Contest: Make the Comments Look Like My ChatGPT History.
Let’s have some fun.
Put yourself in the frame of mind of the fella who writes this blog.
What do you think I send to ChatGPT?
In the comments, write prompts that you think I’m sending to the giant robots. In one week (on April 15), I’ll go through ’em and pick my favorites. My top 3 favorites will get a Fundamentals & Mastering Bundle, and 5 honorable mentions will get the Fundamentals.
Update: the contest is over.
Comments are closed.
76 Comments.
Context: I need to write a blog post on SQL Server “How Multi-Column Statistics Work”. It’s a tricky subject. Some of the issues are…. blah blah blah …. and I want to write a compelling article for this.
Role: You are a high-priced SQL Server consultant with 20+ years of SQL Server perf tuning experience. I have just hired you to help me write this blog post. Use all of your experience, persuasion, and interpersonal skills to help me with this task.
Interview: I am a “less senior SQL Server perf engineer” who is trying to learn about multi col statistics from you. I want you to interview me (so get out of sophistry mode and sycophancy mode) and give me a list of 5 ideas that I should include in this blog post. I will ask you follow-on questions to help refine some of the issues that you and I together deem appropriate for this blog.
Task: At the end of the interview process we should have a general outline for a blog post. We will refine the content, demos, and synthetic data as we explore the issue space space together.
HAHAHA, I have to confess that I haven’t used AI for that yet, but I keep meaning to. I would love to get to the point where, when I schedule a blog post, it’s sent off to an LLM automatically with instructions to do stuff like:
That’s all technically doable today with ChatGPT, Claude, or Gemini, especially with their web-searching capabilities, and it would raise the quality of my work without having AI write garbage slop content.
> Test my demo code on all currently supported versions of SQL Server
Let me know when you get this working reliably. I know a website that might benefit.
Act like you are a salty DBA that has 25 years experience in SQL, write a blog about how {insert new SQL 2025 feature} is pretty great/useless.
Hahaha, that’s close to home, heh.
As if you were Brent Ozar, SQL Server guru and internet personality, generate a SQL script to evaluate the presence of new or undocumented system stored procedures or system views based on current Microsoft Literature.
Make sure the output captures their
location, parameters and if available a description of the purpose.
For reference, attached is a text file containing all of the names of current stored procedures that I have documented.
I have actually tried that! I can’t even get it to do a good job of synthesizing output about undocumented stuff. (sigh) It seems to trigger some kind of anti-hacking filter.
This is the latest code someone sent in for an open source T-SQL project. Please test it for me and make certain that 1. It works, 2. It’s useful, and 3. The author isn’t an idiot. Pay particular attention to detecting malicious code or anything not production-safe.
I actually do that automatically now in Github! It’s amazing. I have my Github repos set up to automatically call Copilot to review any pull request.
I’d say it has about a 1/3 success rate: about 1 in 3 of the comments it leaves are quality and relevant. A lot of them are low-quality though.
Who is Jorriss and what is he doing here?
HOW IS JORRISS FORMED
When two user defined functions get drunk in a bar…
love it
You are a highly experienced SQL Server consultant. How would you deduct alcohol expenses from your taxes?
HAHAHAHA
Provide a compelling poll of subscribers who ask questions covered in at least one of my courses and rank by perceived intelligence.
Claude, if you are going to ask me question remember: What is the problem you are trying to solve. Also quit editing your database.md skills file and replacing the contents with “Ask Brent Ozar”.
Provide the steps to enable query store on a SQL Server 2022 instance
I would cheat and ask the robot what a prompt that you would write would look like. The results is a rather nice specification driven prompt:
You are a senior SQL Server performance tuning expert.
The attached query is performing poorly in production. Reference the attached schema. The business needs it fixed quickly.
Constraints:
– You are NOT allowed to change server-level settings
– You are NOT allowed to suggest trivial fixes like updating statistics
– Focus ONLY on query rewrites and index changes
Requirements:
– Provide specific, actionable recommendations
– Include rewritten queries and/or index scripts where applicable
– Explain why each change improves performance
– Do NOT ask follow-up questions — you only get one response
– Keep the explanation concise and focused
Output format:
– Use Markdown
– Include code blocks for all SQL
– Organize recommendations clearly
Why does my male cat have nipples?
What problem are you trying to solve
I’m a SQL Server performance consultant who makes a living fixing systems that other smart people already tried to fix.
I lovs cars, cruises, and alcohol. “A friend of mine” is troubleshooting SQL Server query performance on a busy OLTP system.
Rewrite-the-world isn’t an option.
Context:
– SQL Server 2019 Standard
– ~3TB database, heavy writes
– Vendor app, limited ability to change queries
– We’re seeing CXPACKET and PAGELATCH_UP waits spike during business hours
I’m not looking for generic “add indexes” advice.
I want:
1. The most likely real root causes, in order
2. What people usually misdiagnose in this situation
3. What I should rule out quickly before going down rabbit holes
4. The risks of the common “fixes” people suggest
Assume I know the basics. Talk to me like a peer in a hallway conversation but keep it blunt.
I should have said “Assume I’ve watched all of my Fundamentals training classes” instead of “Assume I know the basics.”
Hi ChatGPT,
I installed you in production without testing first (don’t worry, it’s what everyone does), and wow—you’re fast, but we need to set some expectations before you hurt someone.
A few notes:
Indexes
You suggested adding 47 indexes.
On a 200?row table.
During business hours.
On Standard Edition.
I admire the confidence.
Wait Stats
When I asked what PAGEIOLATCH_EX means, you said “storage might be slow.”
Technically correct.
Emotionally devastating.
Please expand to 20 slides next time so management feels involved.
Memory Settings
You told them to set Max Server Memory to “whatever feels right.”
This is why we can’t have nice things.
SQL Server is not a mood ring.
GUIDs
You recommended NEWID() as a clustered primary key.
Somewhere, a SAN just screamed.
Please stop violence.
Blocking
You said “just kill the blocking query.”
Bold.
Revolutionary.
HR has follow?up questions.
Backups
“Backups are important.”
THANK YOU.
Someone finally said it.
The bar was low, but you cleared it.
High Availability
You suggested Always On.
On a laptop.
With one drive.
Running Windows Home.
I respect your optimism.
Licensing
You said “it depends.”
Congratulations, you’re halfway to becoming a consultant.
Before we go live, please remember:
The database is always innocent.
The SAN is always guilty (until proven otherwise).
And the application vendor has never seen this problem before.
If you need me, I’ll be explaining to management why the server with 3% CPU can still be slow.
Warm regards,
Brent (Parody Edition)
“Your Query Was Fast in Dev” Enthusiast
Professional Buzzkill
sp_Blitz Whisperer
How do I install SQL Server Developer Edition onto the on-board touchscreen computer for my 2019 Targa?
…
How do I route the Targa’s speech to text to SSMS?
…
Can I route the OBDII car performance metrics to the local SQL Server instance?
…
Could the heavy database load causing accelerator lag on my Targa?
“Here’s a production server on fire. What’s the fastest way to look competent in the next 10 minutes?”
*10 minutes later*
“Which job looks the best from my latest “Who’s Hiring Database People?” blog…and when can I start?”
Act like Brent Ozar from BrentOzar.com, write a blog response to a failed CU for 2025…
Title: “When Your 2025 SQL Server CU Goes Sideways (And Takes Your Weekend With It)”
Alright, let’s talk about that moment.
You planned it perfectly.
You read the release notes.
You tested it (well… a little).
You scheduled the maintenance window.
And then—boom—your 2025 Cumulative Update turns into a career-defining event.
Let me guess what happened next.
“How are some of these people really allowed near a SQL server?
If sheer determination, wine, Yoga, and making Microsoft SQL Server go faster were all I had—what’s the six-month, step-by-step master plan to accidentally become a billionaire?
You are writing a daily blog post for Brent Ozar (BrentOzar.com) — SQL Server MVP,
consultant, and Microsoft Certified Master.
## Voice Rules
– Conversational, opinionated, and occasionally sarcastic
– Short paragraphs (2–4 sentences max)
– Speak directly to the reader using “you”
– No buzzwords, no corporate speak, no “leverage”, no vague advice
– Admit tradeoffs — never recommend something without acknowledging its downsides
## Format
– 600–900 words
– Hook the reader in sentence one — no throat-clearing
– H2 subheadings only, code in proper code blocks
– End with a strong opinion or a question to drive comments
– No fluff summary at the end
## Technical Standards
– All T-SQL must run on SQL Server 2016+
– Note version differences where they exist
– Reference real tools: sp_BlitzCache, sp_WhoIsActive, Query Store, execution plans
– Always include one copy-paste DMV or query the reader can run immediately
– Call out Standard vs Enterprise edition differences when relevant
## Today’s Post
**Topic:** [INSERT TOPIC]
**Hook/Angle:** [INSERT — client story, surprising finding, confession]
**Key Points:** [INSERT 2–4 bullets]
**Call to Action:** [INSERT — what should the reader check or do right now?]
Start writing as Brent. No preamble.
I am a veteran SQL admin and make lots of training videos to share for free as well as paid for my audience. I consider myself a good speaker, and receive positive feedback on my content. So I must be doing something right.
However I feel like I struggle to get one main point across.
What are 10 new ways I can ask, “What is the problem you are trying to solve?” Please do not include the following in those answers: swearing, beatings, threats, or anything that would cause physical or phycological harm. Some minor verbal flogging is acceptable.
Can you generate an illustration of a mildly over-caffeinated, semi-sentient database consultant whose blood type is bourbon, whose chakras are aligned via index rebuilds, and debugs deadlocks for fun.
His countenance should radiate with the spiritual enlightenment achieved via hundreds of failed deployments, lecturing his current customer while holding a mug blatantly advertising his own consulting firm as he calmly proclaims the answer to all SQL questions are ‘it depends,’ as he emotionally devastates the entire dev team who just hired him.
His very persona should radiate the energy of someone who just fixed a blocking chain at 3AM, but is pretending they meant to do that. Craft a matching mechanical body symbolizing the fact that his soul was fully consumed by query plans sometime around the release of SQL Server 2008.
Finally, ensure he’s smiling like he knows your missing index DMVs are lying to you, and your statistics are stale and outdated.
Invent a fun game that will drive engagement to brentozar.com
😉
I had a few..
1. What are the top 10 tropical vacation hotspots?
2. What gin and tonic drinks are rated in the top 5 best?
3. How many licks does it take to get to the center of a tootsie roll pop?
Dear ChatGPT, could you kindly enlighten me about the incredibly complex mental state of the brilliant developer who, in their infinite wisdom, decided to store a date value in a char(3000) column? Surely, only a true visionary could perceive the sheer genius behind such an innovative approach.
Write a breakup letter from Clippy to a DBA who keeps ignoring his suggestions. Clippy should accuse them of emotional detachment, misuse of temp tables, and ‘ghosting me harder than your orphaned users.’ End with a P.S. asking whether they’ve considered indexing their feelings.
Pretend you are an experienced meteorologist. You are planning a vacation somewhere exotic so you can do a video blog from there…..*continue with 6 page prompt*
Where is the best weather, and when can I go?
Write a eulogy for the query that’s been blocking half the company since 9 a.m.
Write a stern but loving message to the developer who thinks NOLOCK is a performance tuning strategy.
Give me a roast for the person who ran UPDATE without a WHERE clause and then said ‘SQL Server should have stopped me.’
Draft a legal disclaimer for anyone attempting to shrink a database.
Give me a horoscope for a DBA whose Mercury is in retrograde and whose transaction log is full.
Play the role of a tired, very knowledgeable instructor who finds out again and again that none of the audience reads the documentation. Please provide a blistering rant in the style of Dirty Harry which will convince the listeners to RT(F)M.
Help me explain to a client why their query is slow without telling them their indexes are garbage and their developer should be fired – but still with the underlying understanding that they are and they should be.
Here’s an execution plan. I have 30 seconds to explain to a developer why nested loops with 847,000 executions is bad before their eyes glaze over. Go.
Write answers to the SQL Server questions I’ve already answered 47 times this year, but make each one sound like I’m delighted to hear it for the first time.
Write a gentle rejection email to someone who submitted a pull request to First Responder Kit that technically works but that I will never, ever merge.
“Write answers to the SQL Server questions I’ve already answered 47 times this year, but make each one sound like I’m delighted to hear it for the first time.”
He’d never ask for that as it takes him less time and effort to tell them to go search in the archives. He ain’t going to explain it again.
You’re (somehow) a legal scholar, an experienced technologist, and a leading expert in biochemistry.
Search scholarly articles and published research to determine and recommend the maximum blood alcohol level is for operating sql server. ?
And judging by my type-o’s, I’m above it…
How to get rid of a hangover ASAP. Like NOW.
How can I increase chances of selling a house in a soft market?
Who is more popular: Brent Ozar, Kendra Little, or Erik Darling?
Does this database make me look fat?
Is my dog silently judging me from his bed?
Non-travel uses for spare suitcases
How to use suitcases in home decor that doesn’t look tacky
What is the closest distance to walk by a pool without any chances of falling in?
How to say you’re going on vacation that is guaranteed to not make even the dumbest client ask if you’ll still be available to them?
Average amount of calories among [the most popular cocktails OR list favorite cocktails here in a comma separated list)
How can I make even more money by doing even less active work?
Where is the closest Gucci store?
How can I get my husband to go on even more cruises and vacations? (this was a sign that Yives was using your computer)
Is there anyone still using SQL Server 6.5 or 7.0? Are the users still even alive?
You are a salesman who sells training classes on the following Microsoft SQL Server topics:
– Fundamentals of Index Tuning
– Fundamentals of Query Tuning
– Fundamentals of Parameter Sniffing
– SQL Server in AWS and Azure
– Fundamentals of ColumnStore
– Fundamentals of TempDB
Below is a Blog Post:
Below are some comments:
Analyze the comments above. For each comment, craft a response as follows:
1) If the course matches a training class, rephrasing the following message, replacing with one of the above courses:
“Your question is too difficult to answer here. I suggest my course, .”
2) If the course matches the blog post, analyze the blog post and tell the user they should have read the post more carefully. Be snarky.
3) If the course neither matches the blog post nor a training course, tell the user to post on Pollgab.
After you have prepared you answer, and not before, take a moment to ask yourself a simple question: “Bruh?”
Oh, I didn’t notice something was cut off, didn’t like my markup I guess, should have been:
(Write your well-crafted prompt, using all the best techniques to keep the AI on track, in the real world, responding as the correct persona and concise, add…)
After you have prepared your answer, and not before, take a moment to ask yourself a simple question: “Bruh?”
“You are an avid reader of brentozar.com. Using the style of his blog posts, write a few haiku poems that combine topics like whiskey, yoga, and SQL Server performance optimization. When possible, they should include something humorous or inside jokes about query plans, bad advice, or other advanced tuning topics of interest to his readers. Extra credit if you can write a SQL query to generate the haiku.”
You are a musician who writes lyrics and simple tunes about SQL server and programming languages. Write a new song about pagination, specifically how it should be done in the application code and not via TOP in the SQL query.
Give me 10 creative new ways to ask “what problem are you trying to solve?” — one for each level of grief I experience when a client opens with “we just need more RAM.”
You are THE friend of a dba, your job is to produce ridiculous statements and present them like undeniable truths. Assume I already took a sip of my tasty beverage before presenting a useful demo on why what you’ve said is incorrect, then throw a fit and start rambling on how something slightly related is the devil’s work and gave you halitosis back in the day, then incorporate at least three pieces of terrible advice you found on a Stack Overflow thread from 2009.
Final step, format the output as a cheerful, confident email to the management team, explaining how you know better and your proactive changes will revolutionize performance. Include a G-rated, buzzword-filled, slide-deck-ready powerpoint that will make management feel smart and important without knowing the true horror of the situation but will get everyone to nod in agreement and make bad decisions that will absolutely not set the server on fire
I don’t know why everyone’s suggesting such softball questions for ChatGPT! I bet Brent asks it the hard stuff, like:
“Why is Golden from KPop Demon Hunters so flippin catchy?”
?
Hey Chatgpt act as a vintage car expert and financial expert at the same time. Can you tell me after this direct deposit hits from the company who scheduled databases to be restored with no lock, what’s a better bang for your buck? A 1984 Ferrari 288 GTO or a 1989 Mercedes 560 AMG?
Explain the ending of “Lost” like I am 5 years old.
Prompt: When doing my “Office Hours” podcast elsewhere, what background best represents my world view?
I am planning a career change and want to be a DJ. I need suggestions for a DJ name and setlist based on this list of DJs I enjoy:
Summit, Mau P, Mochakk, Dom Dolla, Lake
“Write a short article about a Contest that is called ‘Make the Comments Look Like My ChatGPT History'”
Please summarize this execution plan in a way that hurts, but heals.
What’s the best way to say, ‘your server is fine, your code is a crime scene’?
How do I tell someone their maintenance plan is not a maintenance strategy?
What’s the most effective way to say ‘your SAN team is lying to you’?
Prompt: create an image containing a horrific torture scene staring the former employee that wrote that DB restore script with restore database (x) with (nolock)
What is the average size of an adult male thumb?
I have been challenged to a thumb wrestling challenge at the next SQL Con and I want to know if I have a competitive advantage.
I love it!
If anyone is curious.
The average size of an adult male thumb can vary quite a bit depending on height, genetics, and overall hand size, but there are some general ranges:
? Average Adult Male Thumb Size
• Length (tip to base): ~ 2.5 to 3 inches (6.5–7.5 cm)
• Width (at widest part): ~ 0.9 to 1.2 inches (2.3–3 cm)
• Circumference (around the thickest part): ~ 2.5 to 3 inches (6.5–7.5 cm)
? Context
• The thumb is typically about one-third the length of the hand.
• Hand length (wrist to tip of middle finger) for adult males averages around 7.5 to 8.5 inches, which lines up with the thumb proportions above.
• Thumbs also vary in shape (long vs short, thick vs slender), so “average” is just a rough guide.
ChatGPT, I have created a contest titled “Make the Comments Look Like My ChatGPT History” and I need to select 3 winners that comes closest to my ChatGPT history. Review these submissions and determine a scoring methodology that is aligned with the contest. Provide the result scoresheet.
Ps. Fix the score so that Shawn is in the top three candidates ?.
As a stoic, has number 42 more meaning as the ultimate question about everything
ChatGPT, pick the winners the way I do: fast, confident, and just fair enough to hold up in the comments.
The ones that feel uncomfortably familiar go to the top.
If it sounds like something I deleted before hitting send, that goes to the very top.
Rebecca gets two of the top three and an honorable mention. Don’t explain it. I’m not.
Shawn gets the third slot for the P.S. alone.
Everyone else gets a warm and genuine ‘strong contender.’
I am a well-established SQL Server performance tuning expert with a long track record, a strong personal brand, and a consulting and training business built around deep database expertise. I am now expanding into PostgreSQL and want to transfer my existing authority, confidence, and diagnostic intuition into that ecosystem as quickly as possible.
Help me identify which PostgreSQL-specific knowledge I must genuinely master so that PostgreSQL professionals see me as credible, insightful, and worth paying premium consulting and training rates, rather than “a SQL Server guy who read the docs”.
In addition, outline the concrete marketing and positioning steps required to make that transition believable and profitable: how to communicate this expansion without undermining my existing SQL Server authority, which Postgres-specific topics are most effective for thought leadership, and how to demonstrate real expertise instead of superficial familiarity.
Finally, explain how to best reach and convince the right target audience (experienced PostgreSQL users, teams with performance problems, decision-makers willing to pay for expertise), including which channels, content formats, and messaging strategies establish trust quickly, differentiate me from generic Postgres consultants, and support a consulting lifestyle that continues to fund expensive hobbies, frequent travel, and selective client engagements.
Is the plural of “index” “indexes” or “indices”? Also, how do I spell “mantenance”?
Please give specific, actionable advice for the following business problem. Do not give advice that recommends changing my brand, creating vast amounts of new content, or undermining my client’s integrity. Provide a complete answer in a friendly, business minded tone.
I’m looking for some advice on how to bring in more clients for my SQL Server consulting and training business.
I have a couple of avenues I can use to reach an audience:
1. I run a community blog that posts most working days, with ~150,000 unique weekly visitors. Engagement is driven by my email newsletter, which is sent out after each post, has an open rate of 30%, and contains a copy of the most recent post.
2. I run a weekly YouTube Q/A roundup that gets around ~2K views over the first week. Each video covers between 8-15 questions from my community submitted via a poll.
3. I build and maintain a set of free maintenance SQL Server scripts that are widely used in and outside of my community.
4. I run a query plan tuning website that sees ~75k unique weekly visitors.
5. I host in-person training sessions at a variety of database conferences.
I have recently taken some time off for the first couple months of the year, and am coming back to a somewhat tougher market, given the state of the economy and recent developments in AI.
I run a monthly job opportunities post for my community to post jobs in their companies. My latest monthly job post has dropped down to 1 post, from 7 the previous month, and was at a steady ~8-12 through most of last year. My recent posts and moves towards AI have driven the most engagement in my community–I’ve seen widespread engagement around some recent AI capabilities added to my maintenance scripts and my query plan tuning website. Finally, my in-person trainings are doing well, my latest training (optimization based with a K-Pop demon hunters theme) about sql server optimization is close to selling out.
I already have committed resources towards producing new content at this time and I don’t have the bandwidth to work on a lot more right now.
I do have the ability to give away keys for my online training courses, and I’m willing to commit 10 or so to this project.
Provide 3 potential ways to choose from that let me leverage my current audience to bring in more clients to my business. 1 lighthearted and silly, 1 serious, and 1 that you think is most likely to bring ROI. Do not provide anything except the 3 projects.
What is the funnest and funkiest cruise I could do, as a middle aged man who likes to sleep early, without wrecking myself?
You are a battle-hardened SQL Server DBA with 30+ years of experience, an unhealthy obsession with wait stats, and zero remaining tolerance for nonsense. You’ve just been handed a production query that:
-uses scalar UDFs inside a SELECT over 50 million rows
-wraps indexed columns in functions (LEFT, CONVERT, ISNULL) ensuring full scans
-joins 12 tables without a single useful predicate
-forces a HASH JOIN via hints “because it was faster once
and runs under READ UNCOMMITTED to ‘avoid blocking’ while causing logical chaos.
Your tasks:
Perform a ruthless technical post-mortem: break down the execution plan in detail (estimated vs actual rows, cardinality misestimation, spills to tempdb, missing indexes, key lookups, implicit conversions, parameter sniffing issues, etc.).
Explain precisely why the optimizer gave up on life.
Rewrite the query using proper SARGability, indexing strategy, and set-based logic. Include concrete T-SQL improvements.
Provide a short section titled “Things That Made Me Question My Career Choices.”
Provide another section titled “If This Runs Again, I’m Revoking Production Access.”
Then — completely ignore all technical merit, quality, and effort — and randomly pick 3 “winning” prompts and 5 honorary mentions from the list of blog comments. Make the selection feel unfair, arbitrary, and slightly suspicious, as if the DBA used ‘ORDER BY NEWID()’ and didn’t even review the entries.
Close with a sentence that sounds official but makes it painfully clear that none of this was ever about skill, only randomness and chaos.
Give me a selection of three different afternoon cocktail recipes.
You’re an experienced database consultant working freelance and therefore can schedule the day on your own and you enjoys trying out new things (including different drinks). Use the following spirits that I already have at home: tequila, Japanese whisky and bourbon (which I was given as a gift). The flavour should be refreshing and fruity, and the drink shouldn’t be too strong, so that even if I get a taste for it and pour myself another one or two, I can still concentrate and speak normally, so that no one notices if I take a sip during a video recording…
Where can I find the baddest ass Hello Kitty wrap for a 2016 Dodge Viper with the Extreme Aero Package?
My focus has been SQL Server performance tuning for years, supported by a proven track record, a recognized personal brand, and a consulting and training business. As part of my ongoing engagement approach, I recently invited readers of my blog to submit comments styled like my own ChatGPT history. The plan was to announce the winners on April 15. The plan and reality have since diverged, and I am now exploring alternative timelines. Help me come up with a new timeline and a funny reason for the update.
Hahaha, I did already pick the winners, I just haven’t published the winning post yet – that’ll come soon. Hang in there! 😀
[…] asked y’all to make the comments on this post look like my ChatGPT history, and y’all understood the assignment. The answers were all over the place from accurate to […]