Ebay & Paypal Changes & How They Effect US

While I do not know how ebay operates internally, there are many managers in many companies who make changes under the guise of "new and improved" to justify their job when it is really just new. They have to do something or they will get laid off.

I hate paypal..the have caused me more headaches in the last year..good ridence..

Don,

While I agree that software development is costly and that new requirements pop up faster than they can be implemented, ebay's problem is unnecessary complexity and lack of testing. A test platform, once set up, should identify product regressions and insure that basic functionality remains intact. Software with too many seldom used options or multiple paths to functions inevitably exposes unintended paths or sequences that lead nowhere or expose faulty design. The ebay API is a great example of software not in sync with the product rest of the product. Developers of third-party desktop applications are frequently delayed by tardy API development to support new functionality. Market facilitator sales tax reporting and Managed Payments come to mind.

If ebay had designed their system to use the API to support the web interface, the API and the web would remain in sync. Although potentially slower, if the web interface relied on the ebay API, substantially less code would be required, potentially resulting in a more robust and secure application.

I once inherited a system with about six million lines of c code, much of it duplicative and potentially redundant. Finding and fixing bugs in sprawling systems that have grown over time becomes more like the experience of a medical doctor instead of that of a software developer. Because debugging is not possible, finding and testing scenarios that could cause observed symptoms was the norm.

in my experience, few software developers have the experience or imagination to reverse engineer and fix chronic bugs in large systems. A developer once told me that he did not bother to attempt to fix IBM 370 operating system bugs because most crashes occurred less than once a year and would be unlikely to reappear before he moved on to another assignment.

Although experienced software developers can be found almost everywhere, experienced software designers are harder to find in low-wage countries. Also, I did not point out that scripts commonly used to build web pages may be difficult to test unless the web pages are generated by software. Finally, talented database developers are even more difficult to find.

It would be easy to believe that ebay may not be able to attract top talent working in a stable, productive software development environment in competition with Google, Amazon, Microsoft or any number of startups.

It took PayPal about five years to fix bugs in transaction reporting to sellers. At first, I thought money was disappearing. Once they could produce reports that balanced, I gave up on asking for balance-affecting rows with complete transaction information. The problem with PayPal seems to be complete isolation of software developers from paying customers with no intermediary able to organize problem descriptions and change requests for the development team.

Clark,
Understood, but let's look closer at regression testing and test platforms. If we start with the test platforms and developed a list of all the hardware they would need to test a code revision on, how big is the list?

The list will easily number into the hundreds of thousands of computers and devices. But the real rub here is in the configuration of those computers and devices. Obviously these devices will be running absolutely 'clean' installs of the BIOS and operating systems (which are reset every time a new test is done).

No one in the world could possibly run regression testing on the billions and billions of 'combinations' of BIOS, operating systems, drivers, updates, and applications that actually exist and will be exercised when ebay releases a code revision.

But let's consider just one computer model, say a Dell XPS 13. We would need a number of them since Dell made running changes in the hardware over the life of the XPS 13. I do not know the exact number but let's conservatively call it 10.

Moving on to the operating systems, in the field there are users who will be running all kinds operating systems (Windows, Linux, etc) and versions. But let's ignore all of them and just drill down on the latest release of Windows (Win10). There are currently seven versions; Windows 10 (1809) 10.0.17763, Windows 10 (1803) 10.0.17134, Windows 10 (1709) 10.0.16299, Windows 10 (1703) 10.0.15063, Windows 10 (1607) 10.0.14393, Windows 10 (1511) 10.0.10586, Windows 10 10.0.10240.

Now lets move on to browsers; there is a huge number of browsers but let's just use 5 of them; Chrome, IE, Firefox, Edge, Opera. Which version of these 5 browsers should we test against? IE 11 alone has 3 major releases; 11.0.10240.16384, 11.0.9600.18617, 11.0.9600.16384 so lets greatly restrict the browser testing to the three most recent builds of each browser (a total of 15).
So to summarize the testing plan for just the Dell XPS computer;
- 10 different XPS
- 7 different builds of Win10
- 15 different browsers

We will ignore all the possible combinations of OS hotfixes, updates and any impact that the installation of any of the thousands of application that will be found in the field. Just using the list above, there are over 1000 combinations. The next question is how long does it take to run the test? Computer has to boot up, log into the operating system, start the browser, test the new ebay code, log the results. Assuming a test time of 2 minutes, that is a total of 33 hours of testing for just the Dell XPS 13. Of course some of the testing could be concurrent, but it would take a full day for a company to test just this one computer model.

The above is not how large companies test for obvious reasons, the testing cycle is highly automated and the only feasible way to regression test is to use 'representative' platforms. This confluence of hardware and software and the overwhelming huge number of combinations in the field it represents, guarantees that users will find problems that testing has 'missed'. So it is true to say that testing is inadequate but in my opinion 100% regression testing is impossible. Software testing is an incredibly complex and costly thing for all companies. At one time, Microsoft employed ten-thousand testers! In one MS meeting Bill Gates said, "I know I pay a lot of testers, but I really have no idea what they do".

So is the beef with ebay or the fact that comprehensive testing is impossible given the computing diversity and permutations?

When a company is faced with this kind of testing problem, they do the 'best they can'. They certainly will not be able to help an angry user running Windows XP with 15 years of accumulated old applications and malware. Yet there are a number of posts on this and other forums where this is exactly the case, users blaming ebay or other software companies for the issues they are seeing on their old crusty XP machines.

I sold my ebay stock a while back, I do not like defending ebay. I am highly critical of some of the things they do; for example I have been beating the drum about their badly flawed feedback system for years. But my opinion is that it is naïve to think that ebay has some employee or manager just implement a new idea/code without significant marketing analysis, cost justification, and testing. Large companies ALWAYS group code revisions into larger releases for reasons that should be obvious to anyone (as outlined above). Yet I I have seen posts where people think that some coder has simply changed something because he/she thought it a good idea. I would agree that there are some code changes which get fast-tracked. When country regulations or legal conditions change there are times when a company like ebay has to short-cut their own standard development procedures. But these kinds of situations are not by choice for a company.

My intent is to simply give some balance to the discussion. While it may be popular to hate on ebay and other software companies, there are many aspects to technology development that may not be understood by stamp collectors. I welcome folks who offer their career and business experiences to our discussions whether it is printing, accounting, agriculture, retail, manufacturing, etc. For myself, I enjoy gardening but would not consider my opinion to be as significant as a person who farmed for a living. Hopefully offering a bit of the 'other side of the coin' on technology development helps balance this topic.
Don

Hi

I worked with Point-of-Sales systems. Over the years the original package had so many changes and updates it was a bear to fix anything. A decision was made to re-create the system from scratch incorporating all the "bandages". It took almost a year but the system was much better and in the long run saved money in maintenance.

Today most Computer Science courses really don't teach any methodologies of application design. Most of the software I have seen are "start here, end there". There is hardly any use of common routines and there is duplicated code everywhere in a given application. Therefore, fixing a problem in one spot does not fix the identical problem somewhere else.

Many may think I am an out of date, but I still use flow charting as a development tool. Using it, I can see where common code can be developed, where possible problems may lie and organize my thoughts about the problem solution. Many laugh at flow charting but it is useful and serves a purpose.

I taught a class at a trade school about computer organization and basic application design. The course was dropped as students did not think it was necessary.

Jerry B

GTC is now confirmed for all sites

https://www.ecommercebytes.com/C/bl...1194482.html

Don,

Some common sense is needed here. My remark about regression testing was focused on servers owned or controlled by ebay. The reason I suggested using the ebay API internally was to limit interactions between uncontrolled user environments and internally controlled servers. I don't believe that billions of meaningful configuration combinations exist. Because ebay relies on browser software to handle variations in user configurations, they should focus on interactions with browsers giving priority to supported and up-to-date configurations. There is no excuse for not supporting Google Chrome, Microsoft Edge and other common browsers still maintained by their owners. Adherence to standards is important, especially if failure to do so can create security issues.

Going back to the original post, I was referring to hardware and software controlled by ebay. Regression testing is important for detecting loss of existing functionality from fixing bugs or adding new features. It is obvious that ebay testing procedures are deficient, as evidenced by the broken link in the ebay Spring 2019 update message sent today with a broken link in it.

I am aware that hardware age and configuration differences may expose software flaws. My experience in a multiple server corporate environment was that developers often worked on older out-of-warranty servers while testers were assigned production servers for integration testing. Sometimes this practice would expose timing problems in asynchronous software, but more likely than not, newer, faster production servers had fewer problems than development servers.

I'm just a lowly user but I know I occasionally see a change suddenly pop up on an ebay page I use every day. The change may be good or bad but usually it will go away in a day or two. How does this happen?

Last week an ebay Seller Development Manager reached out to me and wanted to set up a time to talk. I figured he wanted to sell me something but I set up the call anyway. He actually just wanted to ask some questions about my business and how he could make it better. At one point I expressed my concerns about the ebay Payment system. He seemed to be pretty well informed about the opt in experiment and it's problems and actually was pretty candid about his concerns too.

I told him I would not like it if they made it mandatory before Paypal was included because I think it would cost me sales. I also told him that if they did not make provisions for me to use Sixbits it would be very difficult for me because I'm too old to learn new stuff. He could give me no assurances on either issue but said he would take them forward to management. All in all it was a good conversation and I wonder why he contacted me, it has never happened before.

Hi Clark,
Server-side code is the 'pitcher' but client-side is the 'catcher'; regression testing client-server architecture includes both sides.

But if folks find ebay such a disaster the good news is that no one is required to use them. Another thing that some folks do not seem to notice is that forum gleans income from ebay with the affiliate links. Without ebay, I am not sure that this forum would survive. Do the criticisms posted in this community achieve anything productive? Hopefully folks here have tried to establish a relationship with the ebay Stamps Manager (Steve) that introduced himself here a few months ago. While he probably does not influence at the highest levels he offers more influence than making posts here.
Don

I have followed the discussions about, and bashing of, ebay with some interest.

I am strictly an ebay buyer at the present. Occasionally I have considered using ebay to sell duplicate materials and other items of no further interest, but have to confess that the negativity from some of the current sellers give me pause.

But as a Buyer, I can state without reservation that without ebay, I would have never been able to form several collections that I feel are my most significant accomplishments, as they include materials that I had almost never encountered in 40 years of visiting dealer storefronts, philatelic exhibitions, and antique shops. It has opened my eyes to materials that even though I don't collect them, I am amazed and delighted that they have survived and are out there for an astute collector to buy and make into his collection.

So I am for ebay all the way.

If as sellers, you feel the fees are too high, then adjust your business model to account for that. Or go elsewhere. I follow Delcampe regularly, but my purchases there are less than 10% of those on ebay as the opportunities are simply not there. Besides, much of which I find on Delcampe and even Hipstamp are duplicates of what the dealers have on ebay. Why shop there?

I understand that you sellers have to make a living. But all your competition on ebay has the same challenges you face. Deal with it just like you would if you were located in a storefront when the landlord raises your rent. Don't live in the past, when fees were cheap and you made more profit. Those days are gone just like 19c gasoline, and 50c Big Macs.

Searches: I execute about 25 searches each day. All are set to return results in the "newly listed" orientation. I have seen no indication that I am missing search results from one day to the next. I overlap my examination of the search returned items from one day to the next, and feel confident that I am seeing everything offered by all ebay sellers. And often find more to buy than I can afford. What a wonderful feeling.

I agree that when one uses the "best match" option, ebay is manipulating what you see. But the only time I ever encounter the "best match" option is when ebay leads me there by default, and I immediately re-execute the search with my "newly listed" option. Why anyone smart enough to have money to spend on ebay would use "best match" is beyond my comprehension.

End of soapbox.

mml1942 Good post, IMO every stamp collector should learn to sell on ebay. If you start with reasonable expectations you'll do fine. Selling on ebay is a great skill to have.

Took a few days gain some composure last week and started listing again this week.

Have 11 bids with $373 on 11 items in one store out of 154 listings. That is the amount of listings before I start today's reboot.

The other store is still running with some BIN's and there are a couple of buys per day.This 2nd store has 1 auction bid out of 275 listings. In that store I still can not get my head around GTC listings for some. They really do not suit auction listings format.

Opening and closing the BIN listings once a month could incur penalty?

I don't think GTC applies to auctions. I am still running 5 & 7 day auctions.

Hi KRelyea

An auction has to have a finish point.

That is the exact very important issue I got confused by.

Once you get it ,it is all about humble pie!

Does zeroing the amount on offer with a BIN listing a safe way to cancel GTC listting?