Misc The abandoned East European server room of LJ and Kori Ellis

[OK Computer]

It made them up. That's how LLMs work, they string together plausibly connected words with zero awareness. Sometimes you get useful results that are close to reality, sometimes you get complete nonsense like this.

Just adding “search the web for each player’s stats against the Loong Lions” to the prompt will fix that error.

 

[Exstatic]

If not entirely AI generated, at the very least they're mostly AI generated with perhaps a human touch here and there. But feels like very formulaic stuff, I've felt this way for the last couple of years TBH. I don't think this can happen without LJ and Kori Ellis' consent, which coupled with their failure to acknowledge / address everything that's been going on, further reinforces my belief they just don't give a crap anymore.

If a girl doesn't return your messages repeatedly, the lack of communication IS communicating something. Some of us have been digesting this for quite a while, others are probably still holding out hope, but we're past the point of no return IMO and the damage they've done to ST is beyond repair.

They haven’t been involved in almost 10 years. There was a message from 2016 that someone linked here that stated that. No, they don’t give a crap, because it’s no longer theirs,

 

[Neuromancer]

Well, that was kind of half ass'd by you both TBH. You could create a template with basic structure, possible conclusions from certain input, use player profile, scouting reports and advanced stats (even if not openly stated), historical data on players, sample grades from other games, etc. Also not sure what model you used, but not all are created equal. You need a lot more effort initially, but once it's set up, it should generate fairly reasonable reports.

If someone is going to go through the trouble to do all that, might as well just write the damn game grades themselves tbh

 

[Ariel]

If someone is going to go through the trouble to do all that, might as well just write the damn game grades themselves tbh

No, because you set it up once and use it forever. At some point I might take the time to illustrate this.

 

[OK Computer]

Sorry to sound doom and gloom, but this is where humanity is headed and most of the population are probably going to be “boomers” soon.

For those that don’t know, my namesake is a tribute to my favorite music album of all-time. It was made 20+ years ago.

I asked ChatGPT the meaning behind the album (see the irony here?):

OK Computer is about the spiritual cost of modern civilization. It warns that when we outsource our choices, identities, and relationships to systems — technological, political, economic — we risk becoming spectators to our own lives. It doesn’t condemn technology outright; it asks whether we can remain human as we build and inhabit systems that no longer require us to be.

Its enduring relevance comes from that unresolved tension: it isn’t anti-progress, but it insists that progress without humanity leads to alienation, meaninglessness, and quiet self-erasure.

We’re fucked.

 

[drpill]

Just adding “search the web for each player’s stats against the Loong Lions” to the prompt will fix that error.

This is true. The more context and specificity you provide, the better the results will be. But it's still glorified autofill.

 

[NASpurs]

This is true. The more context and specificity you provide, the better the results will be. But it's still glorified autofill.

Bend over, I'll show you a glorified autofill

sorry

 

[Neuromancer]

Just adding “search the web for each player’s stats against the Loong Lions” to the prompt will fix that error.

It told me it cannot search the live web and told me it could use a box score link so I gave it one and it still fucked it up

 

[OK Computer]

It told me it cannot search the live web and told me it could use a box score link so I gave it one and it still fucked it up

Well, now we know you haven’t been nice to your AI and they’ll target you first when doomsday arrives. Mines answered me perfectly.

You gotta say please and thank you, breh. You dun fucked up.

 

[Exstatic]

It told me it cannot search the live web and told me it could use a box score link so I gave it one and it still fucked it up

What AI did you use?

 

[spurraider21]

What AI did you use?

he tweeted at grok

 

[scott]

Well, that was kind of half ass'd by you both TBH. You could create a template with basic structure, possible conclusions from certain input, use player profile, scouting reports and advanced stats (even if not openly stated), historical data on players, sample grades from other games, etc. Also not sure what model you used, but not all are created equal. You need a lot more effort initially, but once it's set up, it should generate fairly reasonable reports.

AI takes a little bit of work to get the right results, but you can get it to do some good stuff.

I put together a quick prompt that read as follows:

I'm going to provide you with three screen shots. The first is the quarter by quarter results of the Spurs v Heat pre-season game from yesterday.The second screenshot is the boxscore of the Spurs players. The third is the boxscore of the Heat players.Please give me a game grades article in the style of LJ Ellis from SpursTalk.com

After this prompt and providing the three screenshots from ESPN.com, ChatGPT gave me this. Seems with a little work this would be really easy to crank out TimGPT articles.

Here’s a “Game Grades” write-up in the style of LJ Ellis (SpursTalk) based on the box score you sent. Let me know if you’d like tweaks in tone or length.

---

Grades: Spurs vs Heat — Preseason Showdown​

San Antonio 112, Miami 107


San Antonio looked sharper than expected in this preseason tilt, eking out a 112–107 win over Miami. The team leaned into balanced scoring, modest ball movement (24 assists), and defensive activity (10 steals, 5 blocks). Below, the grades — with context, praise, and critiques.

---

Starters​

Julian Champagnie — B+
24 min, 5-8 FG, 5-7 3PT, 0 FT, 3 reb, 4 ast, 2 stl, 1 blk, 4 TO, 15 pts, –4 +/–
Champagnie came out aggressive, connecting on high-risk threes and showing confidence off the catch. Defensively, he stayed active in rotations, though a few closeouts were late. His 5-of-7 from deep is a standout. The turnovers are a concern, but overall a strong showing.


Harrison Barnes — B
26 min, 2-7 FG, 2-5 3PT, 2-4 FT, 1 reb, 2 ast, 0 stl, 0 blk, 1 TO, 8 pts, +11 +/–
Barnes was steady — spacing the floor and playing his veteran role. He didn’t do much to take over, but he didn’t hurt the team either. A couple of looks didn’t fall, but he held his own in limited opportunities.


Keldon Johnson — A–
26 min, 7-9 FG, 1-2 3PT, 4-4 FT, 2 reb, 4 ast, 1 stl, 0 blk, 2 TO, 19 pts, +15 +/–
A vintage Keldon night: efficient scoring, aggressive drives, and the kind of two-way attitude you want from a core contributor. The 100% free throw shooting adds polish, and defensively he made solid stands without getting beat often.


Victor Wembanyama — B
22 min, 4-9 FG, 0-1 3PT, 2-4 FT, 0 reb, 4 ast, 1 stl, 2 blk, 5 TO, 10 pts, +5 +/–
Wemby’s stat line is modest, but he showed flashes: a few nice drops into the post, rim protection, and smart passes. The turnovers are high, but expected in preseason exploration mode. He also sat most of the rebounding duties to others — a bit surprising.


Devin Vassell — B+
23 min, 5-12 FG, 0-4 3PT, 2-2 FT, 0 reb, 6 ast, 2 stl, 0 blk, 3 TO, 12 pts, +16 +/–
Vassell had mixed shooting, but his playmaking stood out — 6 assists in 23 minutes is strong, especially in a preseason setting. He looked for the right passes and wasn’t afraid to attack closeouts. On the downside, the 0-for-4 from deep is a missed opportunity.

---

Bench & Role Players​

Luke Kornet — A
16 min, 4-6 FG, 0-0 3PT, 1-2 FT, 7 reb, 4 ast, 0 stl, 1 blk, 3 TO, 9 pts, –
Kornet was quietly excellent. His rebounding (7 boards) and rim presence were impactful, and he delivered good spacing with efficient shooting. He also helped with passing out of double teams. A very strong bench contributor.


Adam Flagler — B
19 min, 4-9 FG, 2-5 3PT, 2-2 FT, 0 reb, 2 ast, 1 stl, 1 blk, 1 TO, 12 pts, –
Flagler showed some lift in his jumper and didn’t shy away from pull-ups. His shot has improved, though his defense needs tightening. Still, a serviceable bench option, especially when he’s hitting threes.


Carter Bryant — B–
19 min, 2-5 FG, 0-1 3PT, 4-6 FT, 3 reb, 2 ast, 1 stl, 0 blk, 4 TO, 8 pts, +4 +/–
Bryant’s defensive activity remains a plus — he shows physicality and effort. Offensively, though, he’s still tentative. The turnovers hurt, but flashes of competitiveness make this a usable performance in camp.


David Jones-Garcia — B
12 min, 3-9 FG, 0-3 3PT, 1-1 FT, 1 reb, 2 ast, 1 stl, 0 blk, 2 TO, 7 pts, +1 +/–
Given limited minutes, Jones-Garcia was respectable. He attacked the rim, made solid reads, and hustled on both ends. The shot selection could be sharper, but he looked the part of a developmental guard pushing for a role.


Lindy Waters III — C+
16 min, 3-4 FG, 3-4 3PT, 0 FT, 1 reb, 1 ast, 0 stl, 0 blk, 2 TO, 9 pts, –7 +/–
Waters was active in his time — knocking down catch-and-shoots and moving in transition. The minus suggests some lineup chemistry issues, but overall a useful spark plug presence.


Harrison Ingram — C
9 min, 1-1 FG, 0-0 3PT, 0 FT, 0 reb, 2 ast, 0 stl, 0 blk, 1 TO, 2 pts, +2 +/–
Ingram didn’t have time to make much impact. He moved the ball well when on court, but lacked the burst or defensive presence to distinguish much in the brief stint.


Bismack Biyombo — C
11 min, 0-2 FG, 0-0 3PT, 2-2 FT, 2 reb, 1 ast, 1 stl, 0 blk, 0 TO, 2 pts, –
Biyombo’s contributions were modest but tidy — a few boards, defense, free throws. Not a game-changer, but did what bench bigs often need to do: stay steady.


Riley Minix, Micah Potter, Stanley Umude — Inc.
Very limited minutes; difficult to grade with much confidence. None did anything egregiously bad, but none had truly memorable moments in their brief time.

---

Coaching & Systems — B+​

In a preseason game, the coaching staff deserves credit for letting the rotation breathe while still pushing for aggressiveness. The balance between letting younger guys explore and maintaining competitive integrity was well struck. The offense found flow in spurts, and defensive switches were mostly handled cleanly. Still, there’s room to tighten transition defense and rebounding habits.

---

Final Thoughts​

This win is encouraging for the Spurs. A double-digit spread in the first half (SA up 33–24 after one) shows the team came out sharp, though Miami chipped away late. San Antonio shot 46.5% overall, 36.1% from three, and 70.4% at the line. They edged Miami in rebounding (49–36) and won the turnover battle (21–23, though it’s close) — both strong signs in a preseason test.


The core — Keldon, Champagnie, Vassell, Wembanyama — looked good in stretches, and the bench showed potential pieces. The team defense was active, though lapses exist, especially in communication. If the Spurs can tighten up turnovers and sustain consistency in their second units, this form is worth building on.


Grade for the night: B+ — more positives than negatives.

---

If you want a shorter “quick take” version or adjustments in tone (more critical, more hype, etc.), I can whip that up too.

 

[scott]

It also gave pretty much everyone Bs... so it nailed the timvp style tbqh

 

[dastexan]

AI takes a little bit of work to get the right results, but you can get it to do some good stuff.

I put together a quick prompt that read as follows:



After this prompt and providing the three screenshots from ESPN.com, ChatGPT gave me this. Seems with a little work this would be really easy to crank out TimGPT articles.

That AI needs to be trained better, giving Vassell a B + doesnt seem accurate.

 

[jeebus]

Kori was on here last month and said all was well tbh.

that's just lefty, tbqhwy

 

[heyheymymy]

For those that don’t know, my namesake is a tribute to my favorite music album of all-time. It was made 20+ years ago.

I was going to ask, was your name a Kubrick reference or a Yorke reference lol. Both work for me.

 

[Neuromancer]

What AI did you use?

ChatGPT

 

[ird80]

Could u feed it the actual game replay/ footage

 

[LeBowen]

That AI needs to be trained better, giving Vassell a B + doesnt seem accurate.

Seems perfectly in line with timgpt, tbh.

Could u feed it the actual game replay/ footage

Can it even get any relevant basketball data from the feed other than boxscore numbers?

 

[ird80]

Well analyzing defense n offensive sets...like how can it say Julian had late closeouts without game footage

 

[Paramore]

hmmmmm

 

[Paramore]

hmmmmm

 

[LeBowen]

I still think Spurstalk/Warhammer analogy is perfect.

The whole idea of the Emperor is that you don't know whether he's alive or dead. The whole Imperium might be running on superstition. There's no guarantee that the Emperor is anything other than a corpse with a residual mental ability. It's got some parallels with religious beliefs and principles.

Then you change it to:
The whole idea of timvp is that you don't know whether he's alive or dead. The whole Spurstalk might be running on superstition. There's no guarantee that timvp is anything other than a corpse with a residual mental ability. It's got some parallels with religious beliefs and principles.

 

[Sugus]

It's been apparent for some time that the Grades articles were AI-assisted or entirely AI-written with maybe some editing afterwards.

The single biggest tell-tale, having read them for years, is that LJ used to pretty consistently give a first-person perspective or comment as to the team's performance in the first two-paragraph introductions. That stopped about 3 years ago, during the onset of Covid, and game breakdowns became a lot more "the Spurs executed well down the stretch but were outmatched by the Heat's 3pt shooting" kind of impersonal narrative style. The kind of comments you don't have to watch the games to write.

In any case, I'm beyond interested in the potential to scrape the Grades archive on the Spurstalk website, and provide it to a LLM -- it shouldn't take much more than a dozen or two articles for it to perfectly mimic TimGPT's style tbh

 

[Sugus]

They haven’t been involved in almost 10 years. There was a message from 2016 that someone linked here that stated that. No, they don’t give a crap, because it’s no longer theirs,

Yeah, that post really put shit into perspective. 2016 really was almost 10 whole years ago... Time fucking flies.

If you asked any one of us whether we're still personally involved with any job or project we voluntarily left 10 years ago, with a family of four kids to take care of in the meantime....... I know my answer is a resounding "no".