apostrophes used in contractions and possessives, however, have no effect on pronunciation; instead, just as in the definition of punctuation, they are used to "clarify the meaning and separate structural units of text."
But we shouldn’t be perpetuating this problem. When a
programmer is writing a regex that can match text in
Chinese, Arabic, or any other human language supported by
Unicode, they shouldn’t have to add an exception for
English.
Thinking that it's possible to do text processing in a language or writing system neutral way is a fallacy. Unicode simply provides an encoding that allows all of these writing systems in a single document, plus a number of algorithms that are designed to be fairly reasonable across the entire encoding, but which cannot be correct for all languages and writing systems without specific tailoring.
Many writing systems do not use spaces between words. Any form of word segmentation for these writing systems will necessarily be language specific, generally involving dictionaries. Using a regex like \w+ on Chinese or Thai text is fairly meaningless, as it will generally match an entire sentence at a time, rather than actually matching a single word.
For godsake, apostrophes are not closing quotation marks!
No, they are not. However, they also aren't modifier letters. If you wanted to provide a distinction for the purposes mentioned here, you would probably need to add a new, distinct punctuation character "curly apostrophe" or something of the sort (since the ASCII range apostrophe can't be reused due to its overloaded meaning). However, even if you did that, you would still need to deal with all of the legacy documents which use ASCII apostrophe and closing quotation marks; you wouldn't actually be able to simplify the implementation by making the assumptions that a closing quotation mark was always actually closing a quotation.
Now having three different characters that looked identical (the modifier letter apostrophe, the closing quotation mark, and the punctuation apostrophe) would additionally add to confusion.
Even if you didn't introduce a new character, and instead used the modifier letter apostrophe as a punctuation apostrophe, you would still have all of the problems with legacy documents; it would take years for this change to make it's way through all of the various word processing programs and text editors, even after it had there would be existing documents using the old conventions, etc.
In short, text processing is hard, because text conventions were designed for human readers who know the language, not computers trying to process text in a language-independent way, and they were designed either through handwriting or manual typesetting, not keyboard entry. You are never going to achieve a perfect text processing model that can handle all of the world's languages simply by using particular global Unicode properties of characters and applying a simple algorithm or categorization on them. A lot of text processing will need to be contextual, language (and locale) specific, and involve dictionaries.
I don't think that switching from the punctuation closing quote character to the modifier letter apostrophe for the punctuation apostrophe is likely to help much; and the confusion caused by nearly 20 years of documents that follow the existing conventions and so having to support both conventions is likely to make the situation much worse, not better.
I'm pretty sure that's intentional. The point of the license is for releasing crappy code that is still necessary for the research in question and validating the findings.
He never says any of these problems are fundamentally insurmountable. Just that you need to take any meta-data with a large grain of salt. Which, well, is exactly what Google does. It uses certain metadata, but doesn't trust it absolutely.
Also, he mentions Google in this essay; this was written after the emergence of Google, when Google was trouncing all of these other services that tried to use metadata embedded in webpages or tried to classify all websites into some big hierarchy.
Nah, his argument goes beyond just the meta data itself. It's about meta data and whether or not we can use it.
A world of exhaustive, reliable metadata would be a utopia. It's also a pipe-dream, founded on self-delusion, nerd hubris and hysterically inflated market opportunities.
The comment at the end of the essay about Google looks to be half throw-away. It's like he showed the essay to someone before he published and they said, "but Google", so he put a comment about Google in there without acknowledging that they were wrecking his essay.
Additionally, he never mentions the positive feedback loop created by companies like Google that tend to encourage good meta data creation by punishing the obvious crap.
CSS/HTML are not meaningfully semantic; they encode presentational attributes of documents and applications. Syntax validation is not semantic; it's syntactic. DOM is not either; it's an API for manipulating HTML, generally used for presentational purposes. RDF is designed for semantics, but no one of interest uses it. HTTP is a transport protocol; it's a way to take an opaque URL plus some persistent state and get a blob of text or binary data back, along with just the barest of semantics about which requests are intended to make persistent changes server side. MathML is this weird semantic/presentational hybrid, of which I have only ever seen the presentational side used; I am aware of no real-world systems that produce or consume the semantic subset of MathML. PNG is an image encoding format.
So out of all of that, RDF and MathML are the only things that even attempt to define any kind of semantic encoding, and as far as I can tell they completely fail at that in the real world. For the vast majority of real world services, if you want to extract semantic data out of them you use JSON.
For the vast majority of real world services, if you want to extract semantic data out of them you use JSON.
JSON is just the encoding, not the data model; it doesn't work at the same level. That's why RDF/JSON and JSON-LD exist.
Without disputing the idea that the implementation of the Semantic Web is a failure, this idea that JSON (or XML, for that matter) is a data format is holding us back. Every provider is saying that they output in a "standard" format (JSON), while we keep writing new clients for each and every service out there, because every data format is different and it's impossible to write common handlers.
Except for a few attributes, a Github user and a HN user could have the same structure, since conceptually they're quite similar concepts. Yet, try handling [1] and [2] with the same code.
The regular claim that programmers can handle more abstraction than regular people just sounds like a sad joke whenever I think about the shortsightedness and lack of self-awareness that has led to the current situation.
Github users are tied to social interaction, job hiring, integration with multiple version control systems, and a pile of other crap, including systems not part of the website itself. Hacker News users are pretty much just tied to posts and comments on the site.
The only piece of data that seems in common is that they both have user names.
There is no reason user data should be similar across different systems. Products are generally unique, or they don't have much value. That means the data tied to a user is often also unique.
Each site has a set of attributes, which partially overlaps. The data formats should reflect this, allowing for software clients that know the base attributes to interact with both, while still having their non-overlapping attributes. Why can't I add both profiles to my PIM software without special adapters?
This is what RDF allows you to do: instead of designing a custom format, or being tied to a rigid standard one, you can simply add standard attributes to the element in an ad-hoc fashion, so the client can use whatever attributes it supports while allowing for the ones it doesn't. And if there is no standard attribute, fine, you can add custom ones as well.
There's no reason the same code shouldn't be able to process HN, Slashdot and Reddit threads, for example - and even a Github issue.
> I have to admit I'm surprised to find that anyone is still trying to make desktop Linux work for normal people, i.e. non-programmers who aren't free-software fanatics.
As far as day to day use goes, modern desktop Linux environments are just as usable by end users as Windows.
The main impediments to adoption tend to be:
1. Software that they need that only runs on Windows or OS X
2. Syncing to devices like iPods or iPhones
3. Lack of easily accessible support infrastructure for when things go wrong, major version upgrades, etc. You can generally take a Windows machine to Best Buy or some other local computer store, or a Mac to an Apple Store, and they can deal with it, but there's little accessible support for the complicated cases on Linux unless you have a friend or relative who's into Linux.
4. Lack of support by IT departments.
So if you have a situation like in the original story, where someone is willing to provide that support voluntarily, and you just need a home computer with internet access and no specialized software, a modern Linux desktop will do just fine.
> My guess is that GNOME 3 and Unity require roughly as much CPU power and RAM as Windows 7.
Lightweight desktops like LXDE and XFCE can run fine on older hardware, and are much lighter weight than Windows 7. If you read the article, you would see that it was running LXDE, not Gnome or Unity.
I have found NetworkManager to be no more difficult to use than setting up networking in Windows. For me, for simple cases, it tends to just work; for more complicated cases with multiple interfaces, it takes some fiddling but can be made to work as well; and pretty similar to the amount of fiddling on Windows. I've found OS X to require less fiddling for some of the simpler and moderately complex cases, but also have problems of its own when you get to some more complicated cases.
It always "just works" at first. The problems start creeping in after a few weeks or months, when NetworkManager inexplicably vomits all over itself with dropped connections, unexplained failures to connect to wireless networks, and duplicate wireless entries (among many, many other things) when the underlying components (namely, wpa_supplicant) work perfectly fine on their own on the same hardware.
Are you trying to imply every OS doesn't face issues like these? A simple google search of "wifi broken after upgrade [insert OS]" will get you thousands of results.
I'd say I had worse wifi issues with Windows than Linux.
This isn't a "oh no, I upgraded and it broke". This is a case of "oh no, it simply stopped working properly without any perceptible cause".
Yes, I'm fully aware how broken Windows' wifi implementation is. NetworkManager, however, tends to make the Windows equivalent look wonderful in comparison.
And yes, every operating system has problems. That's not my point. My point is that we (the free-software-loving community) can do better than the current state of NetworkManager. It's the one pain point that seems to be pervasive across GNU/Linux distributions, and pretending that "oh, every operating system's wireless stack is broken, so we shouldn't be worried" is delusional. Even if NetworkManager is "just as bad" as the others, why shouldn't we be striving to make something better?
Well I've never had any of the problems you're experiencing on Linux in the past 3 years. Only issue I've had was a driver not being included by default. I've had issues on OS X and Windows though, and they put billions of dollars into their operating systems. So if they can't get it right, then maybe the problems go beyond the code and into the actual hardware?
Maybe you're just lucky. Or perhaps I'm just unlucky.
Whatever the case, I know it's not the hardware; wpa_supplicant (the underlying software that allows NetworkManager to function) has (in my experience) generally worked much better than NetworkManager, and OpenBSD's networking stack is also pretty flawless in my experience. So it's something about NetworkManager specifically that needs to be fixed or replaced.
And note that my criticisms about networking are not about Linux as a whole; on the contrary. My criticisms are strictly about NetworkManager. You probably understand this already, but your comment seems to be phrased in defense of GNU/Linux as a whole as if I consider the failings of NetworkManager I've experienced to be the fault of the kernel or somesuch - which is very much not the case - so I figure it's worth clarifying just in case.
I fear that at least one of the big DEs have contracted a severe case of developer paternalism.
In essence the developers involved are attempting to encode every use case they can think of into the program logic, and anything outside of that gets met with a "why would you want to do that?!".
I've been putting off learning it for years and years and years. Finally decided if I'm going to keep doing this as a career I'd better figure out how to use the new tools.
Have a new Cent7 install on a laptop at home.
I spent over an hour trying to get the wifi to simply connect to the network. Much of this was me trying to learn how to use the networkmanager syntax which is neigh-on-impossible to decipher. The help is useless, and apparently the syntax made a huge change from 0.9 to 1.0, and not knowing either means that 90% of the hits I found on Google were utterly and completely wrong and used deprecated flags.
I finally broke down and tried to use nmtui to solve my nmcli woes, but it only made things worse.
The installer was nice enough to create an ifcfg-<myssid>, but when I try to ifup that interface, the system complains there is no DEVICE for it. So I added a DEVICE= line corresponding to my wireless adapter (which nmcli sees), but then the system complains that my wireless device (WHICH NMCLI SEES!) does not exist.
I believe I have traced the problem to a bug report I found on RedHat's bug tracker -- I needed NetworkManager-Wireless installed in order for it to use my wireless adapter. Why lspci sees my adapter, the OS names it, the firmware is installed and up to date, and nmcli even sees the adapter as a wifi adapter, but then can't use it, is beyond me. nmcli wireless scans constantly failed (which is what led me to the bug report).. the bug report says it's fixed and provides better guidance that you need NetworkManager-Wireless installed as of <some version> but that's probably wrong, since I have the latest Cent7.1 (unless the bug report is REALLY recent).
Regardless, after adding the EPEL repo and installing the wireless-utils package which provides the iwlist, I was able to scan for wireless networks.
NetworkManager still was unable to scan, so I did a network restart hoping I could solve the issue without rebooting the whole box. Sadly, I lost my wired connection to the system at that point, so I'll have to wait until I get home to play with it some more.
Oh, and by the way -- even though my wired i/f is set to "ONBOOT=true", it did not, in fact, come up on boot because the cable was disconnected. After I connected the cable it took me 5+ mins to realize I had to restart the network service to get the connection to function, which is not the way it works pre-NetworkManager.
Suffice it to say, the software is massively, massively, massively, massively sub-par, and far less user-friendly than the good 'ol "editing text files" way. I suspect I am soon to have many more similar headaches as I delve into the world of systemd head first (another thing I've been avoiding).
Fortunately, we get to do really exciting things now that NetworkManager has taken over. I can CD to things, and set properties and methods, and all kinds of shit that makes me feel like I'm using Fujitsu's deplorable ALOM:
It's a buggy pile of shit, for starters. Constantly drops network connections, will frequently flood your network history with multiple duplicate entries for no apparent reason, will often mysteriously fail to connect to a wireless network for whatever reason even though others work fine (and other devices can connect to the wireless network in question without issue - even other NetworkManager-using ones)... that sort of thing. I know the underlying utilities (namely, wpa_supplicant) aren't the issue, since wpa_supplicant - in my experience - is much more robust when configured manually, with far fewer problems, but at the expense of having to edit wpa_supplicant.conf in order to manage wireless networks, which is a bit of a pain (though perhaps worth not having to deal with NetworkManager).
If someone figures out a way to port Android's network management system (I'm not sure exactly what it is off the top of my head, but I know it uses wpa_supplicant and that it's not NetworkManager) to general-purpose GNU/Linux distros, that would be a godsend.
OK I must be really lucky. I find NetworkManager just works. It does ask for my admin password when it doesn't need it but I find that if I just ignore that prompt then everything is fine.
The more common issue with passwords is that it incessantly prompts for the user's password or for the wireless network passphrase due to some strange inability to talk to whatever keyring daemon is in use (and thus an inability to retrieve network passwords). I usually work around that by disabling the keyring (either globally or simply preventing NetworkManager from using it).
It tries to bite off a lot more than it can chew, and fails horribly for anything that isn't a simple wifi config.
1. Makes fixed wireless connections less reliable. I plugged an LTE modem into a server for a backup link, and network mangler set it as a default route and munged up resolv.conf. My mistake for not removing it when I installed the machine, but that's hardly an endorsement.
2. Makes complex config out of reach. On my laptop, I'd like wlan0/eth0 to be bridged by default so that I can easily switch existing connections to a wired connection by plugging in. But I'm not going to fight network mangler just to get this feature.
3. Furthermore lots of times I'll want to setup an ad-hoc config on the wired interface (say, configuring a new router). For which it's easier to kill -9 the mangler and setup manually, especially if I want to keep my Internet connection.
4. GUI is the preferred way to config/interact with it, which is completely un-unix. It's one of the few things I have to do manually when reinstalling a machine.
I played with Arch a while back, and wicd felt much simpler and more straightforward. Network Mangler feels like a hack someone came up with to setup a simple wifi connection, which then got adopted into a bigger role than it can handle.
At least as of version 12.04, Ubuntu [1] automagically detects my wireless HP Laserjet printer (albeit with some slightly reduced functionality, IIRC). My wife's Windows 7 laptop struggles with it; the install process was more involved and the printer sometimes refuses to wake up when sending a print job to it.
[1] The modern desktop-oriented distro I'm more familiar with. I'm sure other user-friendly distros handle it similarly well.
At the very most, it might involve having to navigate the CUPS web interface (which, admittedly, is a pile of shit; thanks, Apple) in order to get a printer setup. Still rarely involves having to install drivers (though some vendors are better than others; Brother and HP tend to be the most Linux-compatible, while Canon and Konica Minolta have tended to require driver installation on some distros).
What installation? At the most, you'll be asked to install packages from HP and you click okay and you're done. On Windows, it's still likely that you'll need to download drivers from a website.
He explicitly celebrated it. He talked about it being a natural order. He argued that "the innate character and intelligence of some is more suited to mastery than slavery."
This is pure, racist drivel. This is an evil way of thinking. There is a reason that racism, genocide, slavery are so abhorred in the modern world; they are poisonous ideas that have caused some of the largest atrocities history has known.
He also explicitly argues for a dictatorship beholden only to landowners: http://unqualified-reservations.blogspot.co.uk/2009/07/seces.... He admits that his influences and design are similar to those of the Nazism and Stalinism.
He is a fascist, arguing for a violent dictatorship, and arguing that slavery is a natural order. These are not mere rhetorical devices; this is the core of his argument. His entire argument is steeped in violence.