> ## Content Index
> Fetch the complete content index at: https://davidnunez.com/llms.txt
> Use this file to discover other available public pages before exploring further.

# Outhorse Your Email
- URL: https://davidnunez.com/outhorse-your-email/
- Published: 2026-09-17T13:38:40.000Z
- Updated: 2026-09-17T13:41:10.000Z
- Description: Iceland trained three horses to answer your email. Google, Superhuman and 147 startups have been doing the same job without the hay, and the research on how that lands with the person reading it is not flattering.
- Author: David Nuñez
- Tags: broadcast

They say if you give a monkey an infinite amount of time (and, presumably, an infinite pile of bananas), he'll [eventually peck out the collected works of Shakespeare on his typewriter](https://en.wikipedia.org/wiki/Infinite%5Fmonkey%5Ftheorem?ref=davidnunez.com). (Someone tried. In 2003 a zoo in Devon [gave six macaques a keyboard for a month](https://nowiknow.com/six-monkeys-one-month/?ref=davidnunez.com) and got five pages, mostly the letter S.)

I empathize with that monkey as I try to respond to the limitless stream of email flooding my inbox: The exercise is pointless, the work never ends, and most of it is gibberish.

Email is usually just codified corporate politics tangled in [infinite reply chains](https://www.themuse.com/advice/how-to-end-the-neverending-email-chain-without-being-rude?ref=davidnunez.com), [passive-aggressive use of the cc line](https://slate.com/human-interest/2019/11/email-cc-field-boss-passive-aggressive.html?ref=davidnunez.com), and [offensive signature lines](https://www.theguardian.com/commentisfree/2013/jul/17/email-sign-off-etiquette?ref=davidnunez.com). [Baboons](https://www.inaturalist.org/posts/67769-is-it-baboons-rather-than-chimps-that-really-epitomise-machiavellian-intelligence?ref=davidnunez.com) would be proud, and Shakespeare would have turned it all [into an existential monologue](https://www.youtube.com/watch?v=LdZVR4Ry3jQ&ref=davidnunez.com).

Andrew Scott, "To Be Or Not To Be," Almeida Theatre

In response to bad managers expecting 24/7 email responses, France gave its workers a legal [right to disconnect](https://verdict.justia.com/2017/02/07/new-french-right-disconnect?ref=davidnunez.com) in 2017 and Australia [followed in 2024](https://www.abc.net.au/news/2024-08-26/what-right-to-disconnect-from-work-means-for-you/104234308?ref=davidnunez.com). Iceland, however, went with an equine solution.

![](https://storage.ghost.io/c/2a/66/2a66aa35-855e-4718-a6f9-5e4f6e1dc30d/content/images/2026/09/CleanShot-2026-09-17-at-9.26.24-AM.png)

**Trained in corporate buzzwords. (via* [**Visit Iceland*](https://www.visiticeland.com/outhorse-your-email/?ref=davidnunez.com)**)*

In May 2022 its tourism board [trained three horses to type on a gigantic keyboard](https://www.visiticeland.com/outhorse-your-email/?ref=davidnunez.com). Hrímnir is "assertive, efficient." Litla Stjarna "types fast, but might take a nap." Hekla is "trained in corporate buzzwords." Iceland invited office-workers around the world to let the ponies write their email replies. "Your boss will never know the difference." [Twenty-one thousand people](https://www.oneclub.org/awards/theoneshow/-award/49342/outhorse-your-email/?ref=davidnunez.com) outhorsed an email during the campaign.

![](https://storage.ghost.io/c/2a/66/2a66aa35-855e-4718-a6f9-5e4f6e1dc30d/content/images/2026/09/image-7.png)

**Everyone in this photo is giving the answer. (via* [**Wiki Media*](https://commons.wikimedia.org/wiki/File:Osten%5Fund%5FHans.jpg?ref=davidnunez.com)**)*

[Clever Hans](https://en.wikipedia.org/wiki/Clever%5FHans?ref=davidnunez.com) toured Germany from 1904 answering arithmetic questions with his hoof. In 1907 the psychologist [Oskar Pfungst](https://en.wikipedia.org/wiki/Oskar%5FPfungst?ref=davidnunez.com) figured out the trick. The questioner leaned forward, without knowing it, as the taps approached the right number, and Hans stopped. Hans had learned to read people and his arithmetic skills lacked (ahem) horsepower.

"Reading people" is a foundational skill of the modern office, and it seems like horses are up to the task. We read between the lines of email threads, baking a lot of assumptions into those pithy words (or [overwrought screeds](https://zenhabits.net/snore/?ref=davidnunez.com)). Maybe the [email should have been a meeting](https://medium.com/behavior-design/this-meeting-could-have-been-an-email-from-time-suck-to-unstuck-42642ea6391b?ref=davidnunez.com), after all.

![](https://storage.ghost.io/c/2a/66/2a66aa35-855e-4718-a6f9-5e4f6e1dc30d/content/images/2026/09/CleanShot-2026-09-17-at-7.22.44-AM.png)

As far as I can tell, the Icelandic horse keyboard actually [worked as a functioning input device](https://www.youtube.com/watch?v=Qc8rsnRC1EU&ref=davidnunez.com). That is, if by "worked" you mean hoofs clomping randomly on plywood to make letters appear on a screen. If we are being fair, I suppose this is just as effective communication as most humans pushing random keys to generate text.

![](https://storage.ghost.io/c/2a/66/2a66aa35-855e-4718-a6f9-5e4f6e1dc30d/content/images/2026/09/image-6.png)

I love you. (via [Google](https://research.google/blog/computer-respond-to-this-email/?ref=davidnunez.com))

The horses aren't the only ones writing email, however. Google's Smart Reply was writing [a tenth of the replies](https://blog.google/products/gmail/smart-reply-comes-to-inbox-by-gmail-on-the-web/?ref=davidnunez.com) sent from its mobile inbox four months after launch, once engineers had trained out an early prototype's habit of answering ["I love you" to seemingly anything](https://research.google/blog/computer-respond-to-this-email/?ref=davidnunez.com). Superhuman, which I use, has offered [three finished replies to every message](https://blog.superhuman.com/superhuman-ai-instant-reply/?ref=davidnunez.com) since February 2024\. From the launch post: "Imagine waking up to an inbox where every email has a draft reply." Imagine, indeed. Imagine, as well, how much time we spend editing and deleting the proposed auto-replies because they don't quite "get it right." Whatever *that* means.

Cornell researchers found in 2023 that replies built from AI suggestions read as warmer. They also found that people who *merely suspect* you of using them [rate you as less cooperative](https://news.cornell.edu/stories/2023/04/study-uncovers-social-cost-using-ai-conversations?ref=davidnunez.com), whether or not you let AI help. A 2025 study of managers found that more than 80 percent of employees [read a lightly assisted email from the boss as sincere](https://www.psypost.org/managers-who-use-ai-to-write-emails-seen-as-less-sincere-caring-and-confident/?ref=davidnunez.com). About 40 percent said the same of a heavily assisted one. I take this to mean people are assigning value to email based on the perceived effort of the sender. This seems fair to me as an [email is an implied demand for attention](https://www.nytimes.com/2023/04/13/opinion/email-time-work-stress.html?unlocked%5Farticle%5Fcode=1.B1E.kL8l.uB8Au7MDDIrO&smid=url-share&ref=davidnunez.com). Including this one. Thank you if you've made it this far.

Effort can be a proxy for sincerity. Two months before the Cornell paper, an office at Vanderbilt emailed students its condolences about the shooting at Michigan State. [A line of fine print at the bottom](https://vanderbilthustler.com/2023/02/17/peabody-edi-office-responds-to-msu-shooting-with-email-written-using-chatgpt/?ref=davidnunez.com) read: "paraphrase from OpenAI's ChatGPT AI language model, personal communication." Students objected, rightfully, and the office apologized.

![](https://storage.ghost.io/c/2a/66/2a66aa35-855e-4718-a6f9-5e4f6e1dc30d/content/images/2026/09/CleanShot-2026-09-17-at-9.20.29-AM.png)

**Sorcerer's apprentice mode, est. 2004\. (via* [**rfc-editor*](https://www.rfc-editor.org/rfc/rfc3834.txt?ref=davidnunez.com)**)*

[RFC 3834](https://www.rfc-editor.org/rfc/rfc3834?ref=davidnunez.com), an internet standard from 2004, exists to keep automatic responders from replying to each other forever, a failure it calls "sorcerer's apprentice mode." Tom Fishburne [gives us the modern version](https://marketoonist.com/2023/03/ai-written-ai-read.html?ref=davidnunez.com): one person's AI inflates a bullet point into an email, and the recipient's AI boils it back down to a bullet point. All of it goes unread by actual people.

Human executive assistants have been [replying to email](https://fullfocus.co/the-first-task-your-ea-should-tackle/?ref=davidnunez.com) on their bosses' behalf for about as long as bosses have had email and certainly well before bosses could even type. In 2026 the rest of us are supposed to get one also. Y Combinator alone [has funded 147 AI assistant startups](https://www.ycombinator.com/companies/industry/ai-assistant?ref=davidnunez.com), including an "AI Chief of Staff" and an executive assistant that lives in iMessage and drafts your replies.

I assume these bots are already answering each other's cold outreach, keeping humans out of the loop until they are required. This raises the question of what is actually "required" of humans in the modern office, but I'll leave that for a future email.

![](https://storage.ghost.io/c/2a/66/2a66aa35-855e-4718-a6f9-5e4f6e1dc30d/content/images/2026/09/image-4.png)

Out of Office (via [Wiki Media](https://commons.wikimedia.org/wiki/File:IcelandicHorseInWinter.jpg?ref=davidnunez.com))

In the meantime, the horses are eating their hay and ignoring their inboxes.

/ David

**Listening:** ["A Horse with No Name"](https://en.wikipedia.org/wiki/A%5FHorse%5Fwith%5FNo%5FName?ref=davidnunez.com), America, 1972\. For the record, some horses have email aliases.