This is an R project that aims to verify the probability math behind the identical ballots evidence in Thailand's 2024 Senate collusion case (คดีฮั้ว ส.ว.), and directly rebuts the it's technically possible, so it could be coincidence defense.
โปรเจคนี้สร้างขึ้นเพื่อตรวจสอบความถูกต้องของการคำนวณความน่าจะเป็น เกี่ยวกับข้อกล่าวหาเรื่องบัตรลงคะแนนที่มีรายชื่อและลำดับตรงกันในคดีฮั้ว ส.ว. ปี 2567 รวมถึงวิเคราะห์ข้อโต้แย้งที่ว่า "แม้มันจะเกิดขึ้นไม่ถึงหนึ่งเปอร์เซ็นต์ มันก็มีโอกาสเกิดขึ้นและก็เป็นไปได้เหมือนกัน"
Start here: report.html
Requires R with ggplot2 and rmarkdown:
Rscript R/run_all.R
Rscript -e "rmarkdown::render('report.Rmd')"Does/สิ่งที่สามารถยืนยันได้:
-
Independently re-derives the permutation probability arithmetic behind the public claim that two ballots matching by chance (out of 154 candidates, 10 ranked picks) has probability ~1 in 5.567 × 10²¹, and that 10 identical ballots compounds to ~1 in 5 × 10¹⁹⁵.
-
Validates that method via Monte Carlo simulation.
-
Places those numbers alongside probability thresholds already treated as conclusive in science, forensics, and law (DNA matching, p-values, six-sigma), to show that it's not literally impossible is not a meaningful evidentiary standard at this scale.
-
Simulates (does NOT reproduce real data) why a step pattern in vote scores would indicate coordinated bloc voting.
-
คำนวณความน่าจะเป็นของการจัดเรียง (permutation) ขึ้นใหม่ เพื่อตรวจสอบตัวเลขที่มีการเผยแพร่ต่อสาธารณะว่า หากมีผู้สมัคร 154 คน และผู้ลงคะแนนเลือกผู้สมัคร 10 อันดับ บัตรลงคะแนน 2 ใบที่มีรายชื่อและลำดับตรงกันทุกประการจะเกิดขึ้นโดยได้ด้วยความน่าจะเป็นประมาณ 1 ใน 5.567 × 10²¹ และหากพิจารณาบัตรที่ตรงกันทั้งหมด 10 ใบ ความน่าจะเป็นสะสมจะอยู่ที่ประมาณ 1 ใน 5 × 10¹⁹⁵
-
ตรวจสอบความถูกต้องของวิธีคำนวณด้วยการจำลองแบบมอนติคาร์โล (Monte Carlo)
-
เปรียบเทียบค่าความน่าจะเป็นดังกล่าวกับเกณฑ์ที่ใช้กันในสาขาอื่น เช่น การจับคู่ดีเอ็นเอ การทดสอบสมมติฐานทางสถิติ (p-value) และมาตรฐานซิกซ์ซิกมา เพื่อแสดงให้เห็นว่าคำอ้างว่าสามารถเกิดขึ้นได้ ไม่ใช่เกณฑ์ที่ใช้ประเมินหลักฐานในระดับความน่าจะเป็นในกรณีนี้
-
จำลองสถานการณ์ (โดยไม่ได้ใช้ข้อมูลการลงคะแนนจริง) เพื่ออธิบายว่าทำไมรูปแบบคะแนนแบบขั้นบันไดจึงสอดคล้อง กับการลงคะแนนเป็นบล็อกมากกว่าการลงคะแนนอย่างอิสระ
Does not/สิ่งที่ไม่สามารถยืนยันได้:
-
Verify the underlying claim that 10 ballots actually matched, that's sourced from public reporting (Loy Chunpongtong's public analysis, reported by Posttoday, Feb 28 2025), and not primary EC/DSI data we independently confirmed.
-
Prove criminal intent, money exchange, or anyone's guilt.
-
Claim the simulated score-distribution pattern (Section 5) matches the real 2024 vote data exactly.
-
ตรวจสอบว่าบัตรลงคะแนน 10 ใบที่มีรายชื่อและลำดับตรงกันนั้นมีอยู่จริงหรือไม่ ข้อมูลดังกล่าวอ้างอิงจากรายงานและการวิเคราะห์ที่เผยแพร่ต่อสาธารณะ (การวิเคราะห์ของอาจารย์ลอย ชุนพงษ์ทอง และรายงานของโพสต์ทูเดย์ วันที่ 28 กุมภาพันธ์ 2568) ไม่ใช่ข้อมูลดิบจาก กกต. หรือดีเอสไอ
-
พิสูจน์ว่ามีการสมคบคิด การแลกเปลี่ยนผลประโยชน์ หรือความผิดทางกฎหมายของบุคคลใด
-
ยืนยันว่ารูปแบบการกระจายคะแนนที่ใช้ในการจำลองตรงกับผลการลงคะแนนจริงของปี 2567 เนื่องจากเป็นเพียงตัวอย่างที่ใช้ประกอบการอธิบายหลักการเท่านั้น
- Loy Chunpongtong (ลอย ชุนพงษ์ทอง), public YouTube analysis via Loy Academy, reported in Posttoday, Feb 28, 2025: https://www.posttoday.com/politics/domestic/720251
- Nation TV, June 18, 2026, on the ballot matching anomaly and the distinction between "โพยกันลืม" and "โพยจัดตั้ง": https://www.nationtv.tv/exclusive/special-report/378978941
- THE STANDARD, July 29, 2026, "เจาะเส้นเงินเลือก สว." includes the mathematical evidence cited in the DSI money laundering referral (1 in 5 quintillion figure) and the accused-side legal defense: https://thestandard.co/senate-election-money-trail/