Replacing Crashplan

I’ve been pretty happy with Crashplan for five years or so, although lately not so much. Obviously, one reason to be unhappy with Crashplan is that they’re no longer supporting the family plan I was using to back up all our Macs. Worse, their quality of service seems to have (understandably) slipped as the plan I still have for nearly another ninety days has dropped.

Exhibit 1: my wife bought a new laptop. When she tried to restore files from her old laptop, they were MIA.

So, here we are at a nasty juncture where our supposedly bulletproof fire-and-forget backup system is (a) not working terribly well and (b) shortly going to not work at all.

Possible Crashplan Replacements

The obvious replacement is Carbonite. The problem with Carbonite is that it’s going either (a) be a lot more expensive than Crashplan was (because of our family plan — which is the equivalent of, effectively, n $100/year plans where n is the number of computers you back up; the small business plan works out as being stupidly expensive if you have more than 250GB of data) or (b) require me to do a bunch of work (i.e. set up one computer as the family server, have all the other devices back up to it, and then back it up via Carbonite. So Carbonite will either cost $300 (say) per month or $100 per month but require me to be my own network engineer. Oh, and forget mobile backups.

If I want to do a whole bunch of work I might as well just use Amazon S3. For 2TB of data that’s a mere $46/month. Ouch. I could probably use their cheaper long term storage, but now I’m basically starting my own Crashplan / Dropbox implementation and that sounds kind of like hard work. Forget that.

So, on to the “consumer” options:

Annual Cost$99 for 1TB$540 for unlimited10TB for $60$120 for 2TB$240 for 2TB$100 for 5 users 1TB each$120 for 2TB

Note that all these services support web browser access, so you can get at your files from any device with a web browser, but I’m talking about file-system integration where you can just save your file in the usual place and it’s seamlessly backed up to the cloud.

At first glance, Hubic looks like the standout value-for-money option. (Hubic is essentially like Dropbox, except it’s run by a gigantic hosting company). The problem is that I’ve found Hubic to have a poor user experience, especially with regard to performance. (My own experience is quite limited — and I’m kind of shocked that my blog post on Dropbox vs. Box vs. Hubic ranks quite high on Google searches for “Hubic Review”.) This may be a result of server location, or simply under-provisioning (10TB is so close to “unlimited” that I imagine it attracts a lot of abusive users).

At second glance, Microsoft OneDrive seems like a great deal at 5TB for $100/year. The problem is that it’s 5x 1TB per user, which is effectively 1TB. Still, a great deal if you’re happy with 1TB and want Microsoft Office for your family.

The shock comes at third glance — Apple’s product is competitively priced (cheaper than Dropbox!). It doesn’t support Android (surprise!) but aside from that it’s a great deal, you can share it with your family, it requires no real configuration and — this is the kicker — it’s smart about mirroring stuff to your devices. Want a 1TB photo library in the cloud? Great. Want it on every one of your family computers? Not so much.


Hubic is the clear winner in terms of price per unit storage. Apple is the clear winner in terms of functionality and comes second in price. When I take into account the fact that I pay Apple $3/month for a lower tier of storage just for convenient mobile backups it’s an even better deal (hooray for Opportunity Cost). After some consideration I’m thinking of doing both. Hubic for volume backup and iCloud for convenience.

I’ll let you know how it works out.

Ulysses and the subscription model

Ulysses review are evenly divided between five stars and one.
We wanted table support and the ability to create custom CSS files. We got a subscription model.


The life cycle of software — as exemplified by Photoshop — is:

  1. A new program with an unfamiliar UI but incredibly useful features attracts a devoted following.
  2. The devoted followers evangelize the product they love to more people. For a while.
  3. The developers stop adding new features that anyone cares about and add a subscription model when people stop buying new licenses.

Here’s what I want(ed) from Ulysses:

  • Tables.
  • Better ways of managing images. E.g. the ability to browse all images within a given “folder” and update, export, replace, rename, or locate them.
  • Simpler ways to create custom styles. E.g. the ability to make a document style-guide compliant for a given purpose (such as APA).

There’s other stuff but it’s much lower priority.

Instead what we get is a subscription model and no new features I care about. Indeed, the announcement on the Ulysses web site explains how the subscription model allows them to offer a free trial while the blog post on the subject tells me that they didn’t take this decision lightly, but nowhere do I see anything that helps me — someone who has paid for the iOS and Mac versions and hasn’t seen any useful features added in years. But I do get a discount for subscribing to a product I already fucking paid for.

No thank you.

Post Script: out of a sense of hopefulness and fairness and not wanting to be horribly wrong, I downloaded the new version.

The discount for existing owners is 25%. I.e. $30/year vs $40.

I found one nice bug fix — if you search for something and then edit a file found by searching, it doesn’t whisk the file away and lose context if you delete the search term while editing.

Also there’s a nice new feature — inline image previews. Long overdue but also not that well-implemented.


Voldemort vs. Casmir vs. Every Asshat in the Seven Kingdoms — The Villain Problem

Here’s the tl;dr:

Please, HBO, don’t make a bunch of Game of Thrones spinoffs. I’m virtually certain they’ll suck. How about an adaptation of the Lyonesse trilogy? I’m guessing it would be dirt cheap to get the rights.

At its heart, Game of Thrones is vapid

As I watched a collection of “moments” compiled by HBO for prospective viewers of the final season of Game of Thrones it struck me how few of these “moments” had anything to do with characters anyone cares about, vs. the “white walkers” who have no discernible motivations and appear as little more than teasers and shock-vignettes. If you subtract the zombie army and Bran’s entire storyline, what’s really left?

In a nut, Game of Thrones is a story about a bunch of people viciously squabbling over who gets to preside over famine and death in an inevitable mini-ice-age — “whose skeleton sits on the Iron Throne”, as Ser Davos puts it — instead of, say, stacking firewood and salting meat. When Jon finally points this out, he is met with disbelief, despite the fact that this has all happened before and is well-documented. Heck, there’s even a standing army devoted to defending against it — of which Jon was commander — but it apparently misplaced its instruction manual.

There are lots of unpleasant people doing horrible things, but in the end none of that matters because there’s an army of zombie boogeymen to contend with, and the fact that none of the interesting (but pointless) squabbling that has filled the last six seasons actually matters much, and HBO wants us to know that we need to remember a few tantalizing glimpses of “white walkers” who have said almost nothing and whose motives have never been explored (nor would they likely stand up to any scrutiny).

I started this piece before the premiere of season 7. As I revise this post before posting it, I’ve just watched the third episode of the final season, The Queen’s Justice. So far, there has been little of consequence in the first three episodes of the final season beyond the use of plot devices to tie off loose ends and weaken Daenerys’s hand for purposes of evening her odds against Circe. In essence, so far Season 7 has been The Euron Show. Euron, a leeringly obvious plot device, who since usurping his niece’s throne perhaps a year ago has assembled the “largest fleet in the world” — on desolate islands with no obvious source of timber, which just goes to show how resourceful he is — and then with stunning intelligence (and we assume favorable winds) fights and wins two massive engagements against Daenerys’s navies (and held a victory procession in King’s Landing) in the space of about two weeks.

Wait a second: shall we pause a moment to recollect that Daenerys is served by Tyrion and “The Spider” — the latter commanding a global spy network — 2/3 of the brains in the Seven Kingdoms (the missing third being Littlefinger) — and yet she seems to have no clue what her enemies are up to nor how to avoid tipping her hand to them (apparently the mysteriously empty castle she walked into with no thought to security is full of spies).

Outside of The Euron Show, Sam discovers a mountain of “dragon glass” and cures Mormont, allowing him back in the game. And Jon finally meets Daenerys and discovers that she’s an idiot. (By the way, do you find it a bit annoying that Jon can sail a single ship from the far north to the far south between episodes, but Daenerys sends two fleets which arrive two weeks apart?) And horrible things are done to minor characters to show how mean various mean people are.

Where’s Tom Riddle when we need him?

Perhaps the best major villain in a fantasy blockbuster is Tom Riddle, a.k.a. Voldemort, a.k.a. “he who must not be named”, if solely because he has two qualities utterly lacking in most major villains these days, i.e. a driving motivation (fear of death), and a goal that at least makes some sense both to him and his followers (run the world, put wizards in charge and enslave everyone else).

It all makes me think fondly of Spike, in Buffy the Vampire Slayer, who switches sides from Evil to Good because the Evil team wants to destroy the world, and he “quite likes the world”.

I recently finished reading the entire Harry Potter series to my twins, and it was wonderful. As a child I was so proud to read silently “in my head”; to rediscover the joy of reading as a performance later in life has been a revelation. I think that books are actually better read aloud, but it’s quite an effort to do all that reading.

Also, it’s hard to come up with voices and keep them straight (so I usually don’t bother, but the girls prefer it when I do).

Reading aloud also exposes a lot of weaknesses in a writer that reading quietly in one’s head conceals — e.g. I find Rowling to write dialog very poorly. I often paraphrase her characters for various reasons, but worse I find that she provides the wrong information in the wrong order. E.g. I’m often fooled into thinking person A is speaking when it’s person C (so I find myself having read in the wrong voice), or being told a character is speaking in a particular way too late to do any good (Hermione speaks “shrilly” all the time, a clumsy word and a dubious adverb for a feminist to constantly apply to her best female character).

Even so, my admiration for J.K. Rowling is considerably greater for having read these books aloud. For me, to read aloud as a performance is to pay much closer attention to each word than I ever do when reading to myself, and to feel the rhythm of dialog, and have to actively imagine the emotional takes of each character all the time. I notice and remember many things I missed on previous readings (or reading in this case, since I only ever read each book once before). It’s clear to me that Rowling planned the series carefully and well. I do have a lot of issues with it, but I have a lot of issues with everything. After all, I’m writing a critique of the current holy of holies, Game of Thrones.

Voldemort is, basically, Hitler

Voldemort doesn’t get to kill that many people, but in principle he is a ruthless race supremacist who plans to kill and/or enslave the “inferior” races, and he gathers about him fellow race-supremacists and opportunists. It’s not a subtle construct, but Hitler has about him a great deal more plausibility than most fantasy villains since, well, he actually existed. To understand why people might follow someone like Hitler, one can look at history books, or psychological research such as Milgram’s famous experiment. Hitler didn’t need to use mind control, he followed a well-established despotic populist playbook — presiding over a large scale criminal enterprise with all the usual systems of loyalty and reward. He offered a carrot and a stick. Ordinary German businessmen profited from slave labor (as do businessmen in the South, who use prison labor today, and as does the private prison industry from Trump’s war on immigration).

Even so, Voldemort exhibits some classic villainous behavior that is quite childish. E.g. I know of no evidence that Hitler was personally sadistic. He hired sadists. He created an environment where sadists were able to thrive. He didn’t go around gleefully torturing and murdering people. Indeed, Hitler portrayed himself and saw himself as a good-natured family man (odd though his family was). He understood the value of PR.

My single biggest disappointment with the Harry Potter books is that the resolution depends on only the least of Voldemort’s core failings (he doesn’t understand “love” — awwww) and not his others (e.g. he’s a sociopathic race-supremacist). Indeed, even in the final battle when the House Elves join the fray, it is as comic relief (stabbing people in the ankles with kitchen knives) and ignores their vastly superior magical powers (how about apparating Death Eaters into volcanoes?)

All through the seventh book I couldn’t stop thinking of the climax of Ralph Bakshi’s Wizards where the gnome-like hero faces down his magically overwhelming brother and shows him “a trick mom taught me when you weren’t around”. Harry and Hermione were both raised as muggles, and yet never use their knowledge of the muggle world to significant advantage, despite the fact that muggles have powers superior in most ways to wizard magic (e.g. cell phones, night vision goggles, 9mm handguns). Indeed, zero attention is made to “Muggle Studies” beyond its teacher being one of Voldemort’s gratuitous victims.

Scouring the countryside for evil deeds to commit

There are many wonderful and unique things about Lyonesse as a fantasy epic, but perhaps the single greatest is the central villain, Casmir. (Note that Lyonesse has quite a few antagonists, and of them Casmir is hardly the most “evil”, but none of the villains is as cartoonishly implausible as pretty much every bad guy in Game of Thrones.)

Casmir is an ambitious, calculating, and ruthless medieval monarch. That’s basically it. He’s not a sadist. He’s not a racist. He’s definitely not a religious fanatic. When he does do nasty things, he does it on the down-low because he wants to be loved and feared. Like Voldemort, he’s not a touchy-feely kind of guy, but he’s not even devoid of conscience (e.g. he only carries out unjust sentences to avoid being seen as weak, and he recognizes the truth of accusations against him for his mistreatment of his daughter, Suldrun). He is a competent and diligent ruler (because he thinks it’s important to have the good opinion of his subjects). Indeed, as a ruler he’s distinctly preferable to virtually all of his rivals.

Casmir seems a villain chiefly because of his relationship with his daughter, Suldrun, and to a lesser extent because he has a tendency to quietly murder his former spies (both to save money and avoid risk). He has many admirable traits and a dry sense of humor. In essence, to simply be a competent medieval monarch you need to do a lot of things that seem pretty evil by today’s standards, and he does them with no great gusto. (Aillas, the main protagonist, executes a lot of people, and Shimrod, one of his friends, tortures a man to death.) If we knew nothing of Suldrun, and nothing behind the scenes of Casmir’s reign, we might consider him a capable and ambitious man who had a lot of bad luck. Similarly, if we were to judge him by the people he surrounds himself with, he is not a terrible person.

Furthermore, we can fully understand why people would choose to support Casmir and risk their lives on his behalf, we can picture a world in which he is victorious, we can see why the common people might not care whether he wins or loses, and yet he is an entirely satisfying villain. Not only is this more morally sophisticated than most fantasy novels, it’s more morally sophisticated than most contemporary dramas.

Let’s do a great epic fantasy with a great villain

There are many reasons I prefer the Lyonesse Trilogy to the Song of Fire and Ice. To start with, it’s shorter and much faster-paced. The Elder Isles seem like a place people would generally enjoy living in. The plot is clever but not incomprehensible and the cast of characters large but not unmanageable. The military strategizing does not dominate the story, but the handling of strategy and tactics is both deft and on point. (The competent generals actually care about lines of supply and reconnaissance. Their navies, for example, are not constantly taken unawares.) It also manages to blend historical allusion (it’s set on an archipelago in the Atlantic that has sunk beneath the ocean) with fairytale qualities and brutally plausible pagan practices.

And, not infrequently, it’s also very funny.

Lyonesse is not without flaws — the third book feels rushed (Vance clearly wanted it done with and had worked out the plot arc well in advance), There’s casual and unnecessary homophobia, and there’s no warrior queen in the first book (Ehirme and Glyneth would need to be beefed up, Glyneth could be made a couple of years older, Yane could just as easily be a woman). Vance’s dialog is often hilarious, but over-stylized for some tastes. Nothing a good TV adaptation can’t fix.

But the reliance on faceless, inexplicably and implacably evil major villains who just want to destroy the world, and a motley supporting cast of pointlessly sadistic lesser villains, is the single catastrophic failing of Game of Thrones. I’ll watch it to the end, but seriously. Let’s have a story with actual sides.

System Update Weirdness

I’m simply recording this for the benefit of others who may have similar weird experiences updating to 10.13.6.

I recently updated macOS via the App Store from 10.13.5 to 10.13.6 and it was unusual in that it got about two-thirds of the way through the install fast (while saying it had 16 minutes to go) and then got stuck.

When I noticed the problem a couple of hours later, I

  • forced power off and restarted. It then quickly went up to about the same point it previously hung, rebooted, got to about 90% of the way done, rebooted again, then got to about the same spot and the screen went black and stayed black.
  • forced power off and restarted (I might have had to do this more than once at this point) and then it got to the faux sign-in screen you now get when Macs aren’t fully booted (e.g. after the battery runs flat during stand-by).
  • I logged in and got to the proper login screen, and was was relieved to find that everything from there proceeded normally — the OS asked me if I wanted to send feedback to Apple, etc., which seems to happen after some OS updates but not others, but also wasn’t the “new user” experience where you’re asked to log into iCloud again.

As a side-note, this is an experimental first post via Ulysses.

Please enter date of birth…

I’d like every person who has implemented a date picker control to enter the birthdays of their living relatives one-hundred times using their own date-pickers. Then try some ancestors.

Date pickers are perhaps one of the worst controls one deals with on a daily basis. They’re pretty terrible even when used for their expected purpose (entering dates for appointments). Perhaps the simplest criterion by which to judge them is:

Is your date picker easier to use than a keyboard?

A followup question would be:

If you want to use your keyboard anyway, will you find it harder than if your date picker weren’t there?

The primary reason date pickers exist is not for the user’s convenience. Entering a date using a keyboard requires, typically, 4 keystrokes for a nearby date (e.g. “4/1” or “12/25”) and at most 10 keystrokes for a date with specific year. Now, if you’re picking a date within a predictable and small period of time, you can provide graphical calendars to reduce date entry to a single click, and date-range-entry to a single click-and-drag. So, in a certain restricted set of use-cases, date pickers can be convenient.

The iOS Date Picker
The iOS Date Picker makes me furious every time I use it. The absolute best case is that you want to set something in the next few days. Anything else and you need to remember where you started and then scroll. Now what if you want to schedule your next teeth cleaning in six months? OK, what if you were entering a date of birth? Good. Fucking. Luck. And there’s no option to simply type in the damn date.

So, if I’m trying to enter a new appointment, and the date picker defaults to, say, showing today’s date and time, we’re in pretty good shape. Mostly.

Similarly, if I want to move an appointment from today to tomorrow or from Tuesday to a week from Tuesday we’re again in pretty good shape.

But if I want to select the time I started working at Andersen Consulting (October 1993?) things get a lot less pleasant fast. If I want to select the time I spent living in Canberra (March 1983 to September 1993… ish) it’s getting seriously unpleasant. If I need to enter birthdates for my family… blech. Luckily with America’s healthcare system I probably only need to do that a few dozen times a year.

One reason date pickers exist is because parsing typed or written dates is difficult. To begin with, the US has chosen to use a date format that is different from that used in every other part of the world — month / day / year. Since a lot of dates are ambiguous (if the first two numbers are both 12 or less then you can’t know for sure which is which) a date picker allows the UI designer to make the user’s choice clear to the software.

The simplest option in most cases would be to provide a calendar with a text field. If you type a date, it gets parsed and displayed on the calendar. If you click on the calendar, it replaces the typed date. In most cases, the calendar can be something that’s only visible when the text field is in focus.

Chrome's date picker
Chrome’s date picker is actually one of the best I’ve seen. That said, it has “clever” touches that can be confusing (e.g. if you type “4” as month, it automatically advances to the next field before you type “/” or start typing the day) which in turn means that it won’t let me enter the date in non-US order and flip it, let alone enter the date in non-numeric format (e.g. March 15), and it doesn’t deal well with anything less than a four-digit year.

Another reason date pickers exist is that you often want to display dates, and date pickers allow dates to be rendered in a consistent form, often with useful context. E.g. today vs. the date selected. Also, if you’re going to display something, then it’s nice if you can also directly manipulate it.

But, directly manipulating dates doesn’t really make sense (unless maybe you’re from Gallifrey). What you want to manipulate is events embedded within calendars. In the end, the best solution is probably to figure out how to avoid date pickers altogether.

Mac OS X Calendar
Mac OS X Calendar doesn’t use a date picker. You simply type in your appointment in shorthand (and look, it figured out I had entered date/month automagically). This also happens to have been a feature of the Newton (which would actually figure out that what you had entered was an appointment).