blender
Member of DD Central
Posts: 5,719
Likes: 4,272
|
Post by blender on Jul 31, 2014 17:16:07 GMT
The app is not, afaik, in any way shape or form automated (not having an Ipad I can't check) and it is supposed to use the same API we don't have. The bid bots will be temporarily inconvenienced (mine's staggering back to life, slowly), although I am sure the big players have had advance notice since apparently FC can't live without them. I agree about the loan book - when FC noticed it was still available they went and hid it. I have obviously not understood the app, lazy thinking - but I am happy with PC and keyboard. My main point, as others are suggesting such as Grumpy on another thread, is that FC are no longer straight, as they used to be, or accountable.
|
|
|
|
Post by GSV3MIaC on Jul 31, 2014 20:19:48 GMT
[quote source="/post/18022/thread" timestamp="1406825624" author=" goldservice Re-educate the lot, I say![/quote] This'd be the ex_USSR Version of 'Re-educate' I assume? 8>. |
|
|
|
Post by aloanatlast on Aug 1, 2014 12:21:47 GMT
The new look loans page has many niggles, for example, after placing your first bid the bid button moves down slightly so the pointer may no longer be on it so you have to take aim again before placing subsequent bids. Also, the pop up New Bid! boxes spread down the screen (when several bids are made) until they cover the 'Total bid so far', and then they cover the bid button itself so you have to stop bidding. Annoying. It makes me think that (as happens so often in software development) the FC manager, designer, writer and tester are not actually users of the product themselves and get to think in the end that users are a pain. Re-educate the lot, I say! I keep getting the impression that most software is written by people who don't even use computers.
|
|
oldgrumpy
Member of DD Central
Posts: 5,087 
Likes: 3,233
|
Post by oldgrumpy on Aug 1, 2014 12:42:19 GMT
I haven't seen much difference on FFox. Is this all about a "tablet" version of the site?
|
|
|
|
Post by goldservice on Aug 1, 2014 15:03:15 GMT
I haven't seen much difference on FFox. Is this all about a "tablet" version of the site? I think it was abut the new look loans page for PCs/laptops, not a mobile or tablet version. The new look loans page had disappeared again when I looked at 7:00 today and we're still back with the old look. It's a good thing to take poorly software releases back to the workshop but it's also a bad thing 'cos just as we are trying to get used to a new layout and trying not to grumble about the changes, we have to get used to the old one again and this saps goodwill ... Better to test more thoroughly before release of changes. I know it's easy to scoff at others' work but the profession has methods (development environment, test protocols etc) for ensuring standards in software development ...
|
|
|
|
Post by GSV3MIaC on Aug 1, 2014 15:30:14 GMT
You can still get the new loan page by putting /loans_requests/<auction ref #> in where normally you have /loans/<auction ref #>/auction, what they have done is stopped the automatic redirecting of the old pages to the new ones (for some users on some platforms only, apparently). Of course if you click the links on watchlist or loan request, you get what you are given (currently the old pages) but a quick URL edit will get you the new one (at the moment, until someone at FC pulls the plug .. hmm, why don't I trust them all of a sudden).
I am still doing some testing / debugging on the new pages, but the advantages are slim. It OUGHT (in theory, so they say) be swifter, but it doesn't come out that way when I (or rather my software) benchmarks it. Bit like the Windows logon trick - yeah, the page may appear to load in 2 seconds instead of 3s, but it is not populated and usable for 5.5s, vs 4.5s previously.
|
|
|
|
Post by goldservice on Aug 1, 2014 16:29:51 GMT
On the new loan page, was anyone else irritated by the way that the bid button no longer changes colour after each bid and back again when it is ready for the next bid so we have lost one visual indication of bidding progress? Also, after my last bid for a loan, the button remains the focus which is a bit worrying tho' I still have to check if this really happened (when the new page reappears.)
|
|
mikeb
Posts: 1,072
Likes: 472
|
Post by mikeb on Aug 1, 2014 17:02:23 GMT
yeah, the page may appear to load in 2 seconds instead of 3s, but it is not populated and usable for 5.5s, vs 4.5s previously. As I pointed out in the other place, it loads with information which is just wrong, until it decides to correct it. It's bad enough that information in various areas is wrong (display errors) and stable, but information that appears on first load that is just wrong/placeholder rubbish is not useful at all. The auction I tried it on showed as "ended" with no ability to bid. Some seconds later, it wasn't ended after all, I've got ability to bid. Eh? This is not an update to make things more efficient, really, it is just slower than before.
|
|
|
|
Post by GSV3MIaC on Aug 2, 2014 10:41:04 GMT
In the other place one of the FC developers has explained how they ran this beta test .. basically randomly assigned ~30% of users to get the new page, using a cookie, without any advance warning, and no way of backing out. In my experience (disclaimer .. I used to run a big software development group) our term for this sort of approach would have been something like 'complete cods up'. If you want to test, start with a call for volunteers. If you must stress test using lots of cannon fodder, at least pre-warn them. A bail out mechanism is useful too... heck, even Google (see Google maps, Gmail, etc.) provide that, and they're usually not a shining example to anyone.
The new loan pages have a few problems (likely fixable) but the deployment methodology just purely sux .. unless of course they were deliberately trying to sabotage the bid-bots, and overload the help lines, which seems unlikely.
|
|
sl125
Member of DD Central
Posts: 85
Likes: 64
|
Post by sl125 on Aug 2, 2014 18:04:29 GMT
Gs...
In traditional software development that may well be the case, but when you need to test the effect of incremental changes across whole populations of users that have different browsers, OS, etc, then A/B testing is a good strategy.
A/B testing is similar in principle to a randomised control trial: you assign x% of visits to your website the "new" version, and the remainder of users act as your control. I see that FC use Optimizely, which is a tool specifically for this purpose.
As a fellow FC investor that has built a suite of Selenium / Java scripts to automate my bidding, it is frustrating when the website changes its DOM, etc, but that is their prerogative at the end of the day. I just react by tweaking my code.
|
|
mikeb
Posts: 1,072
Likes: 472
|
Post by mikeb on Aug 2, 2014 18:06:27 GMT
randomly assigned ~30% of users to get the new page, using a cookie, without any advance warning, and no way of backing out That did make my mind boggle, at first, until I realized this is Funding Circle, and testing things out is ALWAYS done on live customers, without warning, and sometimes with no way for FC to back out either  If I discovered I'd been cookied like that, in the middle of trying to actively use the site, I'd been even more peeved than I am. It's not like when Zopa did a secret(ish) A/B test between two new intro pages a while back, to gauge which was most appealing. At least the core bidding/account management stuff stayed working and usable, and unchanged.
|
|
|
|
Post by GSV3MIaC on Aug 2, 2014 20:40:21 GMT
randomly assigned ~30% of users to get the new page, using a cookie, without any advance warning, and no way of backing out That did make my mind boggle, at first, until I realized this is Funding Circle, and testing things out is ALWAYS done on live customers, without warning, and sometimes with no way for FC to back out either If I discovered I'd been cookied like that, in the middle of trying to actively use the site, I'd been even more peeved than I am. It's not like when Zopa did a secret(ish) A/B test between two new intro pages a while back, to gauge which was most appealing. At least the core bidding/account management stuff stayed working and usable, and unchanged. Yeah, but when it's not announced (and apparently not tested on volunteers first) you might find your code had bought you a few £k of some loan you didn't really want, because it got confused by the new web pages. The 'lets pick some random victims' might be appropriate just before release, but this stuff is not ready for primetime! Take a look, for instance, as any of the fixed rate loan requests in the new format, eg www.fundingcircle.com/loan_requests/39705and spot the missing / incorrect info. Of course FC frequently confuse alpha, beta, and zeta testing, so I guess we should be expecting it. I'm always delighted to volunteer to beta test anything (eg the API. 8>.), although I do expect some place to send comments and get feedback, and some way to bail out if the beta is not usable .. I beta tested various games, email programs, etc. for years, when I had more time (the GSV handle is probably still buried in a few game credits someplace).
|
|
blender
Member of DD Central
Posts: 5,719
Likes: 4,272
|
Post by blender on Aug 2, 2014 22:53:56 GMT
randomly assigned ~30% of users to get the new page, using a cookie, without any advance warning, and no way of backing out That did make my mind boggle, at first, until I realized this is Funding Circle, and testing things out is ALWAYS done on live customers, without warning, and sometimes with no way for FC to back out either If I discovered I'd been cookied like that, in the middle of trying to actively use the site, I'd been even more peeved than I am. It's not like when Zopa did a secret(ish) A/B test between two new intro pages a while back, to gauge which was most appealing. At least the core bidding/account management stuff stayed working and usable, and unchanged. Not so Mikeb. FC do not always test things out on live customers. May I point out their exemplary performance in testing the 'upgraded' loan book. Even though the changes required might be thought small by many lenders, FC spent two whole months in very thorough testing of its performance. Far from testing it on lenders to find bugs, they ensured that it was not released for lenders to use and risk inconvenience until they were absolutely sure that we would not be troubled by poor functionality. No doubt it is a fine example of their future intentions - after the bidding page.
|
|
|
|
Post by goldservice on Aug 5, 2014 8:57:40 GMT
Gs... In traditional software development that may well be the case, but when you need to test the effect of incremental changes across whole populations of users that have different browsers, OS, etc, then A/B testing is a good strategy. A/B testing is similar in principle to a randomised control trial: you assign x% of visits to your website the "new" version, and the remainder of users act as your control. I see that FC use Optimizely, which is a tool specifically for this purpose.
As a fellow FC investor that has built a suite of Selenium / Java scripts to automate my bidding, it is frustrating when the website changes its DOM, etc, but that is their prerogative at the end of the day. I just react by tweaking my code. As I understand it, A/B testing makes sense when you are trying to increase sales from a web site. For a site such as FC, I’m not sure that we are buying anything so sales can’t be increased. This also means that one of the A or the B cannot be a control for the other as there is no metric such as sales quantity to compare. Looking at Optimizely’s site, I can’t see what use a service site like FC could make of their product. Is it another example of a software company flogging its inappropriate software to gullible customers?
|
|
mikeb
Posts: 1,072
Likes: 472
|
Post by mikeb on Aug 5, 2014 17:41:41 GMT
From my /etc/hosts file ... 127.0.0.1 cdn.optimizely.com # Funding circle javacrap Blocked this -- whatever the hell it is -- from loading ages ago to try and speed up page operation. I hope it's nothing important. It certainly has nothing to do with optimizing anything. I optimized it away. There were already entries for other "*.optimizely.com" blockers included from the anti-spam anti-advert anti-  entries I've imported from elsewhere. |
|
entries I've imported from elsewhere.