Ah yes, because a probabilistic text generator should be trusted to appraise a rare collectible that wasn't even known to exist as of the time it was trained.
It should be trusted about as much as any Random Internet Stranger's appraisal, as I suppose we're both aware.
I personally do trust it a tiny bit more, actually, because it demonstrably did a modicum of due diligence by searching for comparable NES collectibles that were sold in the past, having a look at some collectors' roughly known budgets, and doing some numbers off the results. This is something Random Internet Stranger might or might not have done before just talking out of their ass.
Honest questions:
After reading TFA, did you not wonder, just a bit, what the box could be worth?
Would you have taken the time to respond to my post if I hadn't told the source and just said it's "probably" worth x, instead of saying "might, according to"?
If I had used ChatGPT in the way you described, I would have likely verified that the numbers sourced from search results it found were used correctly when it was "doing some numbers," since LLMs don't really do numbers at all. At that point, why not just search and do the numbers myself?
Answers to your questions below:
> After reading TFA, did you not wonder, just a bit, what the box could be worth?
No, but admittedly I have a lot of experience selling rare collectibles of several different kinds myself. So I knew enough to know that the only remotely accurate way to estimate the value of something like this (rare, not previously known, possible collector base is very small but dedicated) is to put it up for auction. I recognize that not everyone may understand this, though.
If you browse auctions for collectibles, and dig up the rarest of the rare items, you'll notice that even if the auction house provides an "estimate," they are usually wildly off compared to the actual hammer price, despite coming from the presumable experts in that field. Sometimes the estimates are high, sometimes low; it just depends on the whims of the market.
> Would you have taken the time to respond to my post if I hadn't told the source and just said it's "probably" worth x, instead of saying "might, according to"?
Yeah, I quite possibly would have, and I likely would have responded with an answer similar to my answer to the previous question.
I much appreciate the insightful comment and answers.
To sort of justify my approach, collectibles are a field I transparently don't know much about, apart from the occasional and sensationalist press article. The article left me casually curious, and I was suprised to find no such comment here. Still, I really didn't want to dive into yet another rabbit hole, so this felt like the perfect task to hand off to an LLM to serve me a rough estimate. It's a niche of thought I've come to gladly delegate to an LLM, have it quickly google sth for me.
>Sometimes the estimates are high, sometimes low; it just depends on the whims of the market.
That's probably the nature of the business. But doesn't that mean it more or less evens out on average? I mean, you gotta start somewhere with an appraisal like this.
FWIW, my experience with recent LLM models is that even free tiers are pretty capable of finding and aggregating information available on the internet, and doing some basic arithmetic with it. I may need to mention ChatGPT's answer was a lot more elaborate than the number I quoted. Possible scenarios ranged from a "conservative" $600k up to $1.5-2 million in the event of a bidding war between serious collectors at a premier house.
And frankly, considering how naive the question is, I suppose that guess isn't much worse than yours or mine, isn't it? I understand if you don't indulge in this kind speculation, though.
Thanks for your thoughtful reply! Yeah, I just wouldn’t speculate that much. If someone had put a gun to my head and asked me for an estimate of value, I’d probably give a huge range like $100k up to several million, and the grade also matters as only a handful of them got a perfect 10. I’d also give the caveat that video games are not really my specialty.
> That's probably the nature of the business. But doesn't that mean it more or less evens out on average? I mean, you gotta start somewhere with an appraisal like this.
Not sure, but yes it might even out on average! But that still doesn’t help you answer the question of the market value of one specific item.
One thing I didn’t mention is that there’s another factor at play in this specific case, which is that total the number of items (97) is likely as large or larger than the current number of collectors in existence who would actually buy them at current market prices. I’ve seen this before with other niche/high end collectibles. This means that if you put all of them on the market at once, you’re likely to crash the price of this item and possibly others related to it. So if you want to maintain the current price, you need to trickle the items out into the market slowly.
One thing that’s interesting is that the blog post just admits how many total items were found. This is information that collectors don’t always have in other situations like this, and simply knowing that there are 97 out there may be enough to keep the market price lower than it otherwise could have been. There are other items in this field (Nintendo PlayStation prototype hardware and accessories, first-run pristine sealed copies of classic games like Mario and Zelda, etc) where the total items in existence are thought to be in the single digits.
I personally do trust it a tiny bit more, actually, because it demonstrably did a modicum of due diligence by searching for comparable NES collectibles that were sold in the past, having a look at some collectors' roughly known budgets, and doing some numbers off the results. This is something Random Internet Stranger might or might not have done before just talking out of their ass.
Honest questions: After reading TFA, did you not wonder, just a bit, what the box could be worth? Would you have taken the time to respond to my post if I hadn't told the source and just said it's "probably" worth x, instead of saying "might, according to"?