As long as you present it as an estimation you're fine. The 12 people that you piss off are used to getting screwed and often have alternate arrangements (like a UPS Store box in the nearest big town)
I used to sell recently discontinued stuff in college (mostly Sun and Apple Powerbooks) on eBay, and most of my business was to places where mainstream vendors couldn't or wouldn't ship.
Estimates are not okay, at least in non B2B scenarios if you have a competent lawyer advising you. At a consumer level (where you have to make very serious assumptions about the customer's level of understanding of what an estimate is), you open yourself up to bait and switch claims.
Exacerbating the problem is that there is a huge variability in shipping prices. The parameter space is huge: residential vs business, single-family home vs multi-family building, location relative to shipping/sorting hubs, incidental delivery density (number of unrelated deliveries near you), handling concerns, signature requirements, size/weight X potential lane shipping options (weight dominates costs in ground lane shipping, physical dimensions dominate costs in air shipping). A 50 lb minifridge delivered to a multi-family residential building with a mail room in the same city as the origin warehouse may cost a carrier $1 to deliver. But if it is delivered to rural wyoming it may cost $40 and if it travels via air shipping you can tack on an additional $100. Estimates formed via simplistic assumptions may end up being inaccurate with standard deviations in excess of 100%. At the level of sophistication of the average customer, giving an estimate of $20 when actual costs are $100 screams bait and switch, whereas an estimate of $100 when actual costs are $20 means that the customer never makes it to checkout. If you aren't within a reasonable ~5% of actual, you can't reasonably provide estimates to the customer.
There is a push by carrier customers to provide estimated estimate accuracy for this reason (businesses want to know how accurate an estimate is if they have incomplete information), but ironically that ends up putting the carriers in the same position: data-validated claims about positive/negative bias in estimate accuracy can basically be considered bait and switch in court, even for large B2B customers with very detailed contracts.
The scope of the thread that I was referring to was "small ecommerce companies".
Small companies, depending on your vision of small, are usually pushing this complexity to the carrier via a shipping cost premium or outsourcing fulfillment altogether. I agree that in a B2B scenario, you need to meet whatever standard the contract lays out.
If estimates are not okay and the true cost of shipping cannot be calculated in a short period of time, then why is it okay to display a shipping price during the checkout process right before I click "submit", since this price is presumably an estimate? If presenting an estimate during the checkout process is actually okay, why not display it in the cart before the checkout process?
Getting a shipping estimate doesn't take a lot of time (a few hundred milliseconds at most), and by the time you are in checkout, you should have detailed enough product and destination information to get an extremely accurate estimate from the carrier APIs (which would only change in the event of the vendor misjudging item size or weight or something similar).
I used to sell recently discontinued stuff in college (mostly Sun and Apple Powerbooks) on eBay, and most of my business was to places where mainstream vendors couldn't or wouldn't ship.