Larry Hosken. Coder. Puzzlehunt enthusiast.
331 stories
·
4 followers

AI, My Ass

1 Share

A month or so ago, my wife and I went to Las Vegas for a long weekend.

Valentine’s Day is February 14 and her birthday is February 18, and I learned early on in our relationship that you do not try to combine them. But over the many (many) long (long) years, we’ve mellowed to the point of:

“Are we going out on Valentine’s Day?”
“What, are you nuts?”

…and:

“What do you want for your birthday?”
“Sleep.”

And so with the President’s Day weekend falling between the two, it seemed like the perfect opportunity to not go out on Valentine’s and also have a big bed with black-out curtains. There are also buffets, which is my love-language.

The hotel room was nice — other than the fact that it had Las Vegas right outside the door — and, when I wandered into the bathroom and saw the bidet, I made a sound I haven’t been able to reproduce. Heated seat, glowing blue light, and the kind of steely-eyed menace you get from people who have a job to do and know what’s involved in getting it done. My undercarriage is gettin’ pressure-washed!

Problem, though: the bidet has no instructions. At all. Normally this wouldn’t dissuade me from operating machinery, but here, um, I have some skin in the game. It can’t be that complicated, though, right? You could accomplish the same thing with a garden hose and some privacy. And, hey, there’s a language-agnostic remote control mounted on the wall.

A grey remote control, mounted to a marble-tiled wall.

OK, starting from the top. That’s clearly the universal icon for an ass, and it’s got a fountain pointed at it. Great. The basics. To the right appears to be a more… aggressive… flow.

Below that is… a woman? And a spray that covers everything from the back of her calves to half a foot behind her back. And to the right, an even wider delivery angle. I guess that’s the all-orifice hose-down option.

Next is either a button to spring Wolverine claws, or the dryer. Since I didn’t have to sign a liability waiver before sitting down, I’m going with dryer.

Fourth seems to be an adjustment of the… targeting arm? There is a little hatch in the back of the toilet bowl where a nozzle pokes out of when called upon, like the gatekeeper droid at Jabba’s palace. This button must move it back and forth, like an intimate, adjustable pop-up sprinkler. Below that is what I’m going to assume is the spray pressure control, and not a cigarette to smoke after you’ve had your full bidet experience.

Finally, last, there’s the same nozzle icon used previously, but it’s got… sparkles above it? Where the water normally goes? Maybe it uses carbonated water? Or, I dunno, coffee? Does the nozzle get replaced with a magic wand? And is that licensed from Hitachi? Maybe the cigarette idea is right.

But wait. I’ve seen that before. That sparkle iconography has become really common — it’s the generic graphic that literally every company on earth is using to indicate AI. Is this an AI bidet? What does that even mean? Is this how AI uses all that water? What was the training data? I shudder to think whose intellectual properly rights I might be violating.

I push it. Nothing happens.

So it is AI.

Below that is apparently the volume control.

Read the whole story
lahosken
19 days ago
reply
San Francisco, USA
Share this story
Delete

The Omniscience Expectation and the Mardenfeld

1 Comment

For many leaders the hardest job they have is getting comfortable with not knowing. It is natural to feel like you have to understand everything about the area that you lead. And that’s a feeling that often cascades down through hierarchies. My boss expects me to be able to answer an arbitrary question on the spot, in order to accomplish that I need to be an expert on an increasingly large number of topics. I accomplish this by asking for more and more detailed information from my team, perpetuating this omniscience expectation.

There are two obvious problems with the omniscience expectation (and one non-obvious problem).

Obvious Problem 1: It’s an Asymmetrical Standard. It is always trivial to come up with a new question that hasn’t been anticipated. There is no amount of information you can know to avoid being asked a question you don’t know the answer to.

Obvious Problem 2: It Undermines Psychological Safety. The omniscience expectation breaks down psychological safety as it cascades a profound anxiety through your leaders. To know everything you must be conservative, things must be smaller and tightly scoped, able to be captured in exact detail. (this is also the PGM fallacy. If we could write down in exact language the problem we were trying to solve we would have a computer program)

Non-Obvious Problem: It Focuses Leaders on Trivia. The non-obvious problem with the omniscience expectation is it focuses your leaders on trivia. I don’t know the attach rate of our enterprise partnerships, and I don’t need to know the attach rate. The question isn’t “What’s the attach rate today?”—it’s “Has churn decreased thanks to our new partnership efforts?” If it hasn’t, we persist. If it has, we celebrate and double down. The strategy tells me the answer, not the trivia.

Two (of N) Possible Responses

Two possible approaches to disrupting the cycle that lead to the omniscience expectation and the subsequent anxiety it creates.

Approach 1: Embrace “I Don’t Know”

The first is the hardest: You say, “I don’t know. Here is what I do know ….”. This is a bit different than the standard advice (which also takes courage) on how to say I don’t know which is to say, “I don’t know, I’ll find out, and get back to you on this date.” (closed loop)

Approach 2: The “Mardenfeld” Maneuver

This is an approach to separate a true need for data from “anxiety-driven requests.” And the great thing is that it is plausibly deniable.

Steve Mardenfeld has gone on to have an illustrious career, but once upon a time he was an intern with a sociology degree and a talent for understanding data who ended up building critical aspects of the data and experimentation platform at Etsy. And he had a practice. If you asked him for data he’d share it with you, e.g. in a Google Sheet, and not give you access to read it. If you requested access, great, he would give you access. If you didn’t, then he knew you didn’t really need this information. This was just unmanaged anxiety cascading down the hierarchy. Because the reality is no one is closing the loop on those data requests. I can’t possibly know everything I’m asked, and neither can you, and neither can the person who asked you. So we ask, and then some other request comes in, and we ask about that, and we never loop back for those answers.

And who knows if you forgot to share the permissions because the Google Docs UI is obtuse, or if you shared the permissions, but Sharepoint just forgot like it does 50% of the time, or what.

Challenging the omniscience expectation is important as a leader. You hold space for yourself to be a strategic actor, while protecting the psychological safety of your team. Admitting uncertainity and filtering unneccessary requests aren’t the signs of weakness so many fear – they are leadership tools that allow real creativity, collaboration and growth.

Read the whole story
lahosken
23 days ago
reply
"Steve Mardenfeld... had a practice. If you asked him for data he’d share it with you, e.g. in a Google Sheet, and not give you access to read it. If you requested access, great, he would give you access. If you didn’t, then he knew you didn’t really need this information."
San Francisco, USA
Share this story
Delete

Discover related stories and sites

6 Shares

I want to introduce you to the new Discover Stories and Discover Sites features. Sometimes you’re reading a story and want to know everything there is to know about that topic. You want other stories, but depending on the topic, you might want them from the same site, from similar sites, or from all of your subscriptions. That’s the new Discover Stories feature, and it’s only for NewsBlur Premium Archive subscribers. The Premium Archive subscription is meant for this use case of being able to peer deeply into your story archive and not just what’s been published in the last month.

Second I’m introducing Discover Sites, which is available at the top of every feed and folder to everybody, both free and premium users. Having tried all of the competing discover sites features, I built the popover dialog that has all the features I wanted. It’s an infinite scroll of related sites, showing the most recent five stories, formatted exactly as your story titles are personally styled. You can read stories from unsubscribed feeds and easily subscribe to them while scrolling through the discover stories dialog.

Here’s a set of features I’ve been wanting to build since the very first days of NewsBlur in 2009. I built prototypes of this feature using a few of the modern text tools at the time: nltk (the natural language toolkit), support vector machines, and LDA (Latent Dirichlet Allocation) to group stories by topic. It didn’t work, or it was too slow, and even then not accurate enough. I read the tea leaves and could tell a better tool would come out eventually that was basically a drop-in classifier and topic grouper. Out came word embeddings (word2vec initially, then sentence transformers). And now those transformers are available basically for free.

As you can see, this isn’t your normal related stories feature. It shows all of the related stories, segmented by the folders that a site is a part of. This folder control allows you to filter down to an individual site and up to every feed you subscribe to when finding related stories.

And it’s important to note that none of the data presented in the Discover Stories or Discover Sites dialog is trained on your personal data, like feeds that other people subscribe to in relation to any particular site. All of the data is extracted and grouped by the content of the RSS feed’s title, description, and the titles of the first few stories.

Above we see that Discover Sites is right on the money. An infinite scroll of related sites, showing story previews, and multiple interaction points that let you choose between trying out a site by reading one of the stories, adding it directly to a folder, or checking the statistics of the site. The stats dialog is great in this case because it gives you a feel for what other people like and dislike about the site.

I’m super proud of this release; it took years to build and a decade to plan. And while the Discover Stories feature is technically only available to Premium Archive subscribers, you can see related stories if another Premium Archive subscriber is subscribed to that site. I don’t think hiding those stories from free and premium users is worthwhile.

Please post your feedback on the NewsBlur forum, ideally as an “idea,” but you know I love responding to all feedback. For every person who writes up their thoughts on the forum, there are ten people who are thinking the same thing, so it’s worthwhile to hear from you, knowing the multiplier it represents.

Read the whole story
lahosken
57 days ago
reply
San Francisco, USA
Share this story
Delete

I've left Google

1 Comment

My nearly 2 decades at Google as its Global Privacy Counsel has ended.  I’ve left Google as one of the last few remaining members of the original early Google team.  Google asked me to update social media profiles accordingly, hence my coming back to this dormant blog to say I’ve left Google.  Together with me, other senior members of the Google privacy team have left in recent months.

My career started as Google’s first full-time privacy professional, building a function, and later team, that didn’t exist before.  My job was to try to make Google respect privacy for its billions of users.  You can judge the results, but I am proud of the mission.  Being a privacy leader is a tough job at a company like that.  


The early years at Cool Google were fun, creative, innovative, comradely, and I loved them.  But Google has changed and evolved into Corporate Google, and large committees can now carry forward the work I did, or reverse them; in either case, it’s no longer my business.  


I left on good terms.  No one is in jail now for privacy, including me, and I’m hardly being flippant, speaking as one of those rare privacy professionals who was arrested and sentenced to jail for their employer’s privacy practices.  And I helped build the small company I joined into the largest private processor and monetizer of personal data on the planet. I can’t think of another privacy professional who helped build their data-processing company from the early days to 2 trillion + market cap.  What a ride.  


I will remain active in the field of privacy in many ways.  AI will present existential challenges to the field of privacy, as to so many other domains, and I’m eager to find ways to help organizations develop AI responsibly.  And there are innovators out there who remind me of the fun, creative, responsible environment of my early years at Google, as we wrestled with privacy issues and the then-new online world.  Unless compelled by law to testify, I won’t reveal any non-public information about Google:  I’ll respect my confidentiality constraints as a lawyer to my former client/employer. 


I relish my newfound freedom to share my insights and experience with others in the field and in new ways.  More on that soon.  In the meantime, I wish luck to my former colleagues with MAGA:  Make Alphabet Great Again. 


Read the whole story
lahosken
70 days ago
reply
"AI will present existential challenges to the field of privacy"
San Francisco, USA
Share this story
Delete

Turning off AI in Google search

1 Comment

a robot with a crossed out symbol superimposed

If you don't like how much slower Google search is with AI results, find them useless most of the time, and want to limit the climate damage potential of AI, here’s how to use Google as your primary search engine in Chrome with AI turned off:

  1. Click this link chrome://settings/searchEngines to go the search engine settings page.

  2. Click the Add button.

  3. Enter these values:
    • Google (no AI)
    • withoutai
    • https://www.google.com/search?q=%s+-art1f1c141 

  4. Click Save.

  5. Find your new no AI search engine in the list and click the  icon and choose Make default.
So what's this actually doing? The -art1f1c141 tells Google to not include search results that include that (non) word and this has the side effect of suppressing AI results and also certain other instant results. I've chosen an arbitrary string that (as of the time I wrote this) does not appear in any Google or Bing search results. So excluding that string won't affect the search.

Searching for "time in Antarctica" would normally show the time directly, but with the negative clause it won't. To get that result, just remove the -art1f1c141 in the query. To make that a bit easier, edit the Google search engine and change the shortcut to just g. Then to search google without disabling AI just type g and a space.

These same instructions work in Edge except start at edge://settings/searchEngines in step 1.


Read the whole story
lahosken
142 days ago
reply
feels less brittle than the use-web-results-tab thingy I was doing before
San Francisco, USA
Share this story
Delete

https://mltshp.com/p/1QG8V “always be aware of your ad’s surroundings”

1 Share

mltshp.com/p/1QG8V “always be aware of your ad’s surroundings”



Read the whole story
lahosken
168 days ago
reply
San Francisco, USA
Share this story
Delete
Next Page of Stories