What to do about AI Nudes?

I recently listed to this podcast from the Wall Streeet Journal, Teens Are Falling Victim to AI Fake Nudes: “Last fall, nude photos of a 14-year-old student started spreading around her high school. But they weren’t real… they’d been created with AI. WSJ’s Julie Jargon breaks down how fake photos like these are a growing trend among teens and why it’s difficult to deal with.”

This part of the ordeal stood out for me:

Elliston Berry: I did. I had a couple classes with him in eighth grade, but he was a classmate and we were mutuals on social media. I didn’t see him as a threat. It was really shocking knowing that he did this because he was a peer, he was a classmate.

Jessica Mendoza: The school district declined to comment about the enrollment status of the teenager who made the photos. The local district attorney said he couldn’t give specifics because the student was a minor, but that the boy was sanctioned within the juvenile justice system. Meanwhile, Elliston’s mom Anna decided to take some steps on her own. Could you walk me through your decision to come forward publicly and talk about this?

The offending student was ultimately suspended but his identity was not revealed to the victims. A big part of the WSJ story is what do to in these cases? It’s clearly wrong, but what does it fit under? I assume schools are afraid of legal liability coming from both sides.

In 2017, I wrote a blog post here during the Me Too environment. It talked about something that I did in my high school days that wasn’t meant to hurt anyone. It was more about wanting to get attention. I don’t know if this is the case for the offender, but I can imagine it starting off that way.

And so when everything is preoccupied about legal liability or political correctness, I think we lose sight of what could generally be considered “right.” The offender invaded the girls’ privacy in a way by creating AI nudes and yet his own identity is protected by the “system”. There isn’t this ability to have the person meet the girls, learn about the pain he causes and apologize (if he felt it was appropriate) or explain himself. I feel a big part of navigating life is understanding the real impact on people. If you fire someone, don’t source it to HR. If you break up with someone, don’t text and ghost them. Take the responsibility, as painful as it is. In today’s world, it feels quite easy to do whatever you want and ignore that responsibility. I can send a Tweet threatening to kill someone for example, or say all sorts of things online that I would never do in front of that person. I don’t learn to understand the impact of those actions on the person receiving them.

Just because the offender in the case was suspended doesn’t mean they really get it. And isn’t the purpose of punishment to help us improve?

Here’s an excerpt from that blog post I wrote years ago, which is no longer public:

I have my own story of guilt: 20 years ago in high school, I was going online during the starting days of the internet, the time when people started transitioning away from AOL and finding the internet: Yahoo, Google, Excite. E-Mail was a miracle. This was Web 1.0, the early days of Geocities, where anyone could create his own website and show off a cool animated “hits” counter to show off the number of page visits.

As many teenage boys do, I gawked at the attractive girls in my classes. I had an idea of naming the most attractive girls in my year and putting it on a website for everyone to see.

I learned basic HTML, scanned photos from my yearbook and put it up. Scraping emails from group threads among classmates, I executed my first example of spam / mass emailing / grassroots marketing. I didn’t ask permission from any of the girls, being scared of them (never had I talked to most of them) and their possible responses, yet still wanted the site out there, getting attention.

As that was, perhaps it wasn’t so bad and I started to get some hits. I can’t remember all the details at this point, but I’m sure I continued working on the site, seeing how I could adjust things. Some friends of mine gave feedback, and this is where I made the wrong decision.

A friend mentioned (paraphrased) that one girl had a physical feature that would be great for oral sex. I don’t think I really knew what that meant, but I thought, hey, more content, and something to put on the website.

And so I did. I remember showing the website to my dad at this point, and he made a very astute comment (as usual) that I should reconsider putting comments like that on the website.

But I wanted the attention, I wanted people to visit the site and have a reaction (even though I truly feared a negative one), so I kept it. A few days later, some anonymous students had a negative reaction to the site and I was actually harassed (pretty stressful for my teenage self) for it. I do not know if they were doing so in defense of one of the girls, or some other reason. It does not matter.

I clearly put my need for attention over something that was not nice to put in full public view. It could have affected one of the girls negatively, and it was a form of harassment.

As I reflect upon this over 25 years later, I could say putting yearbook photos on a website isn’t that big of a deal compared to creating AI nudes. One is porn! Attached to a 14 year old. But we live in different times. In my day, internet porn wasn’t readily available. Sex and violence in film and TV was extremely tame compared to today’s standards. I’m not saying that AI nudes are ok, I’m saying that the thresholds of acceptability are moving extremely fast. If you can download an app to make a nude, it’s suggesting that these are becoming everyday things. And I think there has to be some context around that.

If I were in high school today, would the equivalent of scanning yearbook photos and making a basic website of them in the past be the equivalent of today’s AI nudes? I hope not, and maybe it’s not the equivalent today, but what if it is in 5 years or 10 years?

I Finally Won – “Cheating” My Way to a Zwift Victory

This morning, I won my first Zwift race! And to be honest, it wasn’t really close.

After more than 5 years of fluctuating between 3.5 w/kg and 4 w/kg, and never getting close to a B class win on Zwift, I’ll take it. Normally, winning a lower class would be called sandbagging. This is slightly different, I think. I’ve been off Zwift for close to 4 months and have been recently recovering from painful cycling injury.

Since my layoff, Zwift released its new Zwift Racing Score and reset its recommendations for riders. Or at least reset them for me. When I got back on Zwift last week, I got recommended E class. So I took it. I didn’t even podium that one. If you compare my performance in the first race to today, I worked harder (higher heart rate) in the first race, but did 13% more power (this is a lot in cycling) this morning – I was still a bit weak and a bit sick then. I didn’t even win the second race I did, in which I got recommended class D. I finished fourth – my indoor trainer went crazy and I had no power data for 15-30 seconds in the last few minutes of the race. I had expected to podium, but not win – I made the wrong call not to follow someone and my group wouldn’t work to catch up to that breakaway rider.

So again, it’s not like I’ve been destroying the competition taking easy wins.

Today though, I was feeling reasonably decent and when a gap opened up even though I wasn’t working at my limits, I decided to keep pushing and try to stay away from everyone else. It worked and the result was a 45 second victory in the field of 18, pushing about 3.6w/kg in Normalized Power (NP) for half an hour.

Zwift isn’t going to let me compete in C again, so at least I got one victory to show I don’t belong in C.

The Cost of the San Francisco Giants Acquiring Vladimir Guerrero, Jr

As I write this, the MLB trade deadline is tomorrow evening. The Giants stand at 2 games below .500. They’ve waited a long time for their starting pitching to come off injuries, and over the past two weeks, we’re starting to see how impactful former Cy Young winners Blake Snell and Robbie Ray can be. There are lot of rumors that the Giants will trade Snell, but as a fan, I think if you get a true superstar to galvanize the fan base and go for winning without taking two steps back as a team, you go for it.

I don’t typically think of MLB trades, but it’s so much more complicated to understand the dynamics of near and long term planning and salary management in baseball. But I’ve looked at past comparable trades for superstars including the Juan Soto trade to the Pirates to propose this package to the Blue Jays for Guerrero, Jr.

  • Marco Luciano or Bryce Eldridge
  • Carson Whisenhunt or Hayden Birdsong
  • Camilo Doval
  • Lamont Wade Jr
  • One other prospect from the top 5-15 list, Blue Jays choice.

In summary, the Giants would trade their #1 hitting prospect, their #1 pitching prospect, their (2023) All Star closer who has 3 more years of team control, and starting first baseman who has another year of team control, and one more top 15 prospect for one of the most popular players in the game who has one more year of team control. Birdsong has played pretty well in the majors so far, and Whisenhunt and Luciano are both at AAA. Eldridge may reach AA this year, so this is a good assortment of legit major league players and almost ready players.

Wade is a legit good MLB hitter, 20% and 40% better than the average hitter the last two years in OPS+. His main issue is injuries. He’s 30 this year but still improving. Guerrero replaces him directly. Doval just turned 27, throws 100MPH. His control can be his downfall, and I think the Rogers twins could replace him ably as the Giants closers. Someone like Reggie Crawford could become the Giants’ next Doval. The Blue Jays have their own All Star closer who has been injured most of the year and could keep both or trade one.

The Blue Jays only do this trade if they’re non-committal about the massive contract decision.

The Giants could then offer Guerrero 15 years at a backloaded $460M (and replacing the last arbitration year), the second largest contract in MLB history, $100M more than the current second place, Aaron Judge. Guerrero is only 25 years old and while not a good fielder (he does have a Gold Glove), this contract takes him to his 40th birthday. He’s only 6 months older than my crush Heliot Ramos, and is already a 4 time All Star. While $460M is no small sum, the average yearly value is just above $31M. 10 years ago, the most valuable per year contract was Ryan Howard at $24M. This year, 4 players including Judge, but not Ohtani since his payments are deferred, are making at least $40M per year. That’s over a 50% increase.

This is all to say, $30M in ten to fifteen years is likely going to feel more like today’s $20M towards the end of the contract with inflation and the expansion of salaries.

A Guide to Selling Your Car through Daniel Kahneman’s “Thinking Fast and Slow” Methodology

I recently listed my 2020 Tesla for sale and sold it in 1.5 weeks. Autotrader (or Cars.com?) predicted I would sell it in a bit over 3 weeks considering market demand and my pricing.

An overview of my preparation:

  • I learned that demand for the Tesla Y right now is quite soft, and my research suggested that the overall demand for EVs is not strong.
  • I researched third party pricing through KBB and CarGurus and looked at competing listings both locally and nationally across all sites.
  • I listed on Cars.com, Autotrader ($49), Facebook, OfferUp, and Craigslist ($5). CarGurus disabled their private used car listings a few days before I listed.
  • I looked into sell-direct marketplaces and dealership offers and those offers ranged from $21 to $24K.
  • I took over 30 photos after the car was detailed and shared the maximum allowed for each platform. My approach was, no surprises, you should see everything that you might note in person. Don’t wate your time, don’t waste my time. The lighting was a bit darker than I wanted – I was going for the golden hour, but everywhere in the area around me ended up being shaded at that time.
  • I decided to use KeySavvy to act as an escrow service protecting both sides. I really liked that I wouldn’t need to handle securing payment and because I had a car loan obligation, I didn’t want to create complications for the buyer.
  • My friend Sean suggested I consider using the recently departed Daniel Kahneman’s Thinking Fast and Slow in considering how to write my sales ad. As Wikipedia describes, the “book’s main thesis is a differentiation between two modes of thought: “System 1” is fast, instinctive and emotional; “System 2” is slower, more deliberative, and more logical.”

As a buyer, I want System 2, to read and research carefully and to make up my mind even before seeing the car. I have bought 3 cars and 2 houses this way. However, Sean’s comment was more about, most people are more likely to be System 1 in how they respond to your ad. I ended up doing both. I’ll let you see both formats and then share more notes about the overall experience that closed with the sale.

System 1 Ad Format (the quick reacting, knee-jerk TLDR person)

2020 Tesla Model Y Long Range AWD | Tow Package + More | 51K Miles

Price:

  • $27,750 (FIRM)
  • “Great Deal” rating by CarGurus.
  • KBB Value: $26,721 – $31,228.

Key Features:

  • Tow Package, Homelink, Autopilot (Not FSD)
  • Panoramic Glass Roof, Heated Seats, Premium Audio
  • 19″ Gemini Wheels

Condition:

  • 50,894 miles (80% highway) under one owner (me)
  • Clean title, no reported accidents
  • Light paint wear, small scuffs, light curb rash on wheels (see photos)

Extras:

  • Tinted Windows, Charging Cable, Extra Floor Mats, and more
  • Carpool Stickers transferable in CA

Warranty:

  • Battery & Drive Unit until June 2028 / 120K miles

Purchase Process:

  • Secure escrowed transaction via KeySavvy ($99 fee for buyer)

Maintenance Summary:

  • Regular maintenance, new tires recently installed.
  • Complete maintenance records and Carfax available for review.

The System 2 Ad Format (what I wish someone would prepare for me)

2020 Tesla Model Y Long Range AWD | Tow Package + More | 51K Miles

I’m selling my 2020 Tesla Model Y Long Range All Wheel Drive as the original owner. The car is in very good condition with 50,894 (80% highway) miles and current range of 291 miles.

The vehicle has a clean title. Carfax report and all repair/maintenance records are available for review. CA DMV registration paid for 2024.

Includes the tow package, homelink for automatic garage door opening/closing, and other bonuses. No FSD.

Asking $27,750.

Price rated as “Great Deal” by CarGurus Car Values. Kelly Blue Book Value Range is $26,721 – $31,228.

VIN: [REMOVED].

Purchase Process:
We will use KeySavvy for the transaction. KeySavvy acts as the escrow, ensuring a safe and easy process. Buyer pays a $99 fee; I will pay all other fees. [LINK]

Recent Maintenance:
-3,500 miles on Michelin Pilot Sport EV tires (20K warranty)
-3,000 miles since the last maintenance visit

Warranty:
Battery & Drive Unit Limited Warranties expire June 28, 2028, or 120,000 miles. Basic Vehicle Limited Warranty has expired.

Vehicle Details and Key Features:
-Tow Package ($1,500)
-Homelink – Automatic Garage Door Opener ($350)
-Exterior Color: Pearl White Multi-Coat
-Interior Color: All Black Premium Interior
-Autopilot (Not FSD)
-19’’ Gemini Light Wheels
-5-seat interior
-Panoramic Glass Roof
-Heated Seats
-Premium Audio System

Condition:
-Light paint wear, various small scuffs and scrapes. See up close photos of driver side door (1”), driver side rear bumper (4”) areas, passenger side rear wheel (.2”).
-Light curb rash on wheels.
-Repaired driver side damage from a 2022 incident – work done by Tesla and Nationwide-approved shop.
-Windshield replaced in May 2024 by Safelite.

Additional Inclusions:
-2 Key Cards
-Windows: Tinted for heat reflection ($1,200).
-Charging Cable: Official Tesla universal charging cable ($250) with Gen 2 NEMA Adapter ($35).
-More: Various accessories including an extra set of floor mats, new official paint repair kit, new official tire repair kit, official (installed) mud flaps and more.
-Carpool Stickers: Automatically transfers to new CA owner.

Maintenance History: All recalls addressed
-Sep 28, 2023: Wheel alignment, standard maintenance at 47,553 mi
-Aug 22, 2023: 12V battery replaced
-Oct 10, 2022: OCS Filter Module installed
-Sep 22, 2022: Tow package installed, standard maintenance
-Sep 02, 2022: Homelink installed
-Nov 05, 2021: Maintenance items

More photos: [Link]

———————

Here’s the rest of the story:

  • I ended up pricing at $27,750 after some initial thoughts of $28,250. I wanted to be the best price in the market. Was I giving up some money? Should I price higher to have some wiggle room to negotiate down? Perhaps yes to both, but I wanted anyone who was remotely interested in the car to reach out to me first. I wanted my car to be prioritized among all other listings. I didn’t know how long this selling process would take and in a soft market with high interest rates, I didn’t want to be left standing in a game of musical chairs – I had experienced this a couple of years ago in the housing market.
  • Facebook generated, by at least 10X, the most inquiries of all the car marketplaces, but 90% of them were not serious. I knew this would happen beforehand, so I vowed to myself to not get annoyed. If someone gave me the Facebook’s default “is this still available?” message, I would give the thumbs up icon. If someone clearly didn’t read the description, I’d give a short reply and point them back to the description. If someone made a lowball offer, I’d say “no, but thank you”. I treated it as garbage in, garbage out – people who gave me a system 1 response would get one back in return. People who were more detailed and thoughtful got a similar level reply back as well.
  • Which System Ad was more effective? I don’t know. I started off with the System 2 ad, but switched to System 1 the following week. The eventual buyer ended up coming from Cars.com and originally saw the System 2 ad. I liked working with him because he was always respectful and asked good questions.
  • I ended up selling for a few hundred dollars less than asking. I was still confident I could sell it for my asking price, but there’s a real cost of continuing to work on selling the car to new parties. I felt taking the 1.4% discount and giving the buyer a small win was worth sparing myself of future effort, which could easily be tens of hours or more in the following weeks. I did end up liking the System 1 ad more because it reduced cognitive overhead for buyers, even if it then put the responsibility on the buyer to know which questions to ask – if you think that’s immoral, most private ads I saw are ridiculously sparse in information.
  • I got interested buyers from all the platforms, but if I rank the platforms by true leads, they probably all come in tied at 1 or 2 each. That means that if I were to do it again, I’d relist on all of them except perhaps OfferUp. Someone tried to scam me there by offering to take over my payments and told me to call him. At the same time, OfferUp is free. In general, I suggest having a high level of vigilance in terms of junk / scam offers and following my garbage in, garbage out policy of responding.
  • Keysavvy (as the escrow) ended up being very easy to use on both sides, and the buyer can simply pay and take control of the car without delay. It took two business days for my bank loan to be paid off and to get the remaining amount in my bank account, but the car buyer can buy and leave with the car within an hour. I chatted with Keysavvy’s customer support a few times before making the sale, and they were great in the process. I was able to make sure I understood what to do, what would happen, and things happened with no surprises.
  • Right now, buyers in the US with less than $150K annual income can purchase an EV that’s less than $25K in price from a dealer and get $4K back in federal rebates. ($24K car effectively becomes $20K) So many people who messaged me were thinking about this, but they’d approach me with “why don’t you drop your price to below 25K so I can get the rebate”. Their thought was why doesn’t this seller take less money because it’s better for me? For those who were more thoughtful about a creative way to approach this, I reciprocated. Again, garbage in garbage out. Keysavvy acts as the dealer because technically, the buyer buys the car from them, and you sell your car to Keysavvy so anyone using Keysavvy can take advantage of the EV offer.
  • For buyers who immediately asked for a test drive, I added a filtering mechanism. I’m not a private dealer who’s just available for people to come by, test drive the car and lowball me. I felt that if someone were serious, they wouldn’t mind some extra friction, and I was surprised by how many people ghosted as soon as I sent the filtering message below:

Hi [Name], you reached out to me about my 2020 Tesla Y. A few things I would like to check on.

1) I hope you have been able to read through my notes in detail about the car, the purchasing process, etc. Any questions there?

2) Also, are you looking to test drive and finalize the transaction with Keysavvy at the same time if everything is good?

3) For the test drive, I’d like to ask for a copy/photo of the driver’s license and insurance in advance.

4) I’ll be asking you to follow a specific route that should take us around 10-15 minutes, but provide a feel on various road conditions including freeway. I ask that only the one person (driving) join the car for the test drive.

Please help me by confirming agreement with these four items. As a reminder, I am firm on price.

In summary, here’s my advice when selling a car on your own:

  • Know that it’s an ordeal.
  • Garbage in, garbage out. Don’t waste your time explaining or getting exasperated with most people – with just a little practice, you will quickly figure out who is extremely unlikely to buy, regardless of what they claim. You can still be courteous, but not waste your time.
  • Do your research so you know how competitive your car is in the market and have confidence in your research. To the previous point, make buyers do what you want them to (price, buying terms, etc.). Anyone who thinks it’s a hassle will eliminate themselves.

Why do we fall? To learn to pick ourselves up (2024 MCBC Dirt Fondo)

“All Dirt Fondo routes are planned with mountain and gravel bikes in mind. Upright bars and suspension will have advantages on the trails and rutted fire road segments, but a gravel bike is great for the smoother fire road bits and road segments. Either way, be prepared for some challenges along the way and walk tough sections if it looks too difficult – no one will judge you!” 2024 Dirt Fondo in Marin, CA.

Unfortunately, I took a hard fall last Saturday at the Dirt Fondo. I flipped on the bike riding a heavy rock garden-esque descent, landing primarily on my right leg around the knee. If you look at the images in the link, my descent was comparatively much rockier and chopped up – imagine pieces of large rock. I later realized I took some impact on my hands (I assumed I tried to brace myself), right forearm and left leg as well. The shock was immediate. I couldn’t move for a while and my right leg just shook uncontrollably. I had a small 1inch diameter cut below the knee which didn’t look bad but I immediately thought, what if I broke my leg?

I was fortunate that a pair of riders was right behind me and saw the crash. They helped make sure I was ok and went back to the previous rest stop to ask for help before moving on. I waited a bit, but when it didn’t seem like help was coming, I decided to continue downhill to make the next rest stop and get help there. Unfortunately, I never saw another rest stop but I did hear from another rider that I was being searched for and I was able to send an update via a volunteer. I rode through more gravel/dirt/rock sections before taking a detour and finding pavement as I rode through beautiful Sausalito back to the start – another 2 hrs of riding after the fall.

  • Should I have picked a different path? This is probably the critical issue. Going down, the edge of the mountain was on the left. I am guessing a 10 foot wide path total split into two. On the left edge is a loose (smaller) rock path, on the right is a rock garden-ish path. I don’t remember exactly why I took the right, but here are a few notes I’ve reflected uplon. There was someone going up hill on my left. Did I see that person before picking the right side? I’m not sure but that would have been a reason to go right, there wasn’t room. Loose rock is very slippery, and because that was the mountain edge side, I may not have wanted to be struggling with slowing down on slippery rock. As I started descending perhaps I wanted to move to the loose rock, but that would have been very scary to attempt. I kind of knew the rock descent was difficult but the decision would have been to focus on surviving, not attempting a tricky transition.
  • Should I have run different (wider) tires? I ran pretty wide tires for gravel, 42mm S-Works Pathfinders that expanded to 47.2mm on my wheels. MTB tires can be 10-20 mm wider but after I fell, I saw other gravel bike riders (who took the other smaller rock line) on thinner tires.
  • Should I have run lower tire pressure? Anything can be second guessed. I was running 22/23 PSI on my tires, which I feel is pretty low. The pair behind me were running wider MTB tires and suspension bikes. Less air would have meant a softer ride but I don’t think that was the culprit.
  • Should I have gone slower or faster? On unpaved terrain, I have learned that going faster is often better than going slower because speed helps you roll over obstacles better. I don’t think I was going fast at all, most likely less than 8 mph. It’s hard for me to remember because it was a descent, I was for sure braking or in a brake position in the drops and I was in a rocky path in which I was just trying to go over big rocks non-stop. I didn’t want to go too fast because I didn’t have good traction but going too slow would stop the bike because of the rocks.
  • What was my position on the bike? Earlier, I mentioned I flipped over the bike. I don’t think that’s accurate. I didn’t land on my head. I think the bike got stopped on a rock and I had to fall over in the process. I don’t think I was front leaning on the bike, but I’m not sure if I was sitting or in a more crouched/standing position on the saddle.

Sitting down, writing this a couple of days later, I of course want to blame the event a bit. If I’m the only one who fell (I do not know if this is the case), then I have to assume it’s my fault, whether that be due to lack of skill, bad luck, or something else.

As for my injury, I learned I haven’t broken anything, or at least nothing that can be seen on X-Ray. That’s good news so far. The bad news is that I couldn’t walk the day after the event because of the swelling in my leg. Thankfully, my body let me ride and drive home on the event day – I was 1.5 hrs driving away from home.

I feel a lot better two days after the accident (if not a bit tired) and remain hopeful I just need time to heal. I can’t walk unsupported yet.