i spent last weekend in pittsburgh…


…because i was asked by the creative writing program at carnegie mellon (my alma mater) to give an address for their fiftieth anniversary celebration. 

i don’t think i have anything remotely resembling consolation, solace, wisdom, or sense to make of the the awful events of the last few days, but i would like to share what i had to say…

One of my favorite movies is Thomas Alfredson’s adaptation of John Le Carré’s novel “Tinker Tailor Soldier Spy”. I have seen this film twenty-seven times. I have read the novel twice, along with its sequels. I have also seen the 1976 BBC miniseries and the follow-up Smiley’s People, both starring Alec Guinness. 

To this day, however, I really can’t tell you what happens in Tinker Tailor Soldier Spy… other than it has something to do with men so profoundly closeted that they just have to destroy the Soviet Union. 

That much said, there is one moment that sticks with me in spite of my complete inability to understand the film on the level of plot or character… or really anything not having to do with ugly men’s suits from the 1970s. In the scene, Spy Master George Smiley – played here by Gary Oldman – has a painful reckoning with the lifelong friend and colleague he has uncovered as a Russian mole in the upper echelons of British Intelligence. 

When asked why he betrayed Queen and Country, the mole replies – with the quivering upper lip of a man who has never, ever, not once, talked about his feelings – that:

“It was an aesthetic choice as much as a moral one, and the West has become… so very ugly.”

If you think that there is a cheap analogy to be made between that line of dialogue and our current political situation, then let me assure you, you hired the right TV writer for the job.

Like so many sensitive and artistic people who were prone to melancholy even before our democracy went to shit, I feel that the present tyranny is not merely political, but also rhetorical and spiritual. It is a living expression of the belief that the only victories that have value are the ones earned by the fist or the crotch. 

Part of where we find ourselves in our politics, discourse, and relationships is the result of a systemic failure of kindness and empathy in our daily practice of living. It is a failure exacerbated by the pernicious yet seductive fantasy that those who act like the rich and cruel will magically gain wealth, and the concomitant entitlement to further, more ambitious cruelty. Too many have bought into the idea that destroying those whose existence annoys us on principle is equal to “winning”.

So. What does this incredible bummer have to do with the anniversary of a creative writing program?

Well, I am a writer and, increasingly, an artist… which is embarrassing, because, when I went into TV in the 90s, the medium was straight-up disreputable and I was honestly trying to sell out. Also, I work for other writers in an industry that not only tolerates, but often lionizes a great deal of madness, rage, and abuse in its upper management. I work in an industry that deliberately keeps its barriers to entry stocked with cauldrons of boiling oil; especially for women and minorities. I work in an industry that forces people to move to Los Angeles.

A few years ago, me and a man named Jose Molina – who is not only a dear old friend but also a very accomplished television writer/producer – were sitting at a bar on Ventura Boulevard getting drunk before our weekly Dungeons & Dragons game… like you do. As we ordered our second round, we began to list all of the sociopathic abusers for whom we had worked over the years. The resulting list, which could charitably be called “a real cavalcade of sigmoid orifices,” was lengthy, and encompassed the majority of both our careers.

As the third round arrived, we fantasized about what it would take to change the character of so many awful bosses to whom we had given our best and most creative work. What kind of Ebenezer Scrooge experience would it take to shift the minds of so many who felt entitled to cruelty, or availed themselves to it through sheer ignorance, to the real work of uplifting their employees instead of torturing them?

Soon however, we came to the depressing realization that, once someone achieves a position of power, it is as if society has given them a validation that their methods do not need to change. A TV network gave you hundreds of millions of dollars to make a show you created. Surely that means you are a genius and beyond correction. So we came up with the idea of making a podcast to, ostensibly, teach newcomers about the TV business. This was an attractive hook, since information about how to break into TV, and how to behave once you’re in, is notoriously hard to come by, and expensive and seldom accurate when it is.

But that really was just a Trojan Horse for the far more insidious message that you do not have to be an asshole to be successful, an abuser to get your collaborators to perform, or need to destroy everyone out of fear that they will someday come back to murder you and take your place… or worse, tell other people what you were like before you were a Genius.

The podcast has the colossally stupid name “Children of Tendu”, it’s free, and the twenty or so episodes we have produced to this day have been downloaded over 400,000 times. Based on that, we approached the Writers Guild and spent a year teaching a series called “Living in the Middle” which was intended as a sort of continuing education for writer/producers but really was a crypto-altruistic mental inception made in the name of decency, integrity, and collaboration. 

If you know the entertainment business – or try to enter it – then you probably know that a lot of people with far less experience than I will gladly charge you a lot of money for far less and pretend they are giving you the keys to the kingdom. We took forty combined years of hard-earned knowledge and experience and gave it away for free. 

The effect this has had on me as a writer, a manager of creative people, and a human being has been immeasurable. I believe in evolution, and evolution means that if your competition is loud, obnoxious, and seemingly victorious, and you are quiet, introverted, and don’t have a good publicist, you can still prevail by surrounding the bad guys with people who just plain disagree with them. Evolution is a numbers game, and we each have the power to help make more of us than there are of them. 

The efforts I have just described are a tiny little spearhead in that evolution. We arm our audience with knowledge, but also tell them unequivocally that much of what is understood as ordinary behavior in our business is wrong, and if they can’t change it, they’d better not feel entitled to further the cycle of abuse with their own actions. If Jose and I wind up helping one person succeed as a television writer while keeping them from falling for the temptation to act like something that was drained out of a colorectal fistula, then we can both enter our homes justified. 

Why? Because that person will, in turn, model good behavior to everyone who works for her or him – and those employees will know that it is good to work for an ethical, sensitive, and empowering manager, and they will take that into their own path to success – and the rest is evolution.

There are many of us who look at the current political situation – the endless lies and gaslighting, the normalization of casual hate, prejudice, and misogyny – and wonder only one thing… what to do? Some run for office, some go to demonstrations, call their congressmen, read all the newspapers and blogs… 

And some of us have no taste, stomach, or talent for it. I’m not talking about sloth or apathy, I’m talking about the truth that we all have limitations. I have none of the requisite qualities necessary to become a political truncheon; but that doesn’t mean that I don’t live in the same fish tank, breathe the same liquid, and choke on the same poison.

I have one solution. It’s not the only solution – not even close. It’s also not a solution that will force anyone out of the seat of power, bring about an end to the systematic abuse of women in our society, or expedite the punishment of the guilty… but that doesn’t make it not worth exploring. 

If you have a skill, teach it to someone for free. 

This is especially necessary if you are a writer. Because if you make it, it won’t just be because you are great (though I am sure you are, each and every one of you, the reincarnation of Mavis Gallant) it will be because all of the gatekeepers opened the locks for you. 

Anyone lucky – yes, lucky – enough to make a living as a writer has a list of people who recognized their talent, helped them hone their craft, gave them honest but compassionate feedback, introduced them to an agent, explained publishing to them, advised them on how to deal with editors – or network and studio executives – and perhaps taught them how to handle both success and failure with grace.  

And if you were to share your knowledge of how those pins and barrels were made to turn for you, imagine how many might be able to walk into the promised land through a path you described. How many stories might see the light that may otherwise have not. 

Every time you give knowledge away for free, every time you teach, every time you empower someone who is not, you are creating something that cannot be measured in terms of capital but is nevertheless a profoundly political act:

You are increasing the density of decency in the world. 

Every time you are generous with knowledge that is considered privileged, you not just saying that you have faith that the arc of the universe bends toward morality, you are becoming one of the billions of moving pieces whose trajectory defines that arc. Even more importantly, you are modeling kindness as a value.

The more we seed the world with the idea that empathy is not a hindrance to victory, but rather its vehicle – the more we declare through action that there is more to “winning” than profit and dominance – the more likely we are to create a majority for whom what is “cool” and “aspirational” is not for sale. Some of us stand up to bullies by quietly recruiting more and more into the ranks of warriors for that which is soft, and gentle, and far more necessary to the soul than “winning.”

And trust me, I’m not advocating that you put your professors or this university out of business. If you are doing it right, your experience of a place like this will be irreproducible. I’m also not telling you to give your secrets to “the competition.” What you bring to the page is the one thing you can’t teach or give away; the only way to truly understand that is to put everything else on the table. 

Some of you may now be saying “but what if I’m too busy changing the world through my amazing writing?” Then I would suggest you study the life of one of the most successful classical composers in history. A man whose voice dominated opera between the eighteenth and nineteenth century. A man who retired one of the wealthiest composers in history. A man who taught composition to Schubert, Liszt, and Beethoven. 

His name is Antonio Salieri. His story is a cautionary tale that should be familiar to all you cultural studies double-majors who have been force fed Roland Barthes. Today, this man’s name is known only because – a hundred and fifty years after his death – another artist wrote a grotesquely inaccurate script framing him for the murder of Mozart, holding him up as poster child for the sin of envy, and canonizing him as patron saint of mediocrity.

What happens to your work and your story in the ashes of time is completely beyond your control. The only thing that survives eternity is your place in an unbroken chain of acts that encourage others to surpass you. In the immortal words of Jedi master Yoda: “we are what they grow beyond, that is the true burden of all masters.”

  But ultimately, it’s a little weird that you should be listening to me on this topic when, not one square mile from here there lived and worked a true warrior for social justice – a writer, musician, spoken-word performance artist, film-maker, pacifist, and humanist – who said it better, and far more succinctly than I ever could. A man who has been vilified by Fox News and many other standard-bearers of speaking-power-to-truth – and who has frequently accused of everything from mass murder to child molestation – for the crime of putting into the world such heresies as this one:

“As human beings, our job in life is to help people realize how rare and valuable each one of us really is, that each of us has something that no one else has – or ever will have – something inside that is unique to all time. It’s our job to encourage each other to discover that uniqueness and to provide ways of developing its expression.“ 

His name was, of course, Fred Rogers, and he remains one of my personal idols to this day because he talked the talk, walked the walk, endured the cruelty and ridicule of many with the humane smile of one who knows truth beyond reproach, and embodied day in and day out – without shame, or fear, or guilt – a reality that, sadly, grows more urgent with each passing day… 

In this moment, there is nothing more revolutionary, counter-cultural, and just plain badass than simple human decency.

On Tumblr

It’s neural net Halloween costume time


People use neural networks for translating languages, recommending movies, delivering ads, and more, but this here is one of my favorite applications: utter surrealism.

Last year I trained a neural network on 4,500 Halloween costumes that you readers helped me collect. This year, I teamed up with the New York Times to train a different neural network on over 7,100 costumes. Jessia Ma illustrated lots of the costumes, even the ones that make NO SENSE and they are all SO GOOD.

Please please go look at the rest of them. You will be so glad you did.

And yes, there will be ill-advised sexy costumes.

Thank you, everyone who contributed a costume! You can still contribute to the dataset (maybe I’ll revisit it next year) by filling out this form.

Want even more Halloween costumes? Costumes like Dr. Poop and Ms. Frizzle’s Robot? The answer is yes. Sign up here to get them, plus optionally more bonus material every time I post.

vintagerpg:Serious nerd history lesson incoming.The first Dungeons & Dragons videogame came out in …


Serious nerd history lesson incoming.

The first Dungeons & Dragons videogame came out in 1982
for the Intellivision, but the burgeoning industry was already under the tabletop
game’s influence. By 1980, two games represented a kind of fork in the
philosophical road for computerized RPGs. Rogue focused on the dangers of
dungeon crawling and complex rule sets that verged on the mystical – it was
essentially a simulation of D&D mechanics where stories emerged from the
action without narrative guidance. Down the other path lay Zork.

Zork was developed by students at MIT from 1977 to 1979. It was
inspired by Will Crowther’s 1975 mainframe game Colossal Cave Adventure that,
though it lacked monsters, was directly inspired by Dungeons & Dragons
sessions (which included Zork writer and Infocom founder Dave Lebling). Zork
was definitely fantasy, though, with a vast underground empire to explore,
treasures to find and monsters to fight (or be eaten by, if we’re talking about
the darkness dwelling Grue).

Zork is an interactive fiction, that is, everything is
presented as text. You direct your actions by typing them into the command line
and a bit of code known as a text parser acts as a kind of dungeon master (Zork
III’s subtitle actually is Dungeon Master, come to think of it), interpreting
your commands and telling you their consequences. If the Dave Arneson school of
D&D thought sought to have players inhabit the fantasy stories he read and
loved, then Zork is perhaps the closest we’ve come to that Platonic ideal.

I love Zork. It is as old as I am, has no flashy graphics,
and yet remains my favorite videogame of all time. It stoked my imagination as
no other videogame has, but in ways similar to D&D. As a kid, peering at
the green monochrome screen, trying (and mostly failing) to work out the devious
puzzles. I didn’t make much of a distinction between Zork and Dungeons &
Dragons. Even though they didn’t share a brand name, I knew they were both
facets of a larger world.

Interactive fiction mostly died out in the late 80s, leaving
the mechanical influence of D&D to dominate videogames until recent years,
when technology has allowed complex narrative to remarry rules systems in
something that approximates the experience of telling a story with friends
around the gaming table. Sort of. But I’m getting ahead of myself…

This AI is bad at drawing but will try anyways.


There was a paper recently where a research team trained a machine learning algorithm (a GAN they called AttnGAN) to generate pictures based on written descriptions. It’s like Visual Chatbot in reverse. When it was just trained to generate pictures of birds, it did pretty well, actually.
This AI is bad at drawing but will try anyways.
(Although the description didn’t specify a beak and so it just… left it out.)

But when they trained the same algorithm on a huge and highly varied dataset, it had a lot more trouble generating a picture to go with that caption. Below, I give the same caption to a version of their algorithm that has been trained to generate everything from sheep to shopping centers. Cris Valenzuela wrapped their trained model in an entertaining demo that attempts to generate a picture for any caption.
This AI is bad at drawing but will try anyways.
This bird is less, um, recognizable. When the GAN has to draw *anything* I ask for, there’s just too much to keep track of – the problem’s too broad, and the algorithm spreads itself too thin. It doesn’t just have trouble with birds. A GAN that’s been trained just on celebrity faces will tend to produce photorealistic portraits. But this one, however…
This AI is bad at drawing but will try anyways.
In fact, it does a horrifying job with humans because it can never quite seem to get the number of orifices correct.
This AI is bad at drawing but will try anyways.
This AI is bad at drawing but will try anyways.
This AI is bad at drawing but will try anyways.
It’s fun to ask it to draw animals though. It knows the texture of giraffes, but not quite exactly their shape. And it knows that boats are on the water, but not necessarily that they are boats.
This AI is bad at drawing but will try anyways.
It also (like many other image recognition algorithms) gets a bit confused about the difference between sheep and the landscapes they’re found on. Other algorithms recognize sheep in pictures of empty green fields. And this one, when asked to draw sheep…
This AI is bad at drawing but will try anyways.
That’s different, though, from asking it to draw *a* sheep. In that case, it knows exactly what to do. It draws the sheep, and then just to be safe it fills the entire planet with wool too.
This AI is bad at drawing but will try anyways.
It really likes drawing stop signs and clocks. Give it the slightest opportunity to draw one, and it will chuck those things all over the place.
This AI is bad at drawing but will try anyways.
This AI is bad at drawing but will try anyways.
Other than its horrifying humans, this algorithm can actually be pretty delightful.
This AI is bad at drawing but will try anyways.
This AI is bad at drawing but will try anyways.
This AI is bad at drawing but will try anyways.
This AI is bad at drawing but will try anyways.
This AI is bad at drawing but will try anyways.
This AI is bad at drawing but will try anyways.
Try it for yourself!

I had way too much fun generating these and ended up with way more than would fit in this one blog post. I’ve compiled a few more of my favorites. Enter your email and I’ll send you them (and if you want, you can get bonus material each time I post).

So let’s bury this bullshit about how women didn’t grow up on Star Wars.



This is my friend TJ, wearing a costume she made for Halloween, 1977.  She was 16 at the time.

Now, keep in mind: there was no internet to search for images.  She could not have rented and paused the movie, because it wasn’t released on video until 1982.  No, TJ just went to the movie a bunch of times, took notes with a flashlight, drew a bunch of sketches, and put this together.

In 19-fucking-77.  So let’s bury this bullshit about how women didn’t grow up on Star Wars.

Some dude tried to explain why TLJ is an insult to real fans and since I liked it I wasn’t a real fan and SON, I GOT A DOT MATRIX PRINT OUT OF Star Wars: New Hope, The Journal of the Whills, Part 1 IN A DRAWER SOMEWHERE, COME AT ME.

My mother is 74 years old and knows what a Corellian blood stripe is. The ladies have been here the whole time.

On Tumblr

staff:“Keeping the internet open is critical for us. It powers social movements, and provides …


“Keeping the internet open is critical for us. It powers social movements, and provides a global platform for people of color, LGBTQ folks and the most marginalized communities to tell their own stories, run their own businesses and route around powerful gatekeepers.”—Candace Clement, Free Press Action Fund Campaign Director via @fight4future

Starting today, June 11, U.S. internet providers will be legally allowed to censor and block websites and apps, and force you to pay extra fees to to access your favorite places online. Your internet sanctuaries, the communities you are part of, the ones you have help build up, could be decimated.

Will it happen today? No. Next week? Probably not. The changes will not be swift. They will come piece by piece. A slow, tempered death to the free and open internet we love.

It doesn’t have to be this way. You can still make a difference, Tumblr. We need the House of Representatives to sign a discharge petition in support of the Congressional Review Act that would force a vote on the floor.

Contact your reps—let them know you support net neutrality.

It’s so easy. Just go to BattleForTheNet.com, fill out the form, and follow their directions from there.

They have an updated widget for you to throw on your websites to urge others to make a difference. You can put it on your Tumblr. Let your followers know what you stand for, encourage them to do the same. It’s so easy to do. Just copy and paste their small line of code right into the customize theme page on the web.

Go, go, go, go. We know you have that passion in you. We’re fighting right alongside you.

A.I. scream for ice cream


Or, the time a class of middle schoolers kicked my butt at neural network ice cream naming.

The other day I got an email from Anita Johnson, who teaches coding classes at Kealing Middle School in Austin, Texas. She explained that her students had been reading the neural network experiments on my blog and had decided to do their own.

The middle schoolers (about 11-14 years old) had downloaded textgenrnn and had generated some new flavors that they wanted to share with me. Did I want to see them?

As it turns out, I had just trained textgenrnn on ice cream flavors myself. But there were some problems with my attempt:

1. I only had a dataset of 200 flavors (that I had a dataset at all is thanks to Salli Wason of Rosanna’s Ice Cream in Portland).
2. textgenrnn allows transfer learning, meaning that it remembers some of what it learned from its previous dataset.
3. My previous dataset had been metal bands.

So the flavors I had generated were not, shall we say, appealing.

Silence Cherry
Strawberry Cream Disease
Chocolate Sin
Bloody Coffee
Sock Caramel
Chocolate Raven
Moan Chocolate
Chocolate Chocolate Blood
Colon Bane

The 11-14 year olds, however, had collected nearly 1,600 flavors, due in part to their coding skills, and in part to sheer numbers, time, and motivation. Their results were significantly better than mine.

Lots of the flavors they produced were sweet and fun. The kind you might find  at a trendy ice cream or yogurt shop near you.

It’s Sunday
Cherry Poet
Brittle Cheesecake
Honey Vanilla Happy
Bubble Bun
Triple Bun
Holy Lemon Monster
Cookies & Red Hot Lover
Vanilla Nettle
Sundana Rainbow
Team Cherry
Cherry Cherry Cherry
Chocolate Breath
Pig Nut
Bumble Cookie
Oh and Cinnamon

Other flavors sounded a bit weirder. Approach with caution.

Chocolate Finger
Caramel Book
Cupsie Core
Washing Chocolate
Peanut Cinnamon Budge
the United Bacon de Vanilla
Texas Boy Nut
Key de Smoke
Crackberry Pretzel
Salted Pie Breekberry Sundae
Texas Charlie Covered Stunt
Seat Strawberry
Butter Sweep
Bunny Out
Strawberry Moons
Pretzel Egg

Others: a quite worrisome level of ambiguity

Nuts with Mattery
Brown Crunch
Sticky Crumple
Cookies and Green
Sea Cheesecake
Mango Cats
Lemon Cream Grassplay

Whereas these are almost certainly bad news.

Pumpkin Trash Break
Peanut Butter Slime
Walnut Dirt
Peanut Skin Tuesday
Gravy Cashew
Ants-Almond Cheesecake
Chocolate Gingerbread Bum
Strawberry Turfle
Cake Pitt
Caramel Livers
Elk Peanut Southe
Sand & Cream
Toffee Frog
Snake Vanilla Cream Cheesecake

The 6th graders (the 11-12 year olds) tended to favor the very weird flavors.

Garamel Phankie Cookies & Peach
Cark Nutty Banana Croced Banana & Crazz
Vervette’s Caramel Borfle
Oatleak with Ninterbise
Barming French Cambarcot
Chocolate Blackbumple
Herbetures with The Chillin
Pie Lime Mint Thrippine
Praline Pelletral Liver
Banana Cookies & Jarange Core
Peach Peacket Marsh Blue

The 13-14 year olds, though? They preferred my first attempt, actually. And added a few of their own to the list.

Orange Chocolate Killa
Blood Chip
Explosion Stick
Die White Pistachio
Funge Ecide

For the complete list of the coding class’s ice cream flavors, as well as a few PG-13 flavors that weren’t quite appropriate for the main blog, enter your email here.

staff:🚨This is a Red Alert for net neutrality 🚨Last December, the FCC voted …


🚨This is a Red Alert for net neutrality 🚨

Last December, the FCC voted to to kill net neutrality. If we do not take action, this will kill the free and open internet as we know it. The internet needs you—all of you—to make sure your voices are heard NOW.

We need all hands on deck for this one. It may be our last chance. If you’re feeling under-informed and overwhelmed about why net neutrality is so incredibly important, we have this handy guide just for you.

Here’s what you can do to save the internet:

  • In mid-May, the Senate will vote on a resolution to overrule the FCC using the Congressional Review Act (CRA). We only need one more vote in the Senate to win. Write or call your Senators or Representatives. You can also text BATTLE to 384-387 to get more information on how to write to your reps. You can do this, Tumblr.
  • Join us and dozens of your other favorite companies like Etsy, Vimeo, Reddit, and GitHub to raise awareness with the Red Alert campaign being run by Battle for the Net. Just add this small widget to your Tumblr to let your followers know how they can contact their reps. It’s as easy as copying and pasting the small line of code right into the customize theme page on the web.

This is important. This matters. It’s up to you to help. 

D&D character names – generated by a neural network


There are algorithms called artificial neural networks that can learn to imitate examples of just about anything. They’re used in all sorts of everyday programs, translating languages, identifying photos, colorizing drawings, delivering ads, and tons more.

It turns out neural networks may also be a dungeon master’s best friend.

I’ve trained neural networks to invent new Dungeons & Dragons spells (part 1, part 2) and also trained them to name new D&D creatures. It worked very well (Shield of Farts, anyone?), thanks to the spellbooks and monster manuals I could use as datasets. But there weren’t any datasets for another big aspect of Dungeons & Dragons: all the characters who populate these worlds. So, over the past few months, readers have been helping me to build a dataset – which has now reached a staggering 20,908 entries.

For each character, people entered a name, a race (human, dwarf, elf, etc), and a class (wizard, rogue, bard, cleric, etc). Some of the races and classes got to be quite inventive – there’s a penguin, a fey corgi, a black pudding, and a sentient bucket. So I gave this huge weird list to a neural network to see how convincing it could sound.

With nearly 21,000 examples, the neural network could indeed sound convincing. Much of the time, the names matched the character type – at least as often as in the original dataset (which had 5 characters named Frank and 12 named Tim).

Rose – Human Assassin
Dwarg – Half-orc Paladin
Liandra – Elf Wizard
Oron “The Star” Cartere – Dragonborn Sorcerer
Silvar the Blackblade – Half-elf Barbarian
Hank – Half-orc Ranger
Jayne Arryn – Half-elf Wizard
Annata Shortscale – Dragonborn Witch
Fyrry – Half-Elf Ranger
Rinas Mistfern – Human Ranger

Other names made perhaps less sense.

The Cart – Kenku Rogue
Nine Case – Dark Elf Fighter
Rump – Kenku Cleric
Gubble Daggers – Tabaxi Monk
Bog – halfling wizard
Jameless – Dwarf Champion Barbarian
Rune Diggler – Halfling Rogue
Borsh the Bardlock – Human Paladin
Spullbeard – Dwarf Fighter
Tovendirgle – Human Ranger
Pinderhand The Bugs – Gnome Wizard
Rune Wash – Human Wizard
Stumbleduckle – Human Paladin
Dawne Shift the Monkz – Dwarf Barbarian
Magnus Tieforian the magnificent von Cloriam Cyital DuP Ever – Dwarf Barbarian
E Ch BISHL NEBe Garte II Cr D McLGHJ T U E AA t Rat lek TF Horn hand tree Whistle – half-orc barbarian

One thing I like is all the new character races and classes that the neural network discovered. I don’t know what most of them are, but you’ll be the only one in your party.

Kelph – Burryman Ranger
Arczi-Sian – Human Dogminer
Jho the Chrishpup – kuborg fighter
Archein Morgurowood – Human Weaponic Bloodlind
Bubblebottom Donder – Half-faerie Dewlze Cleric
Altis Helder – Mander Human Star-Caver Pottlebard
Bender – half-alf paladin
Devith “Kurgbore” Mustwost – Fetchlen Cleric
Varian Amerth – blackbear Bard
Merellios Rose – Rope Gnome Wizard
Mothrek McKingfoot – halfling inquisitive
The Cowben – Human Opera
Ayrell – Forest gnome Arcane Wood Hunter

One type of name the neural network did very well: silly compound names. This pretty much settles the question of whether a neural network would be totally on board with naming something Boaty McBoatface: it totally would.

Here is what it thinks dwarves should be named.

James Crucklebottom – Dwarf Wizard
Frank Firethorn – Dwarf Wizard
Willian Stonefrown – Dwarf Fighter

Actually, you know what? Pretty much everyone needs a name like this.

Kavar Blunderwood – Goliath Monk
Hadrie Trumbledutch – Halfling Rogue
Prinkina Timberspull – gnome sorcerer
Arrina Cuprest – Human Sorcerer
Tretcher Twestybeard – Dwarf Witch
Ponny Stonecharles – Human Monk
Ashrata Dangstrider – Ratfolk Rogue
Den Splatterwoof – Halfling Druid
Wolfrit Rockhole – Human Sorcerer
Beddar Jacklebottom – Halfling Cleric
Azrara Stoutfrogg – Half-orc Monk
Lord Filedawn – Halfling Warlock
Gripple Ravenhorn – Human Assassin
Balfeart Wolfspleam – Dwarf Fighter
Eldric the Bizzlebree – Human Warlock
Pig Haystalker – Human Assassin
Ladie Barewalker – Tiefling Warlock
Fay Blutterlocket – Dwarf Paladin
Millian Kricklebottom – Kobold Sorcerer

I’ve posted the entire original dataset here, and you can access a huge export of generated characters there as well. If you want the list plus a few extra that I deemed not quite appropriate for the main blog, enter your email here and I’ll send them to you.

Also! I’m still crowdsourcing a dataset of character bios (I used some of the names for this experiment). If you’d like to help, use this form.

Neural network-named tomatoes you won’t find at the farmer’s market


As gardeners know, a tomato is not just a tomato. They come in all shapes, sizes, and especially colors, from tiny red gems to big and purplish brown to pear-shaped with zebra stripes. And their tastes and textures vary wildly as well, with some sweet and soft, some tough-skinned and tart, and some dense and destined for processing.

These varieties all have names, of course, and good names are at a premium. Darren Abbey, who runs the blog The Biologist Is In, breeds new tomatoes and needs to name them from time to time. He gave me a list of 11,629 existing names, ranging from the highly experimental (Rts-119) to the classic (Sungold).

I used these names to train a recurrent neural network, which is a kind of algorithm that learns patterns that let it imitate the data it sees. Could it learn to produce plausible tomato names?

Yes, as it turns out! On Darren’s blog post you can read his expert opinion, but the verdict is: these are totally usable names.

Speckeled Boy, Sweet Lightning and Golden Pow

Sweet Lightning
Speckled Boy
Pinkery Plum
Market Days
Fancy Bell
Mountain Gem
Garden Sunrise
Honey Basket
Cold Brandy
Sun Heart
Special Baby
Golden Pow

However, not ALL the names were plausible. These are a lot weirder.

Smol Pineapple, Bumbertime and The Ball

Bear Plum
The Bango
Sun Burger
First No.4
Smoll Pineapple
The Ball
Golden Cherry Striped Rock
Old German Baby
Frankster Black
Ranny Blue Ribber
Adoly Pepp Of The Wonder
Cherry, End Students
Small Of The Elf
Champ German Ponder
Pearly Pemper
Green Zebra Pleaser
Flute First

These *seem* weird, but they aren’t actually all that weird for tomato names. There are at least 50 oxheart tomatoes in the original dataset.

Matt’s Sandwich
Indigo Tree
Sun Burger
Striped Hollow Potato Leaf
Lelly’s Yellow Stuffers
Terra Pink Strain
Greek Boar
Ton’s Oxheart
Babla’s German Paste
Mortgage Lifter, Honey Blues

These however? These tomatoes don’t sound promising.

Speckled Garfech, True Grub and Panterosa

Speckled Garfech
Green Dork
Cluster Gall
Shirve’s Gigant Bullburk
True Grub
Black Crape
Roma Wasting Star
Scar Giant
Bug Beauty
Banana Placente
Speckled Bake
Green Boor
Wonder Bagg
Sun Bung
Bolosavyi Scarling West Colon
San Blue Tard
Red Krap
Shart Delight
Solad Piss

For more in-depth analysis of these names, check out Darren’s blog.

I also trained a neural net on fruit names earlier – read the results here.

And this post’s bonus content is well worth checking out, because not only does it contain many more tomato names, but I also tried an experiment where I trained a neural net first on tomato names, and then on the names of metal bands. The results were glorious, including such wonders as

Angels of Saucing
Serpent Green
Market Death
Sweet Sever
Devil Potato Large Death

Fill out the form here and I’ll email you more ill-advised tomato names.

I wish we could put some of that snow in our mountains…



Indian hawthorn flowers in bloom, spotted on my lunch hour walk. It’s been a warm winter in California, and we’ve finally gotten some rain over the last few weeks (not enough, but some is better than none).

What a gorgeous photo, and I’m so envious of those blossoms!  We still have snow 🙁

Thanks! I wish we could put some of that snow in our mountains…

On Tumblr

Wikipedia articles invented by a neural network


Wikipedia has a page where they list, for entertainment purposes, the titles of a bunch of pages that didn’t meet the cut. These are mostly pages that were submitted as pranks, although a few of them are clever enough that you can’t quite tell. Reader Emily Davis sent me a list of them – here are a few real deleted articles that humans wrote.

List of movie posters with lamps in them
How to trick people into thinking you’re a wizard
List of people who died with tortoises on their heads
People Who delete My Articles have no sense of Humor
I like eggs
Do scented candles burn faster than unscented candles
An article that contains nothing but a full stop
List of differences between apples and oranges
Category:Farts in literature
Category:Political posters using an octopus
Woo woo woo woo woo woo wah ooooo wah
List of all Wikipedia lists that do not contain themselves

It makes a terrible dataset for a neural network – only 1112 unique entries, some of which are quite long, and big variation in style and subject matter. I decided to try it anyway.

I trained a character-level recurrent neural network (that is, it uses individual letters as building blocks) with a very small memory to prevent it from memorizing the small dataset so quickly. Even so, most of the generated names were either incomprehensible or memorized from the original dataset. Those that weren’t, however, fit right in. It turns out text-generating neural networks are great at mashups and non sequiturs.

Popal chickens
List of U.S. pants
List of the Hamburgers
Category:Athletes with maps
Why Inited States
Evil chicken
Liquid cheese
List of bands with pies on them
Ant Fields are bear hair fetishism
Monster Diseases
Why Won’t Space
Tire bear (country)
What hoop
This page is a very short article
Poople who don’t have beer from sydney
Goat that cookie
Near Dogs
Donkey words in the cartoons
Poople who woo wah the pilot
Death of chicken
What is the day
What fame butt
List of fictional characters with the ball
Who is not leaders
List of parps
Proper programming language
Turdis programming language
Article with a cat
Friends and existence
How to draw a coconut
Tree donkey
Category:People who can’t speed
Beer for chickens
Tree Wars

Whoever it is who likes to enter long strings of repeated characters as pranks (I’m looking at you, Sand Person), the neural network shares your obsession. Repeated text is easier to learn, and so the neural network tends to latch onto it easily and, especially when I give it a short memory, takes repetition to even wilder excess (see: The Cow With No Lips).

Beneral Pissednessessessinessismasticlesismsomic comotute

Woo woo woo woo woo woo woo woo woo woo woo woo woo wah ooooooooooooooooooooooooooooooooooo ooooooo ooo on other intortational characters with removable travel

Wich chemical appearaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa

Note: please do not actually attempt to create these articles on Wikipedia.

Bonus material: sign up here to get some more article titles, including a few that were a bit too rude to post here.