Simon Tong

[CrazyShyyt @CrazyShyyt]

Researchers proved you can predict actual purchase intent at 90% accuracy by asking LLMs to roleplay customers.

if you just ask an LLM directly: "rate this product 1 to 5," it fails completely. it gives garbage distributions because models love defaulting to safe, middle-of-the-road answers.

to fix this, the researchers developed a method called Semantic Similarity Rating (SSR):

- The Persona Drop: You condition the LLM with a specific demographic profile (age, income, background).

- Free-Text Impressions: Instead of a number, you make the AI write natural, qualitative impressions of the product first.

- Embedding Mapping: You map those text responses to a 1–5 Likert scale using embedding cosine similarity against reference anchor statements.

They tested this across 57 real consumer surveys with 9,300 actual human responses from a major corporation.

The results are insane:

- Achieved 90% of human test-retest reliability.
- Maintained near-perfect distribution matching (KS similarity > 0.85).
- Captured subtle demographic nuances like age and income behavior patterns better than traditional models.

the implications are massive. you can now A/B test 1,000 product concepts overnight, simulate market reactions before manufacturing a single item, and get rich qualitative feedback explaining why people would buy it, all for pennies.

the synthetic consumer era is officially here.

[Simon Tong @simon_tong]

@mattpocockuk Spec yourself before you reck yaself

[Ryan Carson @ryancarson]

Woke up to 690,000+ views on my Ralph article (what?)

So I'm going to double down and show you a real-world example of how I used Ralph to ship a feature this morning.

Also, I created a public repo that makes it super easy for you to get Ralph running: https://t.co/V2Qu7SMyxh

1. Created the PRD
https://t.co/SURYq8Xizc

Note how I used my create-prd markdown file from https://t.co/YXuNG6XKYz

2. Created the Ralph user stories
https://t.co/5Q7F6xLjUh

Note how I used my Ralph Skill to create the user stories
https://t.co/xdw8ur90m6

3. Started Ralph with max 25 iterations

./scripts/ralph/ralph.sh 25

Iteration 1
https://t.co/l38qgqyQeG

Iteration 2
https://t.co/xj05rXNqDO

Iteration 3
https://t.co/zOknKrGE4O

Iteration 4
https://t.co/vId1EtNGST

Iteration 5
https://t.co/KfGzjNq9AR

Iteration 6
https://t.co/fyMA7fnix9

Iteration 7
https://t.co/HGmrhbiaUc

Iteration 8
https://t.co/V4iww9gNUQ

Iteration 9
https://t.co/4eKssSyTW3

Iteration 10
https://t.co/GdIEliVL28

Iteration 11
https://t.co/jCOjNDbMdB

Iteration 12
https://t.co/a2I1g1EXwP

Iteration 13
https://t.co/xhQTR3R1lg

Iteration 14
https://t.co/F2TNGDxvao

After it finished, I tested manually and found a few edge-case bugs, which Amp quickly fixed.

[Simon Tong @simon_tong]

@Zephyr_hg News

[DHH @dhh]

The quickest way to ruin the productivity of a small company is to have it adopt the practices of a large company.

[Simon Tong @simon_tong]

https://t.co/qvuWh8qwjt

[Simon Tong @simon_tong]

@timClicks @rustlang Heyb Tim, where is this talk gonna air?

[Simon Tong @simon_tong]

Working my way through this brick - https://t.co/6vrrQ06x22 - highly recommend as a next read after "The Rust Book". Rust features and concepts are broken down into an easy to digest format. Pleasure to read!

[Simon Tong @simon_tong]

@jimblandy Hey bud, currently studying Rust and wanted to know if there is a second edition of your book in the works? If not, is the first one still relevant with all the new Rust updates?

[Simon Tong @simon_tong]

@Truth_Investor Into it!

[Simon Tong @simon_tong]

Anyone out there seeking a social media coordinator / marketer for their biz? PM me #socialmarketing #fashion

[Simon Tong @simon_tong]

@b0rk hey just trying to read your latest article on IP address routing but it appears to be sending me raw html rather than rendering the page

[Simon Tong @simon_tong]

https://t.co/yRCEiGyLwS

[Simon Tong @simon_tong]

@Logitech where can I get my hands on a K810 UK layout in 2018?

[Simon Tong @simon_tong]

I see. Any chance of opening up the API to the public? https://t.co/i3SkxXe5Pu

[Strong - Workout Tracker 🏋️ @strongapp]

@simon_tong sorry for the delayed response on this Simon, but to answer your question -- not at the meantime as we plan on tightening our integrations with Apple Health and Google Fit. Sorry about that!

[Simon Tong @simon_tong]

@GDAX Trying to register as an institute and getting 'Internal Server Error' on the list beneficial owners page.

[Simon Tong @simon_tong]

@strongapp Love the app. Any future plans for integration with Samsung Health?

[Bill Gross @Bill_Gross]

The reason it's called a "patch" in software is because you could make small corrections in a programmed sequence by "patching over" the paper tape and re-punching holes in that section..

[Axel Rauschmayer (also on 🦣) @rauschma]

The new homepage for my JavaScript books is online. All of them are free to read online!
https://t.co/nI0meq4N2R

[Simon Tong @simon_tong]

@sainsburys Hi, they have already gone in the bin. I'll go tommorrow again and take a picture. Its the sainsburys on Westminster Bridge Rd