Friday 29 January 2010

My Website Sucks

The result of adding haphazardly to a system, illustrated with a stack of mains power adaptors. Don't try this at home.
I know why it sucks ... it's not because I don't know how to write a proper website ... I do ... it's because it's a personal site, and I kinda tinker with it and add things from time to time, and experiment ... and because I've been using HTML for too long.

I was designing the site for someone else, I'd be less indulgent and more brutal with it. I'd insist that the owners had a clear brief of exactly what they wanted the site to do, and how to judge success. It'd be focused and lean and mean. I'd decide a visual theme, and a hierarchy, and apply it strictly. But when it's your own site, the tendency is to drift and add things and sections and use the pages as a sandbox for playing with different techniques until you end up with an indulgent hodge-podge of themes and style ideas that don't really gel.

If it was someone else's site, I'd tell them to delete the whole thing and start again. Don't just fiddle with the layout, start with a blank sheet of paper and a pen, doodle a brand new layout based on CSS, set up some default templates and rebuild the site from the ground up.



When you drift and add bits and pieces haphazardly, you end up slipping into old habits. I started writing webpages before we even had html tables. My first site (Erk's Relativity Pages) was a 300-page monster written entirely in Windows Notepad, and back then, a site designer had to learn all sorts of odd layout tricks (like using invisible GIFs as spacers) to produce efficient layouts. When tables were implemented by Netscape (and then by MS), we redesigned our pages to suit, with nice orderly auto-resizing panels – they were a pain to begin with, but the quirks and incompatibilities smoothed out with time, and we ended up using them everywhere. Tables became the answer to everything, from navigation panels to equation-setting. Then there was a craze for breaking a page up into sections and writing those sections as separate webpages embedded in frames. I managed to avoid that one (since I could see the long-term search-engine problems), but for a few years, using frames everywhere was supposed to be the mark of a "pro" designer. And then a couple of years later, the importance of search-engine optimisation became obvious, the fashion swung into reverse, and any frame-based sites began to look terribly dated.

Back in the 1980's and 1990's, the way to produce a flashy (but legible) site was to use a dark background with light text. The old CRT monitors tended to be strongly curved, with a display area that didn't extend quite to the edges, so a dark background made your page appear larger. With low-res CRT displays, "inverted" light-on-dark text was often easier to read, because the the outward blurring of light from the letters produced a sort of natural antialiasing effect. With dark text on pale backgrounds, the surrounding light tended to bleed over the characters, making them more difficult to read. Adding background patterning made the pages look more exciting, made the screen defects less distracting, and helped the user forget that they were staring at a fairly nasty little computer screen.

In 2010, things have flipped. Legibility isn't a problem on modern LCD displays, and because the screens are now flat, stark white rectangles actually look good. The monitor glass is thinner, so "snow blindness" due to light-scattering from large bright areas isn't so much of a problem, and you no longer need to add a faint background texture to pale or white backgrounds to disguise the "bitty" red, green and blue phosphor dots of a low-res CRT screen.

Nowadays, we practically squander space. On large screens, we use column layouts that waste most of the screen display, so that the central vertical column corresponds to what the user sees if they try to view the site from an iPhone. The web in the 1980s was content-starved, and you'd try to impress visitors with how much you had on your site and how much you could cram onto a small screen. In 2010 the visitor is spoilt for choice, so now designers try to keep things minimal and direct their visitors as quickly as possible to the information they actually want, otherwise they'll just click back to Google and try somewhere else.

After tables and frames, we now have Cascading Style Sheets. CSS is genuinely cool, and I really ought to rip up the existing pages and redo all their elderly table-based layout completely using CSS. Trouble is, it'll require a certain amount of work, and the immediate result will be that certain existing things (like same-height panels) won't work so well. There are bodges and workarounds, CSS isn't quite perfect yet.

The site's "look" also badly needs an overhaul. It was originally going to just be a few pages supporting the book, with a navy blue block across the top and down the left side referring to the book cover art (front and spine). On the subsequent pages, that morphed into a "program window" theme, with a title bar and an icon in the top left corner. I never quite worked out what to do with the spine. It's now an inconsistent mess, with pages on almost unrelated subjects like fractals, and should really be torn down and rebuilt.


Relativity theory is in a similar mess. A number of themes have come and gone, and left their mark on the subject. There are artefacts and traditions in the way that theory is presented that don't really make sense in the new context, and older methods that aren't compatible with newer principles. We teach special relativity as having destroyed aether theory, but we still teach SR using the length-contraction idea, which was an old aether theory concept borrowed from Lorentzian electrodynamics.

In theoretical physics, we probably have a feeling deep down that we know that we really ought to be tearing up the current system and starting again. But it'd require a lot of work without an immediate payoff, and some of the things we currently do would stop working for a while as the new system found its feet. The current system is bodgy and patched and held together with string and duct tape, but we know how to use it, and over time the bugs and fudges have started to feel like old friends. We invested a lot of time in special relativity (like website designers spent a lot of time learning the quirks of HTML tables), and now that we know that system, we tend to use it everywhere. With special relativity, we've gone further and actually redefined some key parts of relativity theory in such a way as to make SR inevitable and unavoidable, and this lock-in frees us from having to make awkward upgrade decisions.

So while it may seem that I'm sometimes a bit harsh on the theoretical physics community for being welded to obsolete and archaic systems that don't really make sense in the C21st, I do actually sympathise and empathise with their problem. They ought to rip up their SR-based structure and redesign, just as I ought to rip up my table-based webpage layout and redesign. But there's a difference between knowing that you ought to do something, and actually rolling up your sleeves and starting work, especially when there's no external deadline forcing your hand, and you always seem to have other more pressing things demanding your time.

So to help the theoretical physics community, here's a time-point. The book came out in late 2007, and sketches out the principles and the rough shape of the suggested next-generation replacement for our current general theory of relativity. This is early 2010, and the book's now been out for two years. That book is the roadmap to what comes next. So perhaps we can have a concerted start on plotting out at least a rough preliminary schedule for a replacement to general relativity, some time in 2010?

Meanwhile, I'll try to think of a way of cleaning up the website.

Friday 22 January 2010

Einstein's Cosmological Constant

LambdaBack in 1916, Einstein was still working to the assumption that the universe should be neat and tidy, and since he was now using a more mathematical approach, this meant "infinite and unchanging".
If you were solving the equations of general relativity, and getting solutions in which the universe appeared to be unstable, then you could throw those away. Chaos was bad. Order was good. Stability was good. Static solutions were better than dynamic ones.

Since it seems that gravitational mass is always positive, gravitational effects are cumulative, and over a large enough region, the combined background curvature should be enough to curve space right back on itself. The combined attraction also ought to be trying to make the universe contract, so we've appreciated for a while that unless there was some other effect in play, the universe should either be expanding and slowing, or collapsing in on itself (see: Erasmus Darwin, 1791).

Einstein wanted his universe to be pretty much flat at very large scales, so he got rid of the effects caused by cumulative curvature by adding an additional squiggle to the equations: an invented long-range repulsive effect whose purpose was to counteract the cumulative long-range effects of gravitation, allowing a tidy, constant, unchanging, static universe. If the rest of the equation generated long-range curvature effects and evolution over time, the upper-case Greek letter Lambda (Λ) represented the necessary compensating effect that might exactly cancel these effects.

Einstein referred to this as the Cosmological Constant.

Unfortunately, Einstein had made his model too tidy. A few years later, Edwin Hubble successfully measured a distance-dependent trend in the spectral shifts of light from a range of galaxies (Hubble shift), and we realised that the complicating large-scale effects that Einstein thought he'd eliminated with his Cosmological Constant seemed to be physically real. After taking some time to think the matter over, Einstein agreed that a Riemann-type solution (without Lambda) gave a cleaner and more natural implementation of General Relativity. He later described his early decision to invent the Constant to force large-scale flatness onto GR as "The biggest blunder of my career".

End of story.



However, the subject seemed to kick off again in the 1990's when a lot of headlines started appearing in in the popular science press (and in scientific papers) to do with the idea of dark energy, and the idea that the universe seemed to be expanding faster than GR1915 predicted – these articles usually declared that "Einstein's Cosmological Constant" was back, and had excited-sounding researchers competing to see who could give the best quote about Einstein having been "right all along".

This wasn't really true: Einstein's Cosmological Constant had been a mathematically-derived thing that only had one allowable value, and whose justification was to set the strengths of a range of effects in the model (large-scale curvature, distance-dependent redshifts, change in size over time) to zero. It had been there for purely logical reasons, in the context of a static universe, because a static universe seemed to need it. It existed to explain an assumed physical equilibrium that turned out not to exist, in a universe that wasn't ours. It was derived from bad assumptions, but at least it was derived.

The modern counterpart was almost the opposite. The antigravitational "dark energy" cosmological constant applied to an expanding universe that seemed to be expanding too fast for GR1915, and the effect initially had no fundamental logical, mathematical, geometrical or theoretical basis. It was, essentially, a parameter describing the extent to which the result of our GR predictions "missed" the actual data.
More recently, some researchers have tried to put the dark energy idea onto a more "theoretical" footing by arguing that perhaps the constant might not have a fixed arbitrary value, but might be a measure of the universe's expansion. That'd make the "modern" CC less fudgey, but it'd also mean that, as well as the thing not being Einstein's, it wouldn't be a constant, either.

So why did we initially get all those news stories announcing things like: "Eighty years later, it turns out that Einstein may have been right ... So he was smarter than he gave himself credit for." [*] ?

Putting it brutally, it was about PR. Attaching Einstein's name gave a false sense of historical provenance and a false sense of respectability. It let researchers use Einstein's name as a shield to deflect awkward questions about the apparent arbitrariness of their new expansion effect, and it turned a fairly boring and slightly negative story about GR failing to agree with the evidence into a snappy human-interest story about the throes of the scientific process coming out right in the end, and Einstein being right, and GR being right.

The "Einstein's Cosmological Constant returns: Einstein was right after all!" stories generated a lot of news headlines, and let researchers give interviews to magazines and appear on the telly and improve their departments' media profiles. Suddenly there were a lot of editors and journalists wanting quotes on the cosmological constant, because they wanted to print the same reader-grabbing "Einsteiney" headline, but didn't want to put their name on the claim, as reporters, because it was dodgy. So they rang round the universities and found a bunch of cosmologists happy to give the right quote if it meant getting their name in a magazine or getting onto the telly.

The story was junk. It was researchers collectively gaming the news media, and manufacturing and repeating a story that they knew would work, in order to get more media exposure. And unfortunately, that's the sort of behaviour that makes the general public more inclined to distrust scientists.

Friday 15 January 2010

Clever, Bright, and Smart

There's no single scale that adequately describes someone's abilities. People can excel at some types of task and be hopeless at others, and we have a range of different words for different types of aptitude.

Three of the most popular ones are clever, bright and smart.

Cleverness is about tool manipulation. It's about having a library of information and methods at your disposal that you can call upon to attack a problem. It's about the toolset. "Clever" researchers tend to be great at solving well-known types of problems, or well-defined problems that are attackable with existing approaches. It's a matter of going through the toolset until you find something that works. Clever people tend to be good at technical subjects that involve absorbing a lot of jargon and detail. They're not always so good at solving or understanding problems that aren't well defined, or seeing the bigger picture, or starting with a blank page.

Brightness is about being able to appreciate larger patterns and relationships that don't necessarily conform to an existing approach or definition. Bright people tend not to be so dependent on clearly-defined goals or methodology, and can take a more "free-form" approach to work, where the project parameters and characteristics emerge as the project progresses.
A computer programmer needs to be clever, but a software designer needs to be bright.

"Clever vs. Bright" is like comparing soldier ants with butterflies. The soldier ant, working with other soldier ants, manages to overcome a lot of problems even if each individual ant doesn't really know where they fit into the larger scheme of things. The butterfly arguably has the better world-view, but can't always do very much with it.

Smartness is about being able to understand and exploit opportunities to gain advantage and achieve goals. It's possible to be clever and bright without being smart. Having "smarts" means that you learn from experience and think ahead strategically, to plan how the workings of a system can allow you to achieve your desired outcome.

Smart people are often also bright and clever, but they're also smart enough to realise that their success doesn't depend on cultivating those other skills to the same extent, because once they've become moderately successful, they can "hire in" clever and bright people to do that part of the work, and delegate. Successful entrepreneurs tend to be smart, and bright, and clever, but their focus is on being smart.

Military R+D usually wants researchers who are extremely clever, but not necessarily too bright or smart. A "bright" employee might query what their work is to be used for, notice how their research fits together with others to produce a device that they aren't supposed to know about, or query the legality or ethics or consequences of the project they're involved in. A smart researcher might realise that the market value of their work is more than their current employer is paying, leave to take a better job when they realise that the project is in trouble, or try to wrest control of the project from the existing managers.



Now, this is where it gets complicated:

People who describe themselves as smart (outside a limited peer group) usually aren't.
Smart people tend not to publicly identify themselves as as smart, because it's usually not a smart thing to do. Clever people sometimes describe themselves as smart, because nobody's actually told them what the words mean, and they're not bright enough to work it out for themselves. They follow the lead of the other clever people in their group that they've heard describing themselves as smart. The bright people also don't normally describe themselves as smart, because they only hear the word being used self-referentially by people with poor social skills who are "clever-only", and they decide that they don't want to be lumped in with them.

So if you're studying monkeys in a zoo that are picking grubs out of a log that have been put there by the zookeeper, the clever monkey will become adept at using a stick to extract the grubs, the bright monkey will watch the zookeeper and only go grub-hunting when the log's just been refilled, and the smart monkey will congratulate the other two on their cleverness, assume a management position and a share of the grubs, and then patent the stick.

Different skills.

Thursday 7 January 2010

Relativity Four Point Zero

'4.0'logo and icon for the 'Relativity four point zero' website (www.fourpointzero.org)Okay, here starts a new decade. I've started a simple site sketching out the basic principles of the suggested revised general theory:

I figured that if the work of Galileo & Newton counts as "Relativity v1.0", special relativity changed some key equations and counts as v2.0, general relativity altered and added some fundamental principles and did away with SR's concept of global lightspeed constancy, and therefore counts as v3.0, then since this isn't compatible with the current textbook definitions of GR (because it eliminates the "compulsory" SR component), it counts as another "discontinuous" iteration and earns a further major version number, 4.0 .

You can't get to 4.0 without breaking a few eggs. That's what makes it 4.0 .

I was thinking of giving the new site ~twenty-six sections listed in alphabetical order, one page per letter, but I think I might just stop at five or six (the current pages A-E seem to work quite well as a logical progression). I'm trying not to fall into my usual trap of writing realms of material that most people won't want to read, and keeping things pretty minimalist, so there's a lot of the more juicy stuff left out. I think this sort of "skeleton" overview probably serves a useful purpose, so don't expect a lot of updates to the "4.0 org" site, unless other people get involved.