Aaron Tay✓

Mastodon @aarontay@mastodon.cloud

⚔ Compare ☆ Watch Own this channel?

🔄 Data last refreshed 6 hours ago

🤝

Make a deal with Aaron Tay

🔒 Paid proposals held safely in escrow — released only when the work's approved.

Followers

673

Account age

9 yrs

🧰 Free analysis for Aaron Tay

🕵️ Fake follower check 📊 Engagement rate 💰 What they charge

📊 Post engagement

Avg engagement / post

0.3%

Engagement vs followers

Apr 2017

On Mastodon since

🔥 Top post: A researcher friend released this https://github.com/kennethkhoo · 5 likes + reposts

📊 Activity & format

Posting cadence

85.7 / week

A very active account — posts many times a day.

Content mix

Mostly text

Recent: 12 text · 0 image · 0 video.

Follower / following

3.1×

Follows 216 back. A more reciprocal / networked account.

Recent posts

View on Mastodon ↗

🔥 Top post A researcher friend released this https://github.com/kennethkhoocy/lit-review-orchestrator that searches multiple sources, dedupe and has a cross validation layer that checks against crossref, OpenAlex, semantic scholar. I've run it 3 times but I find it rarely catches anything… ★ 5

Sigh people love to blame AI/LLM for everything. In a piece about how using LLMs for literature search might lead to the Matthew effect & gender bias (highly speculative in my view) a female commenter suggested that she saw the same thing … As of now I am sure are elite people say the 90 percentile in the field are fully capable of noticing errors in even frontier LLMs. But by 2030? If progress continues at same rate? Probably not I saw this guy test it on his thesis from a while back and was shocked to see it spotted a huge error. Mathematicians with field medals are impressed with some of the proofs it comes up with ..and yet we librarians think we know better .. ★ 2 I was reading this editor saying they were using llms to check manuscripts and they found the llms had too high standards and were relentless at spotting even the smallest errors. So they had to adjust it to let through stuff as long as it… Actually I think most people will be similar to me. Eg even a year ago I would notice errors in people's work / stuff i read etc and when I asked a LLM to assess it would spot exactly the same issues , sometimes more. Frontier models are… ★ 1 Honestly the models in the last 6 months are really crazy, I'm not sure if there's any domain where im really clearly better than them. I do know some limited stuff they don't know yet because its not on the web yet but other than that ..… ★ 1 It has also gone way past my ability to improve because i am really not a evidence synthesis specialists and i was sort of testing on gold standard sets of Boolean string but a lot dont look very good to me lol Doing such projects is very humbling. I realise I lack real expertise in both coding AND domain knowledge of evidence synthesis and yet it does way better than people trying to prompt engineer a chatbot. It literally can't hallucinate Mes… ★ 1 My stuff is very amateurish cos its 99% vibe coded and I didn't intend to put it on github at first. But what the hell. Here's another one of mine https://github.com/aarontaycheehsien/pubmed-search-builder super over engineered skill to fo… ★ 2 Its main weakness is its going to be article heavy so i would plugin something to search your institutions catalogue eg https://github.com/aarontaycheehsien/primo-mcp-server or maybe worldcat ★ 2 In the old days when we got some obscure ILL request we didnt instantly yell hallucination, we would probably guess the reference was either incomplete or wrong somehow. These days it must be hallucination

🐘 Community & instance

Home server

mastodon.cloud

A large general-interest Mastodon server. On the fediverse the server a creator picks hints at their community — this one spans all topics.

✅ Link-verified

Verified link

Proved ownership of a website linked on their profile — Mastodon's green-check verification, a real identity signal rather than a paid badge.

On Mastodon since

Apr 2017

An OG fediverse account from Mastodon's early years — a long, established presence.

💡 Facts

🗓️Joined Mastodon in 2017 — 9 years ago.

👁️Averages 2 views per post.

📤Posts about 85.7× per week.

🕵️ Fake follower check

Estimated

52/100

Fair Credibility score

76%

Mixed Real audience

Medium Fake-follower risk

High Data confidence

Est. 76% real, active audience · Medium fake-follower risk.
Low engagement (~0.3% of followers engage each post) — a sign of an inflated or inactive audience.
Verified account.
Established account (9+ years old).

Heuristic estimate from engagement, follower ratios, account age & growth — a screening signal, not a guarantee.

About

Academic Librarian, Singapore. Blogger at musingsaboutlibrarianship.blogspot.com and now aarontay.substack.com

🔀 Audience overlap

Estimated

Estimated shared audience with similar creators — useful for avoiding overlap (or doubling down) when planning a campaign.

AntGuyOnMaui 🏳️‍🌈🐜🐜♾ 18%

Sondaggi BiDiMedia :verified: 17%

📷Sayuri.H📱 16%

lobo :nix_snowflake_logo: 14%

Mulit-em-dash-ate 13%

Orkan der Rechtspflege 10%

More like this

Find more →

Mastodon

Mulit-em-dash-ate ✓

@starchy@infosec.exchange

673 Followers

Mastodon

Orkan der Rechtspflege ✓

@orkan_der_rechtspflege@mastodon.social

673 Followers

Mastodon

AntGuyOnMaui 🏳️‍🌈🐜🐜♾

@etnom@mastodon.world

673 Followers

Mastodon

📷Sayuri.H📱

@sayuri_t_h@photog.social

673 Followers

Mastodon

Sondaggi BiDiMedia :verified: ✓

@bidimedia@mastodon.uno

673 Followers

Mastodon

lobo :nix_snowflake_logo:

@lobocode@hachyderm.io

673 Followers

Auto-generated from Mastodon's public data — no affiliation with or endorsement by SocialDB. Is this you? Claim it · Remove

Aaron Tay✓

Make a deal with Aaron Tay

📊 Post engagement

📊 Activity & format

Recent posts

🐘 Community & instance

💡 Facts

🕵️ Fake follower check

About

🔀 Audience overlap

More like this

✉ Message Aaron Tay

🎤 Event / appearance with Aaron Tay